The Forecast Xperts® www.johngalt.

com

Forecast Xpert®
User Guide
Standard & Premium
Copyright © 2018 John Galt Solutions, Inc. All rights reserved.
John Galt Solutions® 17 N State Street, Suite 1890 • Chicago, IL 60602 • Phone: 312-701-9026
Copyright Information

All Licensed Software is protected by the copyright laws of the United States and by
applicable international treaties. No rights under copyrights are transferred to Customer,
except as specifically provided in the License Grant.

Rev 0 2
Contents
Chapter 1 Introduction ....................................................................................................................9
Section 1: Overview .....................................................................................................................9
Section 2: ForecastX Highlights..................................................................................................10
Section 3: New Features in ForecastX .......................................................................................10
Section 3: User Groups ..............................................................................................................12
Section 4: Documentation Conventions ....................................................................................12
4-1: Text ...............................................................................................................................12
4-2: Screen Captures ............................................................................................................12
Section 5: Acronyms and Abbreviations ....................................................................................13
Section 6: Software Editions ......................................................................................................13
Section 7: Software License; Intellectual Property ....................................................................14
7-1: Preamble.......................................................................................................................14
7-2: License Grant ................................................................................................................15
7-3: Restrictions ...................................................................................................................15
7-4: Ownership of Media .....................................................................................................16
Section 8: John Galt Solutions Tech Support .............................................................................16
Section 9: Revision History ........................................................................................................17
Chapter 2 Installation ....................................................................................................................18
Section 1: Overview ...................................................................................................................18
Section 2: Requirements ...........................................................................................................18
2-1: Hardware ......................................................................................................................18
2-2: Software Media ............................................................................................................19
Section 3: Uninstallation............................................................................................................19
3-1: Backward Compatibility ....................................................................................................19

Rev 0 3
Section 4: Installation ................................................................................................................20
4-1: Installing ForecastX™ ....................................................................................................20
Chapter 3 Using ForecastX ............................................................................................................22
Section 1: Overview ...................................................................................................................22
Section 2: ForecastX toolbar in Excel .........................................................................................22
2-1: Forecast Adjust .............................................................................................................30
2-2: Accuracy Report............................................................................................................32
2-3: Forecast Relocation ......................................................................................................34
Section 3: Quick Start Forecast ..................................................................................................37
3-1: Data Cleansing ..............................................................................................................41
3-2: Group Forecasting ........................................................................................................41
3-3: Event Modeling.............................................................................................................42
3-3-1: Event and Promotional Modeling Forecasting ..........................................................43
Section 4: Data Capture Tab.......................................................................................................46
4-1: Data Cleansing ..............................................................................................................48
Section 5: Forecast Method Tab ................................................................................................51
5-1: Transform .........................................................................................................................52
5-2: Adjust ............................................................................................................................53
5-3: Analyze .........................................................................................................................54
5-4: Preview .........................................................................................................................56
Section 6: Group By Tab .............................................................................................................56
6-1: On-Fly Rebalance ..........................................................................................................60
Section 7: Statistics Tab..............................................................................................................61
Section 8: Reports Tab ...............................................................................................................62
8-1: Standard Report............................................................................................................63
8-1-1: Standard Report Grouping Example .........................................................................67
8-2: DRP Report ...................................................................................................................71

Rev 0 4
8-2-1: DRP Settings..............................................................................................................74
8-2-2: DRP Example .............................................................................................................80
8-2-3: DRP Reports and Group Forecasts ............................................................................83
8-3: Executive Report...........................................................................................................84
8-4: Audit Report .................................................................................................................85
8-5: Pivot Report ..................................................................................................................88
8-6: Append .........................................................................................................................90
8-7: Reports Advanced Options ...........................................................................................91
8-7-1: Standard....................................................................................................................91
8-7-2: Collaboration ............................................................................................................92
8-7-3: Tracking Table ...........................................................................................................94
8-7-4: Report Bursting .........................................................................................................95
8-8: Conversion Factors .......................................................................................................97
Section 9: Menu Bar ..................................................................................................................98
9-1: Scenario ........................................................................................................................98
9-2: Tools ............................................................................................................................103
9-3: Help.............................................................................................................................106
Section 10: Customer Feedback.................................................................................................107
Section 11: Upgrade License ......................................................................................................109
Chapter 4 Forecast Methods .......................................................................................................111
Section 1: Overview .................................................................................................................112
Section 2: Adaptive Exponential Smoothing ............................................................................112
Section 3: Bass Model..............................................................................................................115
Section 4: Box-Jenkins .............................................................................................................117
4-1: Performing ARIMA Analysis ........................................................................................120
Section 5: Census X-11 .............................................................................................................121
5-1: Using Advanced Census X-11 with ForecastX ............................................................125

Rev 0 5
Section 6: Croston Intermittent...............................................................................................127
Section 7: Decomposition........................................................................................................130
Section 8: Double Simple Exponential Smoothing Brown .......................................................132
Section 9: Double Exponential Smoothing Holt .......................................................................133
Section 10: Erlang B ...................................................................................................................137
Section 11: Erlang C ...................................................................................................................141
Section 12: Event Model ............................................................................................................145
Section 13: Gompertz Curve ......................................................................................................150
Section 14: Holt-Winters ............................................................................................................152
Section 15: Logistic Curve ..........................................................................................................155
Section 16: Moving Average ......................................................................................................157
Section 17: Multiple Regression ................................................................................................158
Section 18: New Product Forecasting ........................................................................................162
Section 19: Poisson ....................................................................................................................164
Section 20: Polynomial Regression ............................................................................................166
Section 21: Probit Curve ............................................................................................................168
Section 22: Simple Exponential Smoothing................................................................................170
Section 23: Slow Moving Demand .............................................................................................173
23-1: Step by Step Configuration ..........................................................................................174
23-2: Express Configuration ..................................................................................................182
Section 24: Stepwise ..................................................................................................................186
24-1: Stepwise Forward Regression ......................................................................................186
24-2: Stepwise Backward Repression ...................................................................................187
24-3: Stepwise Forward with a Backward Look Regression..................................................187
Section 25: Trend (Linear) Regression........................................................................................191
Section 26: Triple Brown ............................................................................................................195
Chapter 5 Statistics ......................................................................................................................198

Rev 0 6
Section 1: Overview .................................................................................................................198
Section 2: Accuracy Statistics ...................................................................................................200
2-1: AIC...............................................................................................................................200
2-2: BIC ...............................................................................................................................200
2-3: R-Square .....................................................................................................................201
2-4: Means Absolute % Error .............................................................................................201
2-5: Adjusted R-Square ......................................................................................................201
2-6: Sum Square Error........................................................................................................202
2-7: Safety Stock ................................................................................................................202
2-8: Chi-Square ..................................................................................................................205
2-9: Cochran-Orcutt ...........................................................................................................205
2-10: Means Absolute Error ..................................................................................................205
2-11: Mean Error ...................................................................................................................206
2-12: Mean Square Error.......................................................................................................206
2-13: Normality Error ............................................................................................................206
2-14: Root Mean Square Error ..............................................................................................206
2-15: Standard Deviation Error .............................................................................................207
2-16: Theils ............................................................................................................................207
2-17: T-Test ............................................................................................................................207
Section 3: Analysis Statistics .................................................................................................... 208
3-1: Durbin Watson ............................................................................................................208
3-2: Mean...........................................................................................................................209
3-3: Median........................................................................................................................209
3-4: Standard Deviation .....................................................................................................209
3-5: Variance ......................................................................................................................209
3-6: Mean Square Deviation ..............................................................................................209
3-7: Correlation Coefficient ...............................................................................................210

Rev 0 7
3-8: Max .............................................................................................................................210
3-9: Min..............................................................................................................................210
3-10: Kurtosis ........................................................................................................................210
3-11: Skewness ......................................................................................................................210
3-12: Range ...........................................................................................................................211
3-13: Auto-Covariance ..........................................................................................................211
3-14: Auto Correlation ..........................................................................................................211
3-15: Coefficient of Variation ................................................................................................211
3-16: Sum of Squared Deviation ...........................................................................................212
3-17: Root Mean Square .......................................................................................................212
3-18: Mean Absolute Deviation ............................................................................................212
3-19: Mode ............................................................................................................................212
3-20: Ljung Box ......................................................................................................................213
Section 4: Regression Statistics................................................................................................213
4-1: ANOVA Table ...............................................................................................................214
4-2: Coefficient Table .........................................................................................................215
4-3: Coefficient Determination ..........................................................................................216
4-4: Covariance ..................................................................................................................216
Index ...............................................................................................................................................217

Rev 0 8
Chapter 1 Introduction

Section 1: Overview

ForecastX™ is used as an Add-in within Microsoft Excel to forecast and plan. ForecastX™ is
essentially a simplified version of the Viewer; all your work is saved in Excel and not in a
database.

ForecastX is a family of forecasting tools capable of performing complex forecast methods and
the ability to address enterprise- specific business needs. ForecastX tools use data provided by
users to automatically discern which statistical method is best for each specific set of business
needs. The ForecastX product family is completely scalable, with tools and custom solutions
along a clear upgrade path established to provide accurate forecasting solutions for your
company as it grows.

From the Forecast toolbar in Excel, you can control and launch all the major features that are
available within the Demand Management Engine, and like the Viewer, provides the central
location to interact with the forecast.

You can perform the following tasks from the ForecastX toolbar:

View forecasts and associated data in Microsoft Excel.

Rev 0 9
Chapter 1

Edit parts of the Forecast in Microsoft Excel.

ForecastX is part of the Atlas Planning Suite, which consists of several applications that make up
Enterprise Forecasting. The figure below is a depiction of the Atlas Planning Suite.

Section 2: ForecastX Highlights

• Simplified version of the Viewer.

• Easy access toolbar in Microsoft Excel.

• No database necessary.

Section 3: New Features in ForecastX Version 10

This section provides information on new features found in this release of ForecastX 10:

• Support for Office 2016

• Support for Office 365 - Desktop.
• Automated License Management System
• DRP Settings: Include new setting as Min. Quantity Order, Max. Quantity Order and On
Order Quantity. Also support: Statistical Safety Stock, Reorder Point and Coverage Safety
Stock on Min Quantity Type.
• New Report: Allow save max. 3 lags, clear and generate Lag Accuracy report.

Rev 0 10
Chapter 1

• Forecast Relocation:
o Transactional: Allow forecast relocation on transactional format.
o Statistics: Allow forecast relocation of error measures.
o Include Aggregate Series: the user will include aggregate level when the grouping
is enabled.
• Other enhancements:
o Reports: Show fcst separation line in the standard, Executive and Audit report
graph.
o Redesign DRP report to show opinion lines similar to DME.
o Forecast Relocation: Allow forecast relocation of converted lines without unit
forecast
o Events - Reapply event: Interpret null value as zeroes event.
o Events: Change the default base calculation method to Procast.
o Show data source parameters option in Safety Stock for Slow Moving forecast
method.
o When a Regression method is selected, automatically select Correlation Coefficient
and P-Value on Statistics tab.
o This window should be resized when you adjust screen resolution or change the
size of text and other items larger or smaller.
o Save forecast number in data capture as part of Forecast Method tab
o Save forecast periods parameter grouped with forecasting technique when auto
save is on.

Rev 0 11
Chapter 1

Section 3: User Groups

User groups and privileges are dependent upon permissions granted by the System
Administrator. The System Administrator configures ForecastX for individual users or user groups.
Please contact your System Administrator if you have questions about your user permissions.

Section 4: Documentation Conventions

4-1: Text

The following table lists text conventions in this user guide.

Table 1-1: Text Convention Descriptions

Convention Description

Text represented as a screen display. This typeface represents displays that appear on
your terminal screen, for example:
lom>

Text represented as menu or sub-menu This typeface represents all menu and sub-menu
names. names within procedures, for example:
On the File menu, click New.

4-2: Screen Captures

The example screens in this guide may not represent what you see on your monitor; use them
only as guidelines.

Rev 0 12
Chapter 1

Section 5: Acronyms and Abbreviations

Table 1-2: Acronyms and Abbreviations used in this Manual

Term Definition

ANOVA Analysis of Variance

ARIMA Auto Regression Integrated Moving Average

BOM Bill of Materials

ACF and PACF Auto-Correlation Function and Partial Auto-Correlation Function

DF Degrees of Freedom

PO Purchase Order

DME Demand Management Engine

FG Finished Goods

MAPE Mean Absolute Percentage

Error DRP Distribution Resource Planning

IDR Intelligent Data Recognizer

Section 6: Software Editions

ForecastX comes in three different editions, each of these has certain feature limitations
except ForecastX edition which has no limitations.

This is a list of the features that are available in each of the different versions:

Rev 0 13
Chapter 1

Feature Edition
Student Lite ForecastX
On-Fly DRP X
On-Fly Rebalance X
Forecast Adjust X
Forecast Relocation X
Reapply Event X
Conversion Factors X
Lag Accuracy Report X
Unlimited number of items to forecast X X
Saving forecast custom settings (Scenario X X
Management) to specific location.
Report color customization X X
Application Skins X X
Password protecting generated reports X X
ProcastTM Forecast, this method will X X X
analyze your data and select ideal forecast
method.
Grouping Forecast X X X
DRP Reports X X X
Pivot Reports X X X
Reports output saved as .txt or .csv files X X X

Section 7: Software License; Intellectual Property

7-1: Preamble

Any software provided to the customer is subject to the specific license terms and conditions of
the applicable agreement.

In the event of any conflict between those specific terms described below, the specific terms will
supersede and prevail.

John Galt Solutions agrees to those terms and are provided as follows.

Rev 0 14
Chapter 1

7-2: License Grant

John Galt Solutions grants to Customer a limited non-transferable license to use the Licensed
Software subject to the limitations imposed under this Agreement and to the following:

The Permitted Users shall use the Licensed Software only on the equipment located at the
Site. Customer must obtain a supplementary license from John Galt Solutions (which John
Galt Solutions may or may not grant, at its option) before using the Licensed Software.

Customer may make one copy of the Licensed Software in machine- readable form solely for
backup purposes and shall reproduce on any such copy the copyright notice and any other
proprietary labeling of John Galt Solutions that were on the original copy.

Customer shall comply with all restrictions on the use of Licensed Software that Customer is
subject to as a licensee or sub-licensee of John Galt Solutions under the terms of licenses or
other agreements or arrangements by John Galt Solutions with third parties.

7-3: Restrictions

Except as necessary for Customer to exercise its express rights hereunder, Customer may not
itself or allow any third party to (i) make copies of the Licensed Software, (ii) distribute the
Licensed Software to others, (iii) electronically transfer the Licensed Software from one
computer to another over a network, or (iv) decompile, reverse engineer, disassemble, or
otherwise reduce the Licensed Software to a human perceivable form. CUSTOMER MAY NOT
MODIFY, ADAPT, TRANSLATE, RENT, LEASE, LOAN, RESELL FOR PROFIT, DISTRIBUTE,
NETWORK, OR CREATE DERIVATIVE WORKS BASED UPON THE LICENSED SOFTWARE
OR ANY PART THEREOF.

Rev 0 15
Chapter 1

7-4: Ownership of Media

The media on which the Licensed Software is recorded or fixed is Customer's property. If
Customer receives Licensed Software hereunder that renders Licensed Software that Customer
has previously received redundant, Customer will return the redundant Licensed Software to
John Galt Solutions or certify in writing that all copies of such Licensed Software have been
erased.

Section 8: John Galt Solutions Tech Support

http://johngalt.com/support/

To request applications support, open a tech support request in John Galt Solutions Online
Technical Support.

John Galt Solutions provides Online Technical Support so you can get the help you need when
you need it.

With John Galt Solutions Online Technical Support you can:

• Initiate requests for applications support from the web.

• Receive John Galt Solutions rapid, online access to experts to review and respond
to service requests and applications support requests.

• View status of open technical support requests.

• View tech support history and reports about your application, including uptime,
remote fix and service call trends.

Rev 0 16
Chapter 1

Corporate Headquarters:
John Galt Solutions, Inc.
17 N. State Street
Suite 1890
Chicago, IL 60602
(312) 701-9026

Section 9: Revision History

Rev Date Author Reason For Changes

0 3/2018 PM Release for ForecastX 10.0

Rev 0 17
Chapter 2 Installation

Section 1: Overview

This chapter details how to install or uninstall ForecastX™ as well as describing system
requirements to use the application.

Section 2: Requirements

The following sections detail what requirements are needed to install ForecastX™.

2-1: Hardware

The list below details the required hardware to install ForecastX™.

Table 2-1: Hardware Requirements

Since ForecastX is an Add-in to Microsoft Excel, the minimum and recommended requirements are
the same for ForecastX. All requirements may not be listed below, please reference Microsoft’s
requirements for a full list: https://technet.microsoft.com/en-us/library/ee624351.aspx#section7

Hardware Minimum
Processor Intel Pentium 4, AMD Athlon
64 - 1.6 GHz; we
recommend 2.0 GHz or
faster
RAM 2 GB for 64-bit

Hard Disk 3 GB Available

Rev 0 18
Chapter 2

2-2: Software Media

The list below details the required software to install ForecastX™.

1. Microsoft Windows 7 or above (32-bit or 64-bit)

2. Microsoft Excel 2013 or above (32-bit or 64-bit)

3. ForecastX Installer (32-bit or 64-bit, depending on your version of Excel)

Section 3: Uninstallation

This Section describes how to uninstall ForecastX and also describes the compatibility between
ForecastX 10.0 and previous version of ForecastX.

3-1: Backward Compatibility

The new ForecastX 10 is not compatible with any previous version of ForecastX, so before installing
ForecastX you need to uninstall all previous versions.

To remove ForecastX from your computer:

1. Go into the Control Panel and click on 'Add or Remove programs'.

2. Click on the entry named "ForecastX" and click on

'Remove’ or “Uninstall”.

Your computer then uninstalls the application.

3. Click OK.

Note: ForecastX is not compatible with any version of Atlas Planning prior to version 7.4. If you
have any earlier version of Atlas Planning installed, you will need to remove it prior to
installing ForecastX 10.

Rev 0 19
Chapter 2

Section 4: Installation

The following sections detail how to install ForecastX. Before installing ForecastX please close all
instances of Microsoft Excel that might be open.

Before installing ForecastX, please ensure that you are logged in as a user with administrative
rights to your machine, or have someone from your IT department install it. If you do not have
local admin rights, ForecastX will not install successfully.

4-1: Installing ForecastX

Before installing ForecastX, please ensure that you are logged in as a user with
administrator rights to your machine, or have someone from your IT department install it.
If you do not have local admin rights, ForecastX will not install successfully.

1. Locate the downloaded install file on your computer and right click setup.exe to Run-As
Administrator. If newly purchased, the download link was emailed along with the activation
key.
2. Click Next and the ForecastX license wizard will load
3. Click Activate your License
4. Find your Activation Key in the email that was sent when the license was purchased. Enter
the key and press Activate your license. If you don’t have an Activation code available,
please open a ticket with our Helpdesk at https://johngalt.com/support
5. Click Activate Online
6. Enter the End User Registration information. All fields are required. Click Register, then
Next.
7. Enter the Activation Code and click Activate license key, then click Finish
8. Accept the License Agreement and click Next
9. Enter Customer Information and click Next
10. Install the program and Finish
11. NOTE: If you are moving the license to a new computer, choose Deactivate your license in
Step 3.

Rev 0 20
Chapter 2

Rev 0 21
Chapter 3

Chapter 3 Using ForecastX

Section 1: Overview

ForecastX™ is an add-in within Microsoft Excel, which enables ForecastX™ to interact with Excel
seamlessly. The interface for ForecastX™ is designed with the Excel user in mind and follows a
methodology similar to the analysis tool pack.

This chapter provides an overview of ForecastX’s features which enable you to configure your
data and the outcome of your forecasts.

Section 2: ForecastX toolbar in Excel

To ensure that ForecastX™ has been installed successfully, you can access ForecastX™ by
going to:

Start > All Programs > Microsoft Office > Microsoft Office Excel

If you do not see ForecastX™ on the toolbar, you may have to enable ForecastX™ in Excel.

1. If using Microsoft Excel 2013 or later, click on Tools > Add-Ins... from the Excel toolbar. Skip
to “The Add-Ins window displays.” on page 33.

2. If using Microsoft Excel 2013 or later, click on the on the top left corner and select
Excel Options.

If using Excel 2010 or 2013, click on the File menu at the top left and select Options.

The Excel Options window displays.

Rev 0 22
Chapter 3

3. Select Add-ins from the left-hand pane.

4. In the dropdown list next to the Manage option select 'Excel Add-ins' and click 'Go…'

5. Ensure that under “ForecastX” is listed in the 'Active Application Add-ins' list.
If not, select “COM Add-Ins” from the Manage dropdown box. Thereafter, select the
“ForecastX” from the list, select the checkbox next to it and click OK.

Rev 0 23
Chapter 3

6. Click on the Trust Center option in the left pane, and click on the Trust Center Settings
button.

Rev 0 24
Chapter 3

Make sure the following checkboxes are unchecked: “Require Application Add-ins to be
signed by Trusted Publisher” and “Disable all Application Add-ins“

7. Go back to the Excel Options Window, select Add-ins from the pane in the left, select
'Excel Add-ins' from the dropdown list next to Manage text and click the 'Go…'
button.

8. The Add-Ins window displays.

Rev 0 25
Chapter 3

9. Select ForecastX toolbar from the list.

a. If the ForecastX toolbar Add-In is unavailable in the list, click

Browse.

b. In the File Selection window, go to the location where you installed

ForecastX.

c. Select the WizardToolbar.xla file and then click OK.

10. Click OK on the Add-Ins window.

Rev 0 26
Chapter 3

The ForecastX™ toolbar is then displayed in Microsoft Excel. In Excel 2007 & up, it will be under
the Add-Ins ribbon.

The following table details what each feature means.

ForecastX Description
Toolbar

Forecast Main button in the toolbar. This button is always

enabled and it is used to run ForecastX .

On-Fly DRP Used to recalculate DRP based on changes made to

the forecast. This option can be used to perform “What
if” inventory analysis all within the same report.

Note: This feature is only available in ForecastX

Premium Edition.

On-Fly Used to rebalance the hierarchy of a group forecast

Rebalance after the report has been generated without the need to
create a new report. On-Fly Rebalance is only enabled
if the currently selected workbook contains a report
that supports On the fly rebalancing. For additional
information see 6-1: On-Fly Rebalance.

Note: This feature is only available in ForecastX

Premium Edition.

Rev 0 27
Chapter 3

ForecastX Description
Toolbar

Forecast Adjust Used to make adjustment to the forecasted values

after they have been generated. This feature is similar
to the 'Data Adjust' feature inside the Forecast Method
tab.

Note: This feature is only available in ForecastX

Premium Edition.

Forecast Used to copy a report's Forecast to another location.

Relocation Available for Standard or Standard DRP reports only.

Note: This feature is only available in ForecastX

Premium Edition.

Used to save snapshots of prior months’ forecasts

Accuracy and compare them to actual results.
Report
Note: This feature is only available in ForecastX
Premium Edition.

Rev 0 28
Chapter 3

Reapply Event Used to reapply events when using the Event model
forecast method without having to go to the data
source and regenerate the report. Enables you to
modify the event flags in future periods and
recalculate the forecast based in those new events, it
is used as a "What If" analysis tool to see the impact
of different events in the future.

The Reapply Event feature is only available when

the selected workbook contains a Standard report
that uses the Event model forecast method and the
'Allow Reapply' option is selected in the Event
model parameters.

Note: This feature is only available in

ForecastX Premium Edition.

Samples Opens the Samples folder located in the directory

where ForecastX is installed. It contains many
example data sets to assist you with understanding
the features included in ForecastX.

Help Opens the ForecastX User Guide.

11. Click on the button to open ForecastX.

ForecastX™ displays. You will learn more about how to use each feature later in this
guide.

Rev 0 29
Chapter 3

2-1: Forecast Adjust

You can access the Forecast Adjust feature from the ForecastX™ toolbar. The feature can be
used with a Standard Report and any Forecasting method. The Forecast Adjust button is available
for Standard and DRP Standard reports only.

Note: This feature is only available in ForecastX Premium Edition.

To use the Forecast Adjust:

1. After creating a Standard report, click on .

The Forecast Adjust window displays.

The following table details what each feature means.

Rev 0 30
Chapter 3

Feature Description

Data Series to Select which data series you would like to adjust in the
Adjust currently selected report.

Start Date / End Select the start and end dates. The start and end date
Date display the first and last period of forecast for the
selected report. Any adjustments affect the forecast
periods within these two dates only.

Rev 0 31
Chapter 3

Feature Description

Adjust by Value Used to adjust the Forecast by a specific value. To

modify the forecast by a specific amount, enter any
value into the corresponding box and press the '+'
button next to it. For example, if a series has a value of
1000 for a specific month then adjusting it by a value of
50 changes the forecast for that month to 1050, the
change is reflected in the Actual Opinion Line.

Adjust by Used to adjust the forecast based on a percentage of

Percent each observation. To modify the forecast by a
percentage amount enter any value into the
corresponding box and press the '+' button next to it.
For example, if a series has a value of 1000 for a
specific month then adjusting it by 50% changes the
forecast for that month to 1500, the change is reflected
in the Actual Opinion Line.

Range Total Used to span a total across the selected date range.
For example if three periods are selected using the
start and end date and they total 15000 then when the
box is changed to 11000 the new total for those three
periods will be 11000 and it is reassigned to each
period based on how much each period originally
contributed to the original total of 15000, the change is
reflected in the Actual Opinion Line.

12. Click Apply when finished. Click the Reset link to reset the values to
the default.

2-2: Accuracy Report

The Accuracy Report allows users to track the accuracy of their forecasting process by saving their forecasts and
comparing their prior forecasts to actual data from history. Up to 3 periods of prior forecasts can be compared to history.

Note: This feature is only available in ForecastX Premium Edition.

To use the Accuracy Report:

1. After creating a Standard report and making all forecast adjustments, click
.

Rev 0 32
Chapter 3

The Accuracy Report window will be displayed.

The following table details what each control means:

Feature Description

Clear Lags Deletes all saved prior forecasts. Users will be

warned before confirming deletion.

Save Lag Saves a snapshot of the current final forecast, as well

as the historical value for the previous month.

Rev 0 33
Chapter 3

Separate Labels When checked, each label included in the historical

data will appear as its own column in the Accuracy
Report. This enables easier filtering and sorting of
accuracy data.

Show Report Displays the Accuracy Report. The Accuracy Report

will appear as a separate Excel workbook.

2. Click Save Lag. The message “Lag saved successfully” should appear. If Save Lag is
clicked again for the same forecast horizon, users will be warned that they are about to
overwrite a saved lag.

3. Click Show Report. The Accuracy Report appears. Note that only the Actual line is
populated – this is because there are no saved forecasts to compare to the current
actuals.

4. Repeat this process each period when the forecasting process is complete. After 4
periods, the accuracy report will be comparing 3 periods of prior forecasts to their
actual historical values:

2-3: Forecast Relocation

Used to copy a report's Forecast to another location. Available for Standard or Standard DRP
reports only.

Note: This feature is only available in ForecastX Premium Edition.

To use Forecast Relocation:

1. After creating a Standard report, click on

Rev 0 34
Chapter 3

The Forecast Relocation window displays.

The following table details what each feature means.

Feature Description

Opinions This list displays all the available Opinion Lines. The
Opinion Lines vary depending on which ones where
selected for the report.

Layout Used to select the layout of the resulting Forecast

Relocation. You can select to relocate the forecast in
a table (rows or columns), or a transactional format.

Rev 0 35
Chapter 3

Feature Description

Collate This option specifies whether to collate the data by

Opinion Line or by series. When this option is
enabled, the data is sorted by Opinion Line, and when
the option is disabled the data is sorted by series
name.

Statistics When checked, the error statistics included in the

Standard Report will come out in the relocated report.
Only applies when Layout is set to Rows or Colums.

Include aggregate
values When checked and when grouping is enabled,
aggregate-level forecasts will also be present in the
relocated report. This is generally good when
relocating in Row or Column formats, but less
desirable when exporting in Transactional format.

Append at Location Location where the forecast is copied to; it can be in

another spreadsheet in the same workbook or in a
completely different workbook. If no location is
specified, a new workbook is created with the forecast in
it.

2. Click Apply when finished.

Rev 0 36
Chapter 3

Section 3: Quick Start Forecast

Procast™ is ForecastX's™ one-step forecast with expert selection. Procast™ chooses from
complex forecasting techniques, including most time series forecasting methods, new product
growth curves, and Regression models to produce fast, accurate, and easy to read results.

You can run Procast™ on a single data series or create an unlimited batch forecast. Once
Procast™ has been activated, ForecastX determines the best forecasting technique for each data
series in your particular data set.

To open the Procast™ forecasting technique:

1. Click on and open the ForecastXTrainingSet.xls file.

Note: The ForecastXTrainingSet.xls file is a data example to demonstrate how the Procast
method is used. For your company’s purposes, you will have your own data
available.

2. Click the Procast tab at the bottom of the Excel sheet to select the example
spreadsheet.

3. Click in a cell containing data, and open ForecastX™ by clicking on .

ForecastX displays (screen on next page).

Rev 0 37
Chapter 3

4. Click on the Forecast Method tab. By default, the Procast forecasting technique displays.

5. Select Edit parameters to activate Procast’s parameters. The following table details
what each parameter means.

Parameter Description

Error Term The measurement Procast™ uses to determine

which forecasting method is optimal. The best
method produces the smallest value of the
designated statistics.

Rev 0 38
Chapter 3

New Product There are two options for the New Product Curve
Curve parameter:

• Do Not Include — Select this option to remove

the New Product Curve as one of the forecast
methods options.
• Include when num of obs — Select this to
include New Product Curve as a forecasting
method, if the number of observations in the
series is less than the number you specify.

Rev 0 39
Chapter 3

6. Click Finish.

The Forecast Results window displays.

The Forecast Result window enables you to view certain information

about the generated report such as the number of items that were
processed for the report.

7. To accept the Forecast, click the View Output button.

8. To rerun the Forecast, return to the original workbook and click the
Reforecast button. The Data Capture tab displays to allow you to
select other forecasting techniques.

A Standard Report is generated on a separate spreadsheet. The new

report lists the forecasting method used for each data series. The
fitted and forecasted values are listed below the actual values for
each series.

Scroll to the right of the spreadsheet to see the forecasted values

listed in bold. To the right of the forecasted values, ForecastX™
produces four accuracy statistics that help you determine the
accuracy of the forecast.

Rev 0 40
Chapter 3

The following table details what each statistic means.

Statistic Description

Mean Absolute The overall average of the error measurements

Percentage Error ForecastX™ produced.

R-Square Determines how well the fitted values compare to the

actual values.

Mean The average value of each series.

Standard This is a measure of how widely are the values in a

Deviation group of data spread. It is defined as the square root of
the variance

3-1: Data Cleansing

As an advanced option, you can use Data Cleansing to improve the

accuracy of a Forecast. For additional information, see Section 4: Data
Capture Tab.

3-2: Group Forecasting

As an advanced option, you can group forecast through the Group By tab.
Group By tab provides access to group forecasting’s features and
options. For additional information, see Section 6: Group By Tab.

Rev 0 41
Chapter 3

3-3: Event Modeling

If there are special occasions or promotional events, ForecastX™ takes

them into account while developing your Forecast. When you are able to
identify major promotions that impact sales, you can mark these
occasions and ForecastX can enhance the forecasts based on the
promotional plan.

Note: The following steps provide instructions for running a forecast without
Event and Promotional models, as well as steps with Event and
Promotional models. Thereafter, you can compare the results.

1. Click on and open the ForecastXTrainingSet.xls file.

Note: The ForecastXTrainingSet.xls file is a data example to demonstrate

how the Procast method is used. For your company’s purposes, you
will have your own data available.

2. Click the Event Modeling tab at the bottom of the Excel sheet to
select the example spreadsheet.

The Event Modeling sheet contains a column of Newspaper Sales

and the Marketing Plan associated with those sales.

3. Highlight columns A to B, and open ForecastX™.

4. On the Data Capture tab, ensure that ‘columns’ is selected in the

Data is organized in area.

5. On the Forecast Method tab make sure Procast is the selected

Method.

6. On the Reports tab, uncheck the Standard option and select the Audit
report.

Rev 0 42
Chapter 3

7. Click Finish.

ForecastX™ displays an Audit report and a graph of the results. You

can see the forecast results and the error measurements without
Event Modeling performed.

3-3-1: Event and Promotional Modeling Forecasting

To forecast using Event and Promotional Modeling:

1. Click on and open the ForecastXTrainingSet.xls file.

Note: The ForecastXTrainingSet.xls file is a data example to demonstrate

how the Procast method is used. For your company’s purposes, you
will have your own data available.

2. Click the Event Modeling tab at the bottom of the Excel sheet to
select the example spreadsheet.

Rev 0 43
Chapter 3

3. Highlight the Months and Newspaper Sales data found columns A to

B, and open ForecastX™.

ForecastX displays.

4. On the Forecast Method tab, select the Event Model forecasting

technique and select Edit parameters to activate the Event Model’s
parameters.

5. Define the event flag data range by highlighting all of column C.

6. Click OK.

The Forecast Method window displays.

7. On the Report tab, uncheck the Standard Report option and select
the Audit option.

8. Click Finish.

Rev 0 44
Chapter 3

On the Audit Report, you see the effect of the events on Newspaper
Sales and how your error measurements improve. The Audit Trail
report provides you with additional statistical data, including the event
lift factors.

This report includes an Event Index, which represents each events lift
factor over the base forecast.

By comparing the graphs from Procast™ and Event Model, you can
understand the effect of the events on Newspaper Sales, and see
how the error measurements improve.

Note: You can select the Advanced button in Event Modeling to choose
forecasting techniques, and perform a Monte Carlo analysis for best
case/worst case analysis on each event. Please see Section 12:
Event Model for additional information.

Rev 0 45
Chapter 3

Section 4: Data Capture Tab

The Data Capture tab is used to provide ForecastX with information

on where and how your data is configured among other information..

The table below details what each feature means:

Feature Description

Data is Specify whether your data is in columns or rows. Your

organized in data needs to be arranged in a consecutive line either
by rows or columns. To change the selection, click the
appropriate radio button.

Rev 0 46
Chapter 3

Feature Description

Forecast periods Specify how many periods you want to forecast. The
default is 12 periods. ForecastX™ provides the ability
to forecast up to 9999 periods.
You can enter the number of periods you want to
forecast or use the up and down arrows.

Seasonality Enter the seasonality of your data, or allow

ForecastX™ to automatically detect any seasonality.
For example, if data contains quarterly seasonality,
you could enter 4 in this text box, or leave it blank to
see if ForecastX™ detects the quarterly pattern.
You can enter the seasonality, or use the up and down
arrows.

Data to be Specify the data range to include in the forecast. As

forecast ForecastX opens, the Intelligent Data Recognizer
(IDR) selects the data range automatically. The IDR
will attempt to interpret the layout of you data. It will
guess wheather the data is organized in columns or
rows, the data range to include in your forecast,
whether the data inclused dattes, and how many
descriptive labels exist.

To manually specify the data range:

• Click inside the text box. The Data Selection box
appears.
• Click inside the Data Selection text box. Select a
data range in the spreadsheet by clicking and
dragging, or using Excel short cut keys, such as
shift, control, down arrow key.

Note: Your data range should not include Event Indices

used in Event Modeling or Day Indices used in Census
X-11.

Note: The data to be forecasted can also have multiple

range selections, both column and rows. For example
if you only wanted to forecast some of the items in your
dataset you would first select the row with the headers
and then while pressing the Crtl key you could click and
select the rows for the items you wanted to forecast.

Rev 0 47
Chapter 3

Feature Description

Data set Specify whether your data contains dates and any
descriptive labels. If you have Excel formatted dates,
the IDR will identify them, and then select the correct
time frame.
To manually specify that your data includes dates:
• Contains Dates - click Yes to specify the dates.
• Periodicity - select Hourly, Daily, Weekly, 4 Weeks,
Quarterly or Annual to determine the period.
• Last Historical Date - specify this date if the IDR
incorrectly reads you dates, or if your data does
not contain dates. Enter the last Historical date
from your data in this field so the forecast is
created from that date forward.
• Labels - enter the number of labels to be
implemented. ForecastX can handle up to 20
labels.

Note: You can enable the IDR to specify how

many descriptive labels your data contains, or do
it manually.

Note: When labels are numeric, the IDR thinks the

labels are numbers to be forecasted. This is
typically the reason for editing the IDR selection.

• Parameters - Indicates the number of columns that

may be used to make additional calculations.

4-1: Data Cleansing

The Data Cleansing window contains methods to clean data and optimize
the accuracy of your forecast. To open the Data Cleansing window:

1. From the Data Capture tab, click on the Data Cleansing link.

Rev 0 48
Chapter 3

The following table details what each feature means.

Feature Description

Replace Outliers Select Yes if you would like to replace Outliers in your
data. An Outlier is an exception (i.e., an anomaly in
your data) that distorts the accuracy of your forecast.
For example, a snowstorm in Texas in the middle of
July that affects your snow shovel sales is an Outlier.

• St Dev — The distance between a data point and its

corresponding fitted value is used to determine
whether or not a data point is an Outlier. The
distance is measured by the number of standard
deviations and the default is 2. Therefore, if the
distance is larger than two standard deviations, then
the data point is an Outlier.
• Forecasting Technique — Select the forecasting
technique you would like to use to perform outlier
replacement. ForecastX™ provides the ability to
choose from seasonal, non-seasonal, and growth
techniques.

Rev 0 49
Chapter 3

Replace Values Indicate whether or not you would like to replace

missing or erroneous values in your data.
• Lower — Indicate the low end of the range of values
you want replaced.
• Upper — Indicate the high end of the range of
values you want replaced.

Occasionally, data is incomplete; a period or a range of

periods is not reported. Instead of leaving it blank,
ForecastX™ replaces it with a value calculated by
Exponential Smoothing.ture Description

Remove Zero Indicate the zeros that you want removed from your
Values data.
• Leading — Select this option to remove leading
zeros.
• Trailing — Select Trailing to remove trailing zeros.

Frequently during Batch Forecasting, not all items

have data for all time periods. ForecastX™ provides
the ability to ignore the blank or zero values that occur
before or after the data series.
This is beneficial in a situation where a new product is
being introduced and there are several periods of
zeros where the product is non-existent.

Trend Adjust Trend Adjust increases accuracy by adjusting the trend

of the forecasted values. For some data, the trend in
forecasted values is very steep (in both increasing and
decreasing directions). The Trend Adjust option can
reduce the steepness of the trend to an acceptable
level.
To control the adjusted steepness, a control point is
introduced. At the control point, the trend reduces to
the 1% of its original value.
• Apply Trend Adjust — Select this option if you would like to
use Trend Adjust.
• Growth Period — Specify the period of the control point
if the trend in the forecasted value is increasing.
• Decline Period — Specify the period of the control point
if the trend in the forecasted value is decreasing.

Rev 0 50
Chapter 3

Feature Description

Tracking Signal Select this option if would like to apply a tracking signal
to your forecast.
The tracking signal is used for fine-tuning the overall
accuracy of your Forecast. The tracking signal
determines whether or not your fitted values are under
or over forecasting.
The fitted values are the baseline of your forecast.
When comparing it with the overall MAPE, it improves
the accuracy by less than a percentage point.
• Apply Tracking Signal - Select this option if you
would like to use Tracking Signal.
• Under Forecast Percentage — Specify the under
forecast percentage you want to track.
• Over Forecast Percentage — Specify the over
forecast percentage you want to track.

Range Forecast Select the Activate checkbox if you would like to set the
allowed minimum and maximum forecast values.
• Check Hist - validate the floor and ceiling values
gains the history. The lowest and highest history
values are compared to the floor and ceiling values. If
any history values are higher than the ceiling value,
the ceiling values are ignored. In addition, if the
lowest history value is below the floor value, the floor
value is ignored.
• Floor Value - minimum forecast value.
• Ceiling Value - maximum forecast value.

Section 5: Forecast Method Tab

ForecastX™ enables you to select specific Forecasting techniques and

parameters. Techniques include time series, promotional, regression, and
growth curve models. For additional information in regard to the Forecast
Method feature, please see Chapter 4 Forecast Methods.

To select a Forecasting technique:

1. From ForecastX™, click on the Forecast Method tab.

Rev 0 51
Chapter 3

The Forecast Method window displays with the Procast forecast

technique shown as the default.

2. From the Forecast technique dropdown list, select one of the

techniques available.

The features in the Actions section enables you to analyze and

manipulate your data with several tools. Each tool is explained in the
following sections.

5-1: Transform

There are several techniques in ForecastX™ that test the data to see if a
data transformation is necessary. However, ForecastX™ also enables
you to view a Variance chart. Data Transformation enables you to chart
the variance to see if the variance of the data is constant. For additional
information, please see “Variance” on page 213..

Rev 0 52
Chapter 3

1. From the Actions area on the Forecast Method tab, click on the
Transform link.

The Data Transformation window displays.

2. From the Data Series to Display dropdown, select a data series you
would like to display.

3. From the Transformation Applied to All Series area, select a

transformation method.

4. Click OK when finished.

5-2: Adjust

The Adjust feature enables you to view a graph of your actual data and
perform adjustments, as well as adjust data by either a percentage or a
value.

1. From the Actions area on the Forecast Method tab, click on the Adjust
link.

Rev 0 53
Chapter 3

The Data Adjustment window displays.

2. In the Data Series to Adjust dropdown, select the data series you
would like to adjust.

3. In the Start Row and End Row dropdowns, select the range of data
that you would like to adjust.

4. Determine how you would like to adjust your data by either entering a
number in the Value or Perce textboxes. To apply an adjustment
press the "+" button next to the adjustment type used.

5. Click OK when adjustments are done to apply them.

5-3: Analyze

The Data Analysis feature enables you to test if a data set is stationary by
performing seasonal and non-seasonal differencing. This test is important
test for ARIMA modeling.

1. From the Actions area on the Forecast Method tab, click on the
Analyze link.

Rev 0 54
Chapter 3

The Analyze window displays.

2. In the Data Series to Analyze dropdown, select the data series you
would like to adjust. An ACF and PACF graph displays.

3. Set the non-seasonal or seasonal differencing. The graph reflects the

changes the differencing makes.

4. In the Num of Bars textbox, type in the number of bars you would like
displayed on the graph.

5. Click Redraw to get an updated graph if the 'Num of bars" parameter

has been changed.

6. Click Export if you would like to export the graph to a new Excel
sheet.

7. Click OK when finished.

Note: For more detailed information performing ARIMA analysis please go

read 4-1: Performing ARIMA Analysis

Rev 0 55
Chapter 3

5-4: Preview

The Preview feature enables you to get a graph preview of your currently
selected data.

1. From the Actions area on the Forecast Method tab, click on the
Preview link.

The Preview window displays.

2. In the Data Series to Analyze dropdown, select a data series. By

default the Excel sheet that is open is listed in the textbox.

3. Click OK.

Section 6: Group By Tab

The Grouping feature is beneficial when you would like to view product-
family forecasts. For example, if a group of products has an especially
small demand individually, it may be more practical to forecast the group
as a whole (or one total product) in order to see a more accurate forecast.

Rev 0 56
Chapter 3

There are two different methods for grouping your items:

Group Forecast Description

Grand Total When this option is selected no hierarchy can be

defined. Using this option will generate only one
grouped total that includes all the items in your data
set.

Hierarchy When the Grand Total option is not selected you can
defien a hierarchy using the labels configured in the
Data capture window. This will create totals for each of
the different values within those selected labels.

There are two features that are beneficial in Group Forecast:

Feature Description

Multi-Level Up to three levels of grouping can be supported, the

Grouping number of supported level is defined by the ForecastX
Edition you are running, for more information see
Section 6: Software Editions 12. Label information
must be preloaded prior to grouping.

On-fly This feature allows any changes made to the forecast

Rebalance to be distributed accordingly throughout the levels of a
defined hierarchy when a Report uses the Group by
option. On-fly rebalance can be ran after the report has
been created.

To perform a Group forecast:

1. Click on the Group by tab.

Rev 0 57
Chapter 3

The Group by window displays.

2. Select Grand Total to have only one group; all Series are grouped into
one Series. If you select this option, no hierarchy is available.

3. In the Hierarchy area, you can drag and drop your column headers to
select the fields you want to group your data by; up to three different
fields to group by fields simultaneously can be selected, but the
number of columns you group your data by cannot be the same as
the number of labels configured in the Data Capture section.

Note: The available column can be selected in any order, they do not need
to be the in the same order as your data.

Rev 0 58
Chapter 3

4. In the Allocation type area, select whether you would like to perform a
TopDown or BottomUp forecast.

Allocation Description

TopDown Sums up the detail level series, forecast in the highest

level, and then allocate the forecasted values down to
lower level.

BottomUp Forecasts the detail level first, and then sums up both
the Historical and Forecasted values to the Aggregated
level.

5. In the Allocated by area, select if you would like your forecast to be

allocated by either Historical or Forecasted values.

Allocated by Description

Historical Uses the historical weight of each series to calculate

the Allocation percentage. All Forecasted periods for a
Series use the same percentage.

Forecasted Uses the forecasted values in the lower level to

calculate the Allocation percentage. Each forecasted
period for a series has a different percentage. It is
enabled regardless of the Allocation type.

6. In the Other options area, select one or more options if you would like
to:

Feature Description

Include Include the Allocation percentage in the Standard

percentage report. This feature is only available for "Top Down"
Allocation type.

Remove zeros Removes zeros when forecasting and calculating the

Allocation percentage. When selected, leading zeros
are removed prior to forecasting, as well as calculating
the Allocation percentage for the Historical Allocated
by Type.

Rev 0 59
Chapter 3

Feature Description

Allocate Used to distribute any remaining top-level forecast

difference based on the allocation percentage for each Record.

Do not report Determines whether or not the Detail Series is

detail reported. This feature is applies to Bottom up
Allocation type.

7. Click Finish.

6-1: On-Fly Rebalance

ForecastX can rebalance hierarchies without reprocessing your data. If

On-Fly Rebalance is enabled, after you generate you report that has a
hierarchy applied to it, any changes made to the upper levels can be
driven down to the lower levels and vice versa without generating a new
report. The On-Fly Rebalance is disabled by default; To enable the On-
Fly Rebalance, click on the Standard tab in the Report tab and check the
On-Fly rebalance option.

Note: This feature is only available in ForecastX Premium Edition.

If On-Fly Rebalance is enabled and a Standard report has been

generated, the On-Fly Rebalance button is enabled in the toolbar. This
button is used to rebalance the created hierarchy after changes have
been made to the Forecast, so that any changes made to the forecast at
any level can be appropriately driven down or up. There are two different
ways to rebalance a hierarchy:

Rev 0 60
Chapter 3

Option Description

Based on Any adjustments made to the forecasted values in the

selected series selected series are Allocated down to the lower level
and summed up to the higher level.

A dropdown list with all Series is provided. The current

activated Series is selected by default. If no Series is
being worked on, then the first series is selected. When
used, the new forecasted values in this Series are
allocated down to the lower level and summed up to
the higher level.

Based on whole Any adjustments made to the forecasted values in the

level selected level are Allocated down to the lower level
and summed up to the higher level.

A dropdown list with all Levels is provided. The current

activated Series is selected by default. If no Series is
being worked on, then the first series is selected. When
used, the new forecasted values in this Series are
allocated down to the lower level and summed up to
the higher level.

Section 7: Statistics Tab

ForecastX™ enables you to select statistical analyses to include on

Audit Trail reports. ForecastX™ supports more than 40 statistics.
ForecastX™ supports three different categories of statistics:

• Accuracy

• Analysis

• Regression

Rev 0 61
Chapter 3

The Advanced Statistics window offers advanced statistics for

experienced statisticians. The matrix level statistics and the statistics for
Regression Model are also located in the Advanced Statistics area. To
access the Advanced Statistics, click the More Statistics link on the
Statistics tab. For additional information in regard to statistics, please see
Chapter 5 Statistics.

The depiction below is the Statistics window.

Section 8: Reports Tab

ForecastX™ offers six different types of reports with specific options for
each. You can choose which report to generate from the Reports tab.
However, you can also select more than one report to generate based
on your data. Reports are created in a new Excel workbook. Each report
has its own workbook, but may contain several Series within one
workbook.

Rev 0 62
Chapter 3

Note: Take into account that whenever reports that support on fly
operations (On-fly DRP or On-Fly rebalance) are created at the same
time they will share the same forecast until they are saved and opened
separately.

8-1: Standard Report

The Standard Report is built for speed and handling large volumes of
data. It produces a side-by-side report listing the actual values compared
to the forecasted values. It also includes selected statistics and a few
common statistics: Mean Absolute Percentage Error (MAPE), R-Squared
Value, and Standard Deviation.

Rev 0 63
Chapter 3

The table below details what each Standard report feature means.

Feature Description

Layout This feature is used to determine what the layout of the

generated report will be:
• Rows - data in the report is arranged in rows.
• Columns - data in the report is arranged in columns.

Chart There are two different chart options:

• Classic - selecting this option generates a regular
Standard report that includes the forecast in the first
sheet and it includes a chart for each of the series in
each subsequent sheet.
• Standard - selecting only the Standard option
generates a report that only includes a sheet with
the forecast for the selected data and no charts. If
you wish to include charts in your Standard report
there are two options:

• Chart in table - when this option is selected a

chart is included in the Standard report in the
sheet where the forecast data is. This allows
having your data and the graph in a single
sheet, and there are buttons included to
navigate to the different series in the graph.

• Chart in worksheet - when this option is

selected, each of the series in your data
receives its own chart in a new sheet.

On-Fly On-Fly rebalance is used to specify if the Standard

rebalance report should support On-Fly rebalance. When this
option is selected the report that is generated supports
this feature and the 'On-Fly Rebalance' button is
enabled for that report. For more information see the
On-Fly rebalance section from Group by tab.

On-Fly rebalance is disabled if the 'Group forecast'

option is disabled in the Group by tab.

Note: This feature is only available in ForecastX

Premium Edition.

Rev 0 64
Chapter 3

Feature Description

Opinions Specifies which Opinion Lines should be included in

the Standard report. There are four Opinion Lines
available:
• Error - select this option to include the Error
Opinion Line. For additional information, see the
Error or Residual sections in Chapter 5 Statistics.
• Upper Limit - includes the highest forecast value
based on the calculation of standard deviation
on reports.
• Lower Limit - includes the lowest forecast value
based on the calculation of Standard Deviation
on reports.
• Safety Stock - includes the calculated Safety Stock
value on the report.ription

Options Options are used to specify different options that can

be included in the Standard report. There are four
different types of Options that can be used:
• Auto Filter - Select this option to add the auto filter
ability to the standard report.
• Separate Labels - Select this option to separate
labels in multiple cells as in the input data.
• Track Change - Select this option to add a macro
to the Standard report which tracks any change
and support the consolidation function. Enabling
this option will include a button named 'Reconcile'
in the standard report.

Note: When this option is selected, the Standard

report's worksheet is protected, only the Adjust
opinion line is editable.

• Tracking Tables - enables you to make graphical

and table changes to the forecast. It also enables
you to compress or hide the Historical data so you
can focus on the forecasted values only. This will
include a button in the Standard report that can
change from the three different states: Normal,
Compress and Hide. For additional information,
see the 8-7: Reports Advanced Options.

Rev 0 65
Chapter 3

• Report Outlier - used to specify if outliers should be

reported or not. If this option is selected, any outliers
replaced by ForecastX’s engine are noted in the
report; any cell containing an observation point
which was replaced by the 'Replace Outliers'
feature includes a comment in it, that comment
includes the original value that observation point
had before the replacement. For additional
information, see '4-1: Data Cleansing.

Note: This option is disabled if the 'Replace

Outliers' option from the Data Cleansing section is
not enabled.

Rev 0 66
Chapter 3

8-1-1: Standard Report Grouping Example

The example below details a running a Standard Report in order to

explain the output.

1. Click on and open the ForecastXTrainingSet.xls file.

Note: The ForecastXTrainingSet.xls file is a data example. For your

company’s purposes, you will have your own data available.

2. Click the Group Forecasting tab at the bottom of the Excel sheet to
select the example spreadsheet.

3. Click in a cell containing data, and open ForecastX™ by clicking on

ForecastX displays.

4. On the Data Capture tab, ensure the Labels are set to 2 and the DRP
Parameters set to 0.

5. From the Group By tab, and check the Group Forecast option to
enable Group Forecasting.

Rev 0 67
Chapter 3

6. Left-click on the Region label header and drag it to where it says

"drag a column header here to group by that column."

7. Repeat the same procedure with the Category label header

Note: Multi level grouping is a feature only available in ForecastX

Premium Edition.

Rev 0 68
Chapter 3

8. From the Reports tab, select the Standard checkbox and ensure it is
the only one selected.

9. Ensure that the Active checkbox under the On-Fly rebalance is

selected.

Note: On-Fly rebalance is a feature only available in ForecastX

Premium Edition.

Rev 0 69
Chapter 3

10. Click Finish.

Rev 0 70
Chapter 3

11. Click on View Output.

As shown in the example above, generating a group report creates

several Opinion Lines. Below is a brief description of each of the
elements that will help you handle the grouped report.

This report shows that there are 3 different regions in our data East,
North and West and that each of the groups has 4 items, 3 items and
3 items, respectively.

To expand a whole level, use the numbers on the top left of the row
numbers. To expand a single item, use the "+" sign on the left of each
item.

8-2: DRP Report

The DRP report provides information needed to complete the Distribution

Resource Planning process. Data on this report is controlled by the 'Drp
Parameters' option on the Data Capture screen as well as options on the
DRP Options screen. The ‘Parameters' option in the Data capture tab
indicates how many of the columns from your source data correspond to
key values in the planning process.

Rev 0 71
Chapter 3

The table below details what each DRP report feature means.

Feature Description

Standard The DRP Standard report is similar to the Standard

report but it includes the following Opinion Lines:
Recommended Order, Beginning Projected
Inventory, Stat Forecast, Adjust, Recommended
Order Arrival, Ending Inventory, Min Line, Max Line,
Periods Covered, and Unsupported Fcst.

The workbook in which the Standard DRP report is

generated in is protected; the only editable opinion
line will be the Adjust line.

Select Active to arrange how the data is arranged in the

worksheet.

Select the Active checkbox if you want to generate a

Standard DRP report. When selected, you can
configure the Standard DRP report.

Layout - determines whether the report data should be

in rows or columns.

Rev 0 72
Chapter 3

Feature Description

Standard • Select On-Fly DRP - indicates whether the

(continued) generated report will have the ability to
recalculate DRP without having to regenerate the
report.

Note: This feature is only available in ForecastX

Premium Edition.

When this option is disabled the report that is

generated is static, this means that in order to
make any changes to the reports outcome you
must go back to the data source and edit the
historical data to impact the forecast. On the other
hand, when the 'On-Fly DRP' option is selected
the report generated by the application is
dynamic, this means that after a report has been
generated you can edit the forecast for any item
and recalculate DRP using the newly input values,
all without having to create a new report.

To recalculate DRP in a 'On-Fly DRP' report you

just input the desired adjustments into the 'Adjust'
opinion line, and then click on the 'On-Fly DRP'
button located in the ForecastX 7 toolbar.

• Select On-Fly rebalance - only available

when the 'Group Forecast' option has been
enabled in the Group by tab. This specifies if
the Standard DRP report can be rebalanced
once the forecast has been adjusted.

Note: This feature is only available in ForecastX

Premium Edition.

Rev 0 73
Chapter 3

Feature Description

Pivot If this report is active, it is created inside a new

spreadsheet named in a sheet named 'DRP Pivot
Report'. It is similar to the Pivot report but it includes all
the Opinion Lines included in the DRP Standard report,
the Opinion Lines included are: Begin Projected Inv,
End Projected Inventory, Lower Limit, Periods
Coverage, Recommended Orders, Safety Stock, Stat
Forecast, Turns, Unsupported Forecast and Upper
Limit Select Active.

Select the Active checkbox to generate a DRP Pivot

report. When selected you can configure the DRP
Pivot report in two different ways:
• Chart in table - a chart using the pivot table will be
included above the Pivot table in the same sheet.

• Chart in worksheet - a chart using the Pivot table

is included in a different sheet that the one the
DRP Pivot table is in.

The DRP Pivot report is always generated into a

new workbook in a sheet named 'DRP Pivot
Report'; it is very similar to the Standard pivot
report but there are some differences. The opinion
lines included in this DRP Report are the
following: Recommended Order, Beginning
Projected Inventory, Stat Forecast, Adjust,
Recommended Order Arrival, Ending Inventory,
Min Line, Max Line, Periods Covered, and
Unsupported Fcst. The second difference is that
the DRP Pivot report only generates data for
forecast periods; no Historical data is included in
this report.

8-2-1: DRP Settings

Once the number of Drp Parameters has been loaded in the Data
Capture tab, those parameters need to be configured.

Rev 0 74
Chapter 3

1. Click on the DRP Settings link on the Reports tab.

Rev 0 75
Chapter 3

The following table details what each option means.

Tab Description

DRP Used to specify the DRP Settings. Each parameter can

be configured in two different ways:

• Using a default Value can be used for each

parameter. To use the default value for a
specific parameter select the "(None)" option in
the Input column and enter the desired default
number or option in the Default Value column.

• Using an Input Parameter for each parameter.

To configure it is to use the DRP Parameter
included in the source data, they have to be
mapped to their corresponding field in this
configuration. To configure a DRP parameter
from the original data you must select the
corresponding parameter and fill the correct
Label in the Input column. In this case, the default
value for that specific parameter is ignored and
disabled. The dropdown list available for each of
the parameters will only display parameters
identified in the 'DRP Parameters' box in the
Data Capture tab.

Note: If the list of available parameters does not

include parameters that do exist in your data, verify the
'DRP Parameters' option in the Data capture window is
correct. See Section 4: Data Capture Tab for additional
information.

Rev 0 76
Chapter 3

Tab Description

DRP (continued) The Opinion Lines below are used in the DRP
calculations. They are the only Opinion Lines that can
be configured using data included in your Excel
workbook:

• Initial On Hand - the Initial On hand is the

current inventory value for each Record.

• Min. Period Cov. - this value represents the

number of periods of future demand that should be
available at the end of each period. Used if Min
Cvg Type is C.

• Reorder Point - this is a minimum value to never

drop below. Used if Min Cvg Type is set to R.

Max Per Cvg – this is the number of forecast

periods to order up to.

• Order Multiple - this is the order multiple used

when placing a replenishment order. Any orders
recommended by the system are multiples of this
number.

Min Order Qty – this is the smallest order allowed.

Max Order Qty – this is the largest order allowed.

• Use ROP - this flag indicates whether to use ROP to

calculate replenishment plan or not. The only valid
options for this parameter are Yes or No, regardless
if it used using the default value or a parameter from
your table.

• Min Qty Type – this flag determines which option

will set the Min Line in the report. There are three
options: S, which uses a statistical safety stock,
R, which uses the unit quantity in the Reorder
Point column, and C, which is based on the
periods of coverage defined in the Min Per Cvg
column.

Rev 0 77
Chapter 3

Rev 0 78
Chapter 3

The following table details what each option means.

Tab Description

Safety Stock This section is used to determine how Safety Stock is

calculated, there are three different ways to set Safety
Stock:

• Input Parameters - select this option if the

Safety Stock Lead Time is to be calculated by
ForecastX.

• Include Stat Fcst - this flag indicates that the

forecast value plus the Safety Stock value is
to be reported; this means your Safety Stock
will be a Stocking Level. The only valid
options for this parameter are Yes or No,
regardless if it used using the default value or
a parameter from your table.

Note: Using a parameter for this option or:

• Lead Time - this option defines the

SafetyStock Lead Time.

• Service Level - this value is a target of

demand met. This target is converted to a
multiplier, which in turn affects the safety
stock value. The higher your service level
target, the more Safety stock must be
carried to cover forecast error or unexpected
demand. If you did not select the 'Same as
Safety Stock Lead Time' option in the DRP
section, click on the Input dropdown and
select the data field that contains the safety
lead time value for each Record.

Rev 0 79
Chapter 3

Tab Description

Safety Stock • Slow Moving Output - automatically selected

(continued) when using the "Slow Moving Demand" forecast
method. The Safety Stock calculated by the
Forecast Method is used for the DRP Report.

• Input Value - select this option if the Safety Stock

Lead Time is included in the original data or if
you want to set a default value for the Safety
Stock parameter.

o The 'As a stocking level' is used to force the

beginning inventory for each period to be
equal to the Safety Stock. If this option is not
selected, the ending inventory for each period
to be equal to the Safety Stock. Enabling this
option will include your Safety Stock in your
Recommended Orders.

8-2-2: DRP Example

This section details an example of running a DRP Report to explain the

output.

1. Click on and open the ForecastXTrainingSet.xls file.

Note: The ForecastXSet.xls file is a data example. For your company’s

purposes, you will have your own data available.

2. Click the Procast DRP tab at the bottom of the Excel sheet to select
the example spreadsheet.

3. Click in a cell containing data, and open ForecastX ™ by clicking on

ForecastX displays.

Rev 0 80
Chapter 3

4. In the Data Capture tab, make sure your Labels are set to 2 and the
DRP Parameters set to 10.

5. From the Reports tab, select the DRP checkbox and ensure that it is
the one selected.

6. Click on the 'DRP Settings' link to configure the DRP Parameters.

7. Configure each of the Input Parameters using the available labels

options in the dropdown lists. Enter the last 2 parameters on the
Safety Stock tab.

8. Click OK.

9. Click Finish.

Rev 0 81
Chapter 3

The Forecast Results window displays.

10. Click View Output.

Generating a DRP Report creates several Opinion Lines. The table

below details what each Opinion Line means.

Opinion Line Description

Rec. Order Displays the order quantity recommended by

ForecastX.

Begin Proj. Inv. The inventory level at the beginning of the period,
either based on the current on-hand level or the
forecast consumption.

Forecast The statistical forecast value to begin forecasting.

Adjust Allows users to adjust the forecast.

Rec. Order Arr. Shows how much inventory is projected to arrive in

this period, based on order recommendations and
lead time.

End Proj. Inv. The inventory level at the end of the period, based on
forecast consumption and order arrivals.

Min Line Displays the lowest inventory level allowable based

on the Min Inv Type and input value.

Max Line Displays the target inventory level based on the

number of periods in the Max Per Cvg field.
.

Rev 0 82
Chapter 3

Opinion Line Description

Periods Cov. Shows how many periods of demand are covered by

the inventory value in Begin Proj. Inv.

Unsupported Fcst Shows where demand exceeds inventory for a given

period, and the quantity that is unsupported.

8-2-3: DRP Reports and Group Forecasts

DRP Reports can be generated with or without using Group By Forecast.

The table below details how the Opinion Lines in the aggregate levels
behave when a DRP Report is generated and the Group Forecast is
selected.

Opinion Line Description

Initial On Hand Calculated as a sum of all the child series contained in

that level.

Min. Period Calculated as an average of all the child series

Coverage contained in that level.

Note: When rounding numbers, anything from .6 and

above is rounded to the next integer. For example, 2.3
is rounded to 2, and 2.6 is rounded to 3.

Reorder Point Calculated as a sum of all the child series contained in

that level.

Reorder Quantity Calculated as a sum of all the child series contained in

that level.

Rev 0 83
Chapter 3

8-3: Executive Report

The Executive Report is used for those who are responsible for your
company’s forecast. The Executive Report creates adjustable graphs,
charts, and tables, and includes a few common statistics such as MAPE
and R-Squared values. Tables include a complete list of Forecast values
in monthly, quarterly, and annual form.

Rev 0 84
Chapter 3

The table below details what each Executive report Opinion means.

Opinion Description

History Includes a plot of the Historical values on the graph. It

is checked by default and it is always disabled to the
user.

Fitted Includes a plot of the fitted Forecast values on the

graph.

Forecast Includes a plot of the forecasted values on the graph. It

is checked by default and it is always disabled to the
user.

Safety Stock Includes a plot of the Safety Stock values on the graph.

Lower Limit Includes a plot of the lower limit on the graph.

Upper Limit Includes a plot of the upper limit on the graph.

Error Includes a plot of the Forecast error values on the

graph.

8-4: Audit Report

The Audit Trail Report generates a detailed analysis of the Forecast. It

displays all the information provided in the Executive Report, as well as

Rev 0 85
Chapter 3

the statistics that were chosen and optional tables. The Audit Report is
optimal for justifying Forecasts with statistics.

The table below details what each Audit Trail report Option means.

Option Description

Forecast Results Includes the forecast table in the Audit report, it is

checked by default and it is always disabled to the
user.

Include Include an Executive report within an Audit report. It is

Executive checked by default and it is always disabled to the
user.

Fitted Values Includes a table of the Forecast fitted values.

Table

Rev 0 86
Chapter 3

Option Description

Out of Sample If this option is checked the report will include a table of
Table holdback statistics. You must specify the number of
periods to hold back from you analysis.

Note: The Out of sample option will be disabled in the

following situations; when Grouping is enabled or when
one of the following forecast methods is selected:
Census X11, Decomposition, Erlang B, Erlang C,
Event Model, Multiple Regression, New Product
Forecasting, Poisson, Polynomial Regression or
Stepwise.

Rev 0 87
Chapter 3

8-5: Pivot Report

The Pivot Table enables you to leverage your Excel skills and create a
Pivot Table that displays the results in the form you choose. It is an
inherent ability of Excel, which enables you to pivot on certain fields. In
ForecastX™, these fields are the labels utilized in your Forecast. By
selecting from the drop-down menu, you can pivot, or alter the display,
to view a particular series.

Rev 0 88
Chapter 3

The table below details what each Pivot report feature means.

Feature Description

Chart These are options regarding the charts that can be

generated with the pivot report, if no option is selected
no chart will be included with the Pivot report: There
two types of Standard charts:
• Chart in table - when selected, a Pivot chart is
included in the same worksheet as the pivot table.
• Chart in worksheet - when selected, a pivot chart is
included in a different sheet within the same report.

Feature Description

Opinions Used to define which Opinion Lines are available in the

Pivot report:
• History - includes Historical data.
• Forecast - includes Forecast data.
• Error - includes the error opinion line.
• Safety Stock - includes Safety Stock Opinion Lines.
• Lower Limit - includes the lower limit Opinion Line.
• Upper Limit - includes the upper limit Opinion Line.

Note: When an Opinion Line is added to be included in

the Pivot report it can be hidden from the reports
option, but if an Opinion Line is not included in the
report when it is created the reports needs to be
generated again to include it.

Options This section includes general settings for the Pivot

report:
• Auto format Chart - makes the lines in the chart
thicker.

Rev 0 89
Chapter 3

8-6: Append

The Append feature enables you to either append the forecast to your
existing data or to another spreadsheet.

The table below details what each Option means within the Append
feature.

Option Description

Append to Data Append the Forecast directly to the spreadsheet used

Set as the source of Historical observations. Data is
appended to adjoining cells.

Append to Append the Forecast to another spreadsheet by

Location selecting this option and locating the other
spreadsheet on your computer.

Rev 0 90
Chapter 3

8-7: Reports Advanced Options

The Advanced link is used to access the Reports Advanced features that
provide you with more collaborative ability when generating a report.

8-7-1: Standard

To access the Standard Reports Advanced Options:

1. From the Reports tab, click on the Advanced link.

The Reports Advanced Options window displays with the Standard

tab displayed as a default.

2. Select either Output Text File or Output CSV File or both.

Option Description

Output Text File Check this option to save the Standard report to a text
file (tab delimited).

Note: This text report is created under your “My

Documents\My Forecasts” folder for the user currently
logged into the machine.

Rev 0 91
Chapter 3

Option Description

Output CSV File Check this option to save the Standard report to a CSV
(Comma delimited).

Note: This text report is created under your “My

Documents\My Forecasts” folder for the user currently
logged into the machine.

Outputs the Standard Report with a freeze panes line

already in place, separating the graph & graph
controls from the table.

Includes a dashed red line on the report indicating

where historical data ends and forecasts begin.

3. Click OK when finished.

8-7-2: Collaboration

Collaboration enables you to customize the Tracking Table in the

Executive Report. You can add new Opinion Lines to your report that will
be included in the graphs. Collaboration also adds totals in the reports
generated based on the forecast and the new customized Opinion Lines.

To access the Collaboration Advanced Options:

1. From the Reports tab, click on the Advanced link.

2. From the Executive tab, click on Tracking Table in the Options area to
enable Collaboration.

Note: The Collaboration tab is enabled only when the Tracking Table
checkbox on the Executive tab has been selected for the report.

3. Click on the Collaboration tab.

Rev 0 92
Chapter 3

The Collaboration Advanced Options window displays.

The following table details what each feature means.

Feature Description

Additional Used to define up to four additional Opinion Lines to be

Opinions included in the Executive Report. To include an extra
Opinion Line, select the checkbox then select a name
for that Opinion Line.

Options Provides additional options for the Tracking Table in

the Executive Report.

• Sum - select Sum to include an Opinion Line that

will be the Sum of all the extra Opinion Lines
added in the Additional Opinions section.
• Average - select Average to include an Opinion
Line that will calculate an average of all the extra
Opinion Lines added in the Additional Opinions
section.
• Include Stat Forecast in Sum/Average - Sum and
Average Opinion Lines are included the
Forecast Opinion Line before being calculated.

4. Click OK.

Rev 0 93
Chapter 3

8-7-3: Tracking Table

The Tracking Table enables you to make graphical and table changes to
the Forecast. You can also compress or hide the Historical data in order
to focus on the Forecast values only.

To access the Tracking Table Reports Advanced Options:

1. From the Reports tab, click on the Advanced link.

2. Click on the Tracking Table tab.

The Tracking Table Reports Advanced Options window displays.

3. Select Executive Report to activate the Tracking Table feature in the

Executive Report.

Note: Considering the Audit Trail includes the Executive Report, the Audit
Trail inherits this feature if the Executive Report has it as well.

Rev 0 94
Chapter 3

4. Select one of the following options. These settings are applied to the
Tracking table that can be generated for the reports. The option
selected defines the default view for the tracking table.

Option Description

Compress Select this option to have the Historical data

compressed in the report.

Hide Select this option to have the Historical date hidden in

the report.

Normal Select this option to have the Historical data displayed

in the report.

5. Click OK when finished.

8-7-4: Report Bursting

Report Bursting groups reports by assigned category type so that each

category is reported in a separate workbook; this is beneficial if you would
like to focus on each category separately. You can use up to three labels
to group your data.

To access the Report Bursting Reports Advanced Options:

1. From the Reports tab, click on the Advanced link.

2. Click on the Report Bursting tab.

Rev 0 95
Chapter 3

The Report Bursting Reports Advanced Options window displays.

3. Use the pane in the right to select the desired fields to group your
data by.

Note: When the 'Group forecast' option is the Group by tab is enabled you
cannot define which labels to use to group the report, the same
defined grouping as in the Group by tab is used.

Note: The number of available columns to group data is defined by the

ForecastX Edition, for more information see Section 6: Software
Editions 12.

4. Use the 'Apply to' section to enable the Report Bursting for any
report. The list with each of the reports should be selected:

Report Description

Standard Grouping report feature is used for the Standard report.

Executive Grouping report feature is used for the Executive

report.

Rev 0 96
Chapter 3

Report Description

Audit Trail Grouping report feature is used for the Audit Trail
report.

Pivot Table Grouping report feature is used for the Pivot Table
report.

5. Click OK.

8-8: Conversion Factors

ForecastX allows users to approximate total cost and revenue based on

forecasts and cost/price variables.

To use conversion factors:

1. From the Reports tab, click on the Conversion Factors link.

The menu options are as follows:

Report Description

Apply To Determines which reports will display cost & revenue.

Factors Allows mapping of conversions

Rev 0 97
Chapter 3

Section 9: Menu Bar

ForecastX provides additional features when generating reports. The

following sections detail the options available when clicking on Scenario
or Tools buttons, as well as additional help.

9-1: Scenario

ForecastX™ enables you to create forecasting scenarios to use in what-

if analysis. When you create a scenario, you are simulating what the
future might look like and can then study several potential outcomes.
You can study various outcomes by creating and saving several
scenarios using different forecasting method parameters.

You use the Scenario button to save your current settings as a new
scenario, or edit an existing scenario.

Rev 0 98
Chapter 3

The table below details what each feature means:

Scenario Description
Feature

Reset to defaults Resets all the options and parameters inside the
application to the defaults as if the application was just
installed.

Note: Invoking this feature will not change the options

in the Scenario Management settings; this includes the
default location for the scenario files.

Open Used to open a scenario file (.fx7 file extension only)

that has been saved previously. Once a scenario file is
loaded the application's title bar will change to reflect
the name of the open scenario and all the options and
parameters will default to that scenarios configuration.

Default Location Display all the scenario files that have been saved in
the current default location. The default location is a
directory and can be configured in the Setting option
from the Scenario menu. This menu is a way to easily
change scenarios that have been saved for future use.

Save Saves unsaved changes in any options or parameter in

all the application, these changes will be saved to the
currently open scenario.

Save As... Saves all the current settings from the application in a
Scenario file. Scenario files have an .fx7 extension.

Note: This feature is not available in ForecastX

Student Edition.

Rev 0 99
Chapter 3

Scenario Description
Feature

Auto Save Specifies whether the should automatically save

changes made to the currently active scenario. When
this option is enabled, all changes to any option in the
application, like ranges, parameters, reports to
generate will be automatically saved.

Note: The options in the 'data to be forecast' and 'Data

set' sections in the Data capture tab are affected by
higher priority option inside the scenario Settings. In
order for the Auto save feature to affect the parameters
in these two sections the options in the 'Apply scenario
to Data Capture' section must be enabled.

Settings Contains more options regarding Scenario

Management.

To configure the Scenario’s settings:

1. Click on the button and select Settings.

Rev 0 100
Chapter 3

The Scenario Management window displays.

2. In the Address Saving Options, determine how the Excel file is saved.

Save Option Description

Absolute format Stores all ranges using an absolute notation; this

means it will include the reference to the filename and
use absolute notation for cell ranges. For example, this
is a range stored in absolute format
[ForecastXTrainingSet.xls]ProCast!$A$1:$AF$7].

Relative format Stores all ranges using an relative notation; this means
it will only store cells information in a relative format.
For example, this is a range stored in a relative format:
A1:AF7.

Do not save No ranges are saved in the Scenario.

address

Note: If your scenario configuration is not storing your

addresses as configured for addresses used in the
Data Capture tab, then verify the
'Apply scenario to Data Capture' section options.

Rev 0 101
Chapter 3

3. In the Apply scenario to Data Capture area, select how the data is
applied to the Scenario. ForecastX recognizes your data's layout and
characteristics. The settings are used to override the IDR so that the
Auto save feature can remember your changes in the Data Capture
tab when a scenario is loaded. The following table details what each
option means.

Option Description

Change current Impacts the range of data to be forecasted and also the
data selection checkbox to select if the data is organized in rows or
columns. When this is enabled, the IDR will be
overridden and ForecastX remembers the last saved
options for the Range to be forecasted and the data
organization option. When this option is disabled, the
IDR is enabled and the layout of your data is detected
automatically.

Note: The IDR feature works when no range has been

selected manually and the user only selects a single
cell inside the data to be forecasted.

Change other Impacts the 'Contains Dates' and 'Periodicity' features

data capture inside the Data capture tab. When this option is
options disabled, the Periodicity and the 'Contains Dates'
checkbox are automatically detected using IDR every
time the application is opened. When the option is
enabled the application will remember the last setting
saved for the two sections mentioned.

4. In the Default Location area, this is the directory where the looks for
scenario files. All the Scenario files saved in this directory are listed
under the 'Default Location' option in the Scenario menu in the main
window.

5. Click OK.

Rev 0 102
Chapter 3

9-2: Tools

The Tools menu contains options to configure the way ForecastX and
the reports are generated regarding security and color customizing.

Rev 0 103
Chapter 3

The table below details what the Tools features.

Tools Feature Description

Report Options This section contains options that affect the generated
reports in different areas:

• Data - contains configurations regarding:

o Negative Values Allowed - select this

option to include negative values, rather
than dismissing them as 0, which is the
default.

Note: When grouping forecast is enabled, this

option is checked and disabled.

o Integers only - view all numbers as

integers rather than real numbers with a
fixed number of decimal places.

• Fiscal Start Date - in the Executive report, the

Forecast Table uses "January" or first week
as the starting point of a year, if this is the
behavior you expect leave this field empty. To
change the date on which your fiscal year
starts just configure it in this field.

Security - contains security related options. When a

password is defined that password will be required to
unlock the sheet the report is in. If no password is set
the sheet can be unlocked without any restriction.

Note: This feature is not available in ForecastX

Student Edition.

Rev 0 104
Chapter 3

Tools Feature Description

Report Options • Annotation - contains general settings.

(continued)
o Author- name used in the Executive and
Audit reports in the Reports Detail
Section.

o Note - note used in the Executive and

Audit reports in the Reports Detail
Section.

• Colors - This tab contains options to

configure the color the Adjust opinion line will
display in different situations in the Standard
and Standard DRP reports:

Note: This feature is not available in ForecastX

Student Edition.

• Empty Cell - is the default color for cells that

can be edited by the user.

• Value same as Stat Forecast - defines the

color of a cell when it has the same value as
the original forecast. This will be observed
when reconciling or running, On-Fly DRP or
On-Fly Rebalance, for example.

• Value are in range - defines the color of a cell

when a value is entered into the Adjust
opinion line and it is within the Upper and
Lower Limit.

• Value out of range - defines the color of a cell

when a value is entered into the Adjust
opinion line and it is greater than the Upper or
less than the Lower Limit.

Performance Used to configure how many series ForecastX can

process simultaneously. If your computer has less
RAM memory, this setting should be lower.

Rev 0 105
Chapter 3

Tools Feature Description

Customize Used to configure changes in the User Interface.

Currently, you can select which skin to apply to the
whole application.

9-3: Help

The Help menu consists of additional features for efficiently using the
ForecastX.

The following table details the features available within Help.

Feature Description

Contents Opens the ForecastX User Guide.

Register product Enables you to register for the Forecast Xpert User
Forum.

Technical Contact John Galt Solutions Technical Support team.

support

Upgrade License This option allows you to upgrade an existing

installation of ForecastX to a higher edition or replace
an existing temporary license. For more information
on upgrading to another edition read the following
section link.

Rev 0 106
Chapter 3

Feature Description

About Information about the ForecastX edition and version

number.

Section 10: Customer Feedback

In the event that you experience a problem with the application, an error
message window with the errors description will be displayed, if you wish
to have the John Galt support staff get all the necessary data to help you
out in solving your problem you can automatically send an e-mail to
speed up the process.

When the error message window is displayed, click on the Yes option to
send the information through an e-mail, the following window will be
displayed:

Rev 0 107
Chapter 3

In order for the company to solve the problem fill out all the fields with
accurate and as much detail as possible. If you are able to, please check
the “Include data set used“ in order for us to get a copy of the dataset
used when the error occurred, this information will not be shared in any
way.

Rev 0 108
Chapter 3

Section 11: Upgrade License

This Section will describe how to upgrade the license for a ForecastX
installation. This will allow you to do a few tasks, you could upgrade to a
higher Edition of the application that will enable more features, or you
could reinstall a corrupted license, or you could upgrade your license
from a temporary one to a permanent one.

In Order to do any of the previously mentioned tasks you must follow

these steps:

1. Open up Microsoft Excel

2. Click on the button.

The ForecastX window will be displayed.

3. Next, click on the menu and select “Upgrade License”.

Rev 0 109
Chapter 3

The License Activation window is displayed.

4. Next Select the exact Product edition and version you wish to
upgrade to.

5. Contact John Galt Solutions at (312) 701-9026 to get your Activation

Code; you will be asked the product edition and version you are trying
to upgrade to in order to get you Activation Code. Once you have
been given the Activation Code type it into the box below the
“Activation Code” label and press the “Activate License” button.

If the license was activated correctly you should see an information

message similar to this.

6. Press OK. After closing excel and opening it again you should
already be running the new upgraded version of ForecastX. You can
check this by going into the About option in the Help menu of
ForecastX.

Rev 0 110
Chapter 4 Forecast Methods

This chapter includes:

Section 1: Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Section 2: Adaptive Exponential Smoothing. . . . . . . . . . . . . . 116
Section 3: Bass Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Section 4: Box-Jenkins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Section 5: Census X-11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Section 6: Croston Intermittent . . . . . . . . . . . . . . . . . . . . . . . . 131
Section 7: Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Section 8: Double Simple Exponential Smoothing Brown . . . 136
Section 9: Double Exponential Smoothing Holt . . . . . . . . . . . 137
Section 10: Erlang B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Section 11: Erlang C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Section 12: Event Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Section 13: Gompertz Curve . . . . . . . . . . . . . . . . . . . . . . . . . 154
Section 14: Holt-Winters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Section 15: Logistic Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Section 16: Moving Average . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Section 17: Multiple Regression . . . . . . . . . . . . . . . . . . . . . . . 162
Section 18: New Product Forecasting . . . . . . . . . . . . . . . . . . 166
Section 19: Poisson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Section 20: Polynomial Regression . . . . . . . . . . . . . . . . . . . . 170
Section 21: Probit Curve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Section 22: Simple Exponential Smoothing . . . . . . . . . . . . . . 174
Section 23: Slow Moving Demand . . . . . . . . . . . . . . . . . . . . . 177
Section 24: Stepwise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Section 25: Trend (Linear) Regression . . . . . . . . . . . . . . . . . . 195
Section 26: Triple Brown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Section 27: Weighted Moving Average. . . . . . . . . . . . . . . . . . 2

Rev 0 111
Chapter 4

Section 1: Overview

This chapter details the forecasting techniques available from the

Forecast Method tab in ForecastX. After opening the Forecast
Method window, the Procast forecast technique is shown as the
default.

Section 2: Adaptive Exponential Smoothing

The Adaptive Exponential Smoothing method is a derivative of Simple

Exponential Smoothing. In regard to Adaptive Exponential Smoothing,
the Level value is systematically changed from period to period to
allow for pattern changes in the Historical data. Adaptive Exponential
Smoothing is automated, which makes it a useful method to employ
when large numbers of items are involved. The data is non- seasonal
and shows no trend.

Adaptive Exponential Smoothing is beneficial to use for a product that

has been sold for a long period of time, only to encounter a dramatic
market change in recent periods.

To use the Adaptive Exponential Smoothing forecasting method:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Adaptive Exponential Smoothing method is used. For your
company’s purposes, you will have your own data available.

2. Click the Adaptive Exp. Smoothing tab at the bottom of the Excel
sheet to select the example spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .
ForecastX displays.

4. Click on the Forecast Method tab.

Rev 0 112
Chapter 4

5. In the Forecast Technique area, scroll through the list of methods

and select Adaptive Exponential Smoothing.

The Adaptive Exponential Smoothing Forecasting technique

displays.

6. Select Edit parameters to activate Adaptive Exponential

Smoothing’s parameters.

7. In the Smoothing Constants area, type in a number for Seasonal.

8. In the Reports tab, select the Standard report option to output the
results in an Excel Spreadsheet.

9. In the Chart area of the Reports tab, select Standard and Chart in
worksheet.

10. Click Finish.

There are two areas in the Standard report to view the results.
First, the Excel spreadsheet produced by ForecastX™ shows
three things:

Rev 0 113
Chapter 4

• Actual values of your original Historical data

• Forecast method
chosen

• Forecasted values for both the original Historical data and

the future values. (The forecasted values are in bold.)

Second, near the bottom of the spreadsheet is a small table that

contains basic error measurements: Mean Absolute Percentage
Error, R-Square Error, Mean, and Standard Deviation.

Click the Chart1 tab in the lower-left corner of the spreadsheet.

The table shows you that your forecast accuracy is very good, and
that the fit of the forecast, represented by the R-Square value, is very
good.

The graph enables you to view how the forecast is represented in the
future and also how your data has been behaving in the past. As the
graph indicates, the market changes from April 2005 to June 2007did
affect the sales of your product. Because you chose to use the
Adaptive Exponential Smoothing method, ForecastX™ weighted the
most recent values more heavily than older values. This enables you

Rev 0 114
Chapter 4

to consider how the market is currently behaving, and how that

behavior is going to affect your total sales.

Section 3: Bass Model

This model is another new product model, however Bass Model is not
a candidate method in the new product forecasting method.

To use the Bass Model forecasting method:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Bass Model.

Rev 0 115
Chapter 4

The Bass Model Forecasting technique displays.

3. Select Edit parameters to activate Bass Model’s parameters.

4. In the Parameters area, type in the parameters you would like to

use. The table below details what each option means.

Parameter Description

p Set the Innovation Rate of the Bass Model.

r Set the Imitation Rate of the Bass Model.

qbar Set the long run cumulative value of the forecasted

values. The recommended value of qbar is the
cumulative value of all Historical values.
If left blank, ForecastX™ chooses a value for
you.

5. Click Finish.

Rev 0 116
Chapter 4

Section 4: Box-Jenkins

The basis for the Box-Jenkins methodology consists of three phases:

• Identification

• Estimation

• Testing and applying the ARIMA model

This methodology is a multi-step model building strategy aimed at

optimizing the ARIMA process. ForecastX™ automatically optimizes
the best ARIMA model using Box-Jenkins. ForecastX™ enables you
to perform data transformation and analyze the ACF and PACF charts
for model selection. Box Jenkins is best used on extensively long
Historical data sets with lower volatility.

The table below details the four phases of the Box-Jenkins.

Phase Description

Model In this phase, data is plotted (if needed) for

Identification performing transformation and differencing
analysis. A data transformation stabilizes the
variance within a set of data points. If necessary,
difference the data to make it stationary. Also in
this phase, a first guess at the ARIMA model will be
analyzed.

Estimation and In this phase, the model parameters are chosen

Forecasting based on the model that provides the lowest error
term. The data is then forecasted.

Diagnostic The decision is made based on an analysis of the

Checking statistical accuracy of the model whether to re-run
using a different model or continue using the
ARIMA model diagnosed.

Forecasting A forecast of the time series is generated using the

ARIMA model.

The two basic classes of components in Box-Jenkins are Non-

seasonal (trend) and Seasonality.

Rev 0 117
Chapter 4

When looking for trends within data, look for gradual increases in
sales over time that do not repeat within the time range being
forecasted.This (in combination with higher sales during certain times
of the year (seasonality)), signifies an increase in sales from year to
year that follows a seasonal pattern.

To use the Box Jenkins forecasting method:

1. Click on and open the TutorialAirlineData.xls file.

Note: The TutorialAirlineData.xls file is a data example to demonstrate how

the Box Jenkins method is used. For your company’s purposes, you
will have your own data available.

2. Click in a cell containing data, and open ForecastX™

by clicking on .

ForecastX displays with the Data Capture tab open.

Rev 0 118
Chapter 4

3. Since your data is set up in monthly format, ensure that Contains

dates is checked and monthly is selected.

4. In the Forecast periods textbox, type in 36 to Forecast three

years into the future.

5. In the Seasonality textbox, type in 12.

6. Click the Forecast Method tab.

7. In the Forecast Technique area, scroll through the list of methods

and select Box Jenkins.

The Box Jenkins Forecasting technique displays.

Note: For this example, Seasonal parameters will not be set. ForecastX
will choose the parameters.

8. On the Reports tab, select the Audit Trail report. The Audit Trail
report includes Accuracy and Descriptive statistics.

9. Click Finish.

Rev 0 119
Chapter 4

4-1: Performing ARIMA Analysis

ForecastX™ includes several tools to analyze your data and assist

on ARIMA selection. The following table details the tools used.

Tools Description

Customize The ability to input a non-seasonal and seasonal

ARIMA Model ARIMA model is available in ForecastX™. From
Input the Forecast Method screen, you can select Box-
Jenkins and enter your own ARIMA model.

Data This tool charts the variance and provides the

Transformation functionality to transform your data. By examining
the variance chart, the sophisticated forecaster
can perform a log, square, and square root
transformation.

On the Forecast Method tab, click the Transform

button and take a look at your variance chart. The
goal is to have your data distributed evenly around
the mean.

Data Analysis This chart includes the ACF and PACF chart and
provides functionality to differentiate your data.

The ACF and PACF (auto-correlation function and

partial auto-correlation function) are used in
determining whether your data is stationary. It also
allows you to see seasonality and trend. As
discussed previously, your data may contain both
factors, and therefore must be differenced before
we can choose our ARIMA model. ForecastX™
allows you to put in the differencing parameters
and view the changes in the correlogram as you
make them.

Rev 0 120
Chapter 4

Tools Description

Differencing On the Forecast Method tab, click the Analyze link.

It is here that you are able to set the differencing
parameters. Differencing allows you to remove the
trend by subtracting adjacent data points. The
correlogram displays the ACF and PACF, which
show you the auto-correlations and relationships
that each variable has with the past. If correlation
is high, it means that the variable is predictable.

To perform the ARIMA analysis:

1. On the Forecast Method tab, scroll through the list of Forecasting

techniques and select Box Jenkins,

2. Check the Edit parameters checkbox and type in 1 for the Non-
seasonal difference and 2 for the Seasonal difference.

As you can see, only one or two correlations fall outside of the
upper and lower limits.

3. On the Statistics tab, select a statistic to use. ForecastX

displays the default statistic if none are selected.

4. On the Reports tab, choose the Audit Trail report.

5. On the Audit tab, select the Out of Sample option to display a

fitted values table.

6. Click Finish.

Section 5: Census X-11

Census X-11 is a refined seasonal-decomposition method. It

seasonally adjusts and decomposes your data through a series of
predefined steps:

• Seasonal (Trend)

• Cycle (Random)

Rev 0 121
Chapter 4

A seasonal component of a time series occurs regularly, such as

Christmas; while a cycle has a duration that varies from cycle to
cycle. Economic variables are cyclical. A depression occurs at
irregular intervals.

During the forecasting process, Census X-11 factors in trading days

to refine the forecast and reduce the irregular component of the
forecast by weighing each period by the number of trading days.
ForecastX™ enables you to account for trading days or forecast
without adjusting for trading days. The former produces more
accurate forecasts when trading days are important; this is especially
useful if the number of trading days dramatically affects the amount of
business that is done in a time period.

Trading days also can be specifically known for each time period. For
example, you can determine how many trading days each of the last
12 months contained by reviewing how many official working days
took place. You can also determine how many trading days are going
to occur over the next 36 months, for example.

Census X-11 also identifies two types of Seasonality:

• Additive

• Multiplicative

Census X-11 is best suited for a series where the number of days in
the time period directly affects the values in those time periods.

To use the Census X-11 forecasting method:

1. Click on and open the Tutorial_CX11_Columns.xls

file.

Note: The Tutorial_CX11_Columns.xls file is a data example to

demonstrate how the Census X-11 method is used. For your
company’s purposes, you will have your own data available.

2. Click in a cell containing data, and open ForecastX™

by clicking on .

Rev 0 122
Chapter 4

ForecastX displays with the Data Capture tab open.

Note: Notice ForecastX™ automatically selects your data and indicates it

is in columns.

3. Select the Contains dates checkbox and ensure that Monthly is

selected.

4. In the Forecast periods textbox, type in 36 to Forecast three

years into the future.

5. In the Seasonality textbox, type in 12.

6. Select the Forecast Method tab, and scroll through the list of
forecasting techniques and select Census X-11.

Rev 0 123
Chapter 4

The Census X-11 Forecasting technique displays.

7. Select Edit parameters to activate the parameters.

8. On the Report tab, select Executive Report and uncheck any

other available options.

9. Click Finish.

Rev 0 124
Chapter 4

Notice ForecastX confirms that sales are affected in the months

preceding and following a holiday. According to the forecast, sales
through your distribution outlet continue to increase over the next
three years.

Considering you did not take the actual number of business or trading
days into account, the average error of your forecast is 3.81%.

If you decide to run a forecast that does account for the actual trading
days in each month, your results will be a more accurate forecast.

5-1: Using Advanced Census X-11 with ForecastX

To use the Advanced Census X-11 forecasting technique:

1. Click on and open the Tutorial_CX11_Columns.xls

file.

Rev 0 125
Chapter 4

Note: The Tutorial_CX11_Columns.xls file is a data example to

demonstrate how the Census X-11 method is used. For your
company’s purposes, you will have your own data available.

2. Click in a cell containing data, and open ForecastX™

by clicking on .

ForecastX displays with the Data Capture tab open.

Note: Notice ForecastX™ automatically selects your data and indicates it

is in columns.

3. Select the Contains dates checkbox and ensure that Monthly is

selected.

4. In the Forecast periods textbox, type in 36 to Forecast three

years into the future.

5. In the Seasonality textbox, type in 12.

6. Select the Forecast Method tab, and scroll through the list of
forecasting techniques and select Census X-11.

The Census X-11 forecasting technique displays.

7. In the Seasonal Type area, select Multiplicative. This type

spreads the total values from each month over the actual number
of trading days of each month.

8. Select Need Trading Day Adjust.

9. Click the in the Trading Day Series textbox.

Rev 0 126
Chapter 4

The window displays.

10. To choose the range of trading days, select which cells to include
in the trading days in column D of the Excel sheet.

11. Click OK when finished.

12. On the Reports tab, select Audit Trail Reports option and deselect
any other options.

13. Click Finish.

The Audit Trail report is much more detailed than the Executive
report. For example, there is detailed information about the method
you selected, Census X-11, along with the decomposition type you
specified, multiplicative. You chose multiplicative because you wanted
to spread the values from each month over the actual number of
trading days for each month.

Section 6: Croston Intermittent

Croston Intermittent is the method used to handle sporadic data. For

example, a manufacturer has inventory that is constantly in sporadic
demand; that is, there are no precursors to when orders will be
placed, or it could be that the demand is often zero, even though the
average demand may be for several units. This situation causes an
over calculation of the average or mean demand. Orders do not
arrive at predictable times, so there are times when the manufacturer
is over or under stocked.

Rev 0 127
Chapter 4

Croston Intermittent is specifically designed to deal with such a

situation with a two-step process. The main goal is to provide a good
safety stock, resulting in a situation where the manufacturer is neither
over nor under stocked. This is accomplished by considering two
aspects of the data: the demand size and the demand occurrence.

To use the Croston Intermittent forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastX Examples.xls file is a data example to demonstrate

how the Croston’s Intermittent method is used. For your company’s
purposes, you will have your own data available.

2. Open the Croston spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

4. Select the Forecast Method tab, and scroll through the list of
forecasting techniques and select Croston Intermittent.

Rev 0 128
Chapter 4

The Croston Intermittent Forecasting technique displays.

5. On the Reports tab, select Standard Report.

6. Select the Standard and Chart in Table checkboxes.

7. Click Finish.

ForecastX confirms that sales are affected in the months

preceding and following a holiday. According to the
forecast, sales through your distribution outlet continue to
increase over the next three years.

Rev 0 129
Chapter 4

Below is the graph, Chart1, produced by the Standard report.

Notice how sporadic the data is.

With Croston Intermittent forecasting method, you can control and

smooth sporadic data. This method provides excellent results for
product control and forecasted demand. In the standard report, you
can see the MAPE is 75.05% and the R-Square is 0.

Section 7: Decomposition

The Decomposition model is used to identify underlying components

by breaking the series into its component parts and then
reassembling the parts to construct a forecast.

Rev 0 130
Chapter 4

Smoothing methods do not focus on identifying individual

components of the basic pattern. But many business and economic
data contain underlying components that, when examined
individually, can help the forecaster better understand data
movements and therefore, make better forecasts. Usually, these
components include the long-term trend, seasonal pattern, cyclical
movements, and irregular fluctuations. ForecastX includes the
Decomposition model, and presents the detailed information of each
underlying component in the Audit Trail report.

To use the Decomposition forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Decomposition.

The Decomposition Forecasting technique displays.

3. Select Edit parameters to activate Decomposition’s parameters.

Rev 0 131
Chapter 4

The following table details what each parameter means.

Parameter Description

Type Indicate the type of Seasonality of the

Decomposition method.
„ Multiplicative - Seasonality is Multiplicative.
„ Additive - Seasonality is Additive.

Forecast Method Select the method used to forecast the long-term

for Decomposed trend and cyclical movement in the decomposed
Data data.

4. Click Finish.

Section 8: Double Simple Exponential Smoothing Brown

The Brown's Double Simple Exponential Smoothing method attempts

to create a linear equation. It performs two simple exponential
smoothing forecasts and then adjusts for the linear trend in the data.
It is similar to Double Exponential Smoothing in the fact that the goal
is to create a linear trend, but it does so without adding additional
parameters to the equation. Since forecasts can be expressed as a
function of the single and double smoothed constants, the procedure
is known as Double Exponential Smoothing.

Brown's Double Exponential Smoothing is appropriate for data that

shows a linear trend over time. The smoothing component is level.

To use the Double Brown Exponential Smoothing forecasting

technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Double Brown.

Rev 0 132
Chapter 4

The Double Exponential Smoothing Forecasting technique

displays.

3. Select Edit parameters to activate the parameters.

4. In the Smoothing Constants area, type in the number of levels to

use in the Forecast.

5. Click Finish.

The results display a linear trend over time.

Section 9: Double Exponential Smoothing Holt

Holt's Double Exponential Smoothing method is similar to Simple
Exponential Smoothing. It calculates the level component to measure
the level in the Forecast. It also adds the trend component to create a
Linear trend in the Forecast. This equation is similar to a Linear
Regression line and is useful when a product is experiencing an

Rev 0 133
Chapter 4

exponential growth or decline, while not experiencing Seasonality. If

your data is dynamic and is not affected by seasonal factors, then
Double Holt is an optimal method to use.

To use the Double Holt forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastX Examples.xls file is a data example to demonstrate

how the Croston Intermittent method is used. For your company’s
purposes, you will have your own data available.

2. Open the Double Holt’s spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

4. Click on the Forecast Method tab.

5. In the Forecast Technique area, scroll through the list of methods

and select Double Holt.

Rev 0 134
Chapter 4

The Double Exponential Smoothing Holt Forecasting technique

displays.

6. On the Data Capture tab, click on the Data Cleansing link.

Rev 0 135
Chapter 4

The Data Cleansing window displays.

7. In the Replace Outliers area, select Yes. ForecastX will then

automatically remove the outliers and produce a more accurate
Forecast in result.

8. In the Reports tab, select the Audit Trail report.

9. Click Finish.

After ForecastX™ produces the results in an Audit Trail report, the

statistics are displayed to measure the values of your Double Holt
forecast. The graph that is made of the actual, fitted, and forecast
values enables you to see the trend in your new product.

Rev 0 136
Chapter 4

Notice June 1998 on the graph. ForecastX™ removed a large data

point, the outlier. The only thing you needed to do was click Replace
Outliers on the Advance Option screen.

There are two important statistics to look at in the Audit Trail report for
the Double Holt forecast:

• Constant Smoothing values

• Level and Trend measurements

These values tell you how significantly ForecastX smoothed the

data. The gamma value is high (near 1.00), so the smoothing was
insignificant. However, the Level value is low and therefore a
significant Smoothing Constant. This means that your data has
significant trend.

Section 10: Erlang B

ErlangB is used to work out how many lines are required if the traffic
figure (in Erlangs) during the busiest hour known. The Erlang B
distribution is based on the following assumptions:

• Holding times are constant and exponential

Rev 0 137
Chapter 4

• Blocked cells are cleared

• Calls are taken in random order

• There are an infinite number of sources

To use the Erlang B forecasting method:

1. Click on and open the ForecastX_Erlang.xls file.

Note: The ForecastX_Erlang.xls file is a data example to demonstrate how

the Erlang B method is used. For your company’s purposes, you will
have your own data available.

2. Click on the Erlang B1 sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

Rev 0 138
Chapter 4

4. ForecastX displays with the Data Capture tab open.

5. In the Forecast periods area, type in 4 to forecast for the next four
quarters.

6. Click on the Forecast Method tab.

7. In the Forecast Technique area, scroll through the list of methods

and select Erlang B.

Rev 0 139
Chapter 4

The Erlang B Forecasting technique displays.

8. Select Edit parameters to activate Erlang B’s parameters.

9. In the Compute Type area, select either:

• N to A

• A to N

N means the number of servers, and A means traffic measured

by Erlang B. Select N to A.

10. In the Lost Possibility area, select one of the options for the
percentage of loss that is possible.

11. Click Finish.

As you review the results within the standard report, notice how
ForecastX has forecasted the number of servers

Rev 0 140
Chapter 4

needed for the next 4 quarters, and ForecastX™ has also given
the traffic levels (A).

Section 11: Erlang C

Erlang C assumes that all blocked calls stay in the system until they
can be handled. This model can be applied to the design of call
center staffing arrangements where, if calls cannot be immediately
answered, they enter a queue. The Erlang C distribution is based on
the following assumptions:

• Calls are served in order of arrival

• There are infinite number of sources

• Blocked calls are delayed

• Holding times are exponential

To use the Erlang C forecasting method:

1. Click on and open the ForecastX_Erlang.xls file.

Note: The ForecastX_Erlang.xls file is a data example to demonstrate how

the Erlang C method is used. For your company’s purposes, you will
have your own data available.

Rev 0 141
Chapter 4

2. Click on the Erlang C1 sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

ForecastX displays with the Data Capture tab open.

4. In the Forecast periods area, type in 4 to forecast for the next four
quarters.

5. Click on the Forecast Method tab.

6. In the Forecast Technique area, scroll through the list of methods

and select Erlang C.

Rev 0 142
Chapter 4

The Erlang C Forecasting technique displays.

7. Select Edit parameters to activate Erlang C’s parameters.

8. In the Compute Type area, select one of the following options:

• A, N to PO

• A, PO to N

• N, PO to A

N is the number of servers, P0 is the probability of delay greater

than zero, and A is traffic, measured by Erlang.

9. In the Aid Series for Erlang C area, select one of the options.

10. Click on the textbox.

Rev 0 143
Chapter 4

The dialog displays.

11. Highlight the Probability of Delay column in the Excel

spreadsheet and Click OK.

12. Click Finish.

As you review the results within the Standard Report, notice how
ForecastX™ created the forecast for the number of servers for
the next 4 periods and also calculated the traffic based on the
probability of delay).

Rev 0 144
Chapter 4

Section 12: Event Model

Event Model was designed to enhance your data analysis and allow
you to take promotions and other events into account as you develop
your forecasts. Now you can go beyond analyzing the trend and
seasonality contained in your historical data by assessing the impact
of promotions on your historical and future demand.

Companies conduct promotions of various types. At any given

moment your organization may use push and pull types of promotions
at various times throughout the year. In general, push promotions are
designed to push the merchandise off of the shelves through
increased customer awareness, whereas pull promotions are
designed to attract customer's attention by means of an incentive that
causes them to purchase your product.

To use the Event Model forecasting technique:

1. Click on and open the Tutorial_Event_Modeling.xls

file.

Note: The Tutorial_Event_Modeling.xls file is a data example to

demonstrate how the Event Model method is used. For your
company’s purposes, you will have your own data available.

2. Click on the Base Case sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

Rev 0 145
Chapter 4

The Event Model Forecasting technique displays.

4. Select Edit parameters to activate the Event Model’s parameters.

5. Click on the Event Flags textbox to select a range.

The dialog displays.

6. Click OK.

Rev 0 146
Chapter 4

7. In the Smoothing Constants area, type in a Level.

8. Select Allow Reapply to enable the Reapply Event Modeling

feature. This feature allows you to click on the Reapply Event
button on the ForecastX toolbar. The Standard report of the
Event Modeling Forecast can be reapplied.

9. Click on the Advanced link.

The Advanced Event Model window displays.

The table below details what each feature means in the Method
and Method and Simulation tab.

Feature Description

Basic Forecast Select the forecast method used to Forecast the

Method bottom line of the Event Model.

No Simulation Do not use the simulation.

Random Event Use the simulation, and the percentage of each

Percentage event in the simulation is random.

Rev 0 147
Chapter 4

Feature Description

Fixed Event Use the simulation, and the percentage of each

Percentage event in the simulation is fixed, which equals the
percentage of events in the given event flags.

Number of Enter the number of simulations. Each simulation

Simulations creates one set of events to create the simulated
forecasted values.

Need Estimation Select this option to activate the profit estimation

function, which calculates the profit by the
subtraction of the total promotional expense from
the total revenue.

Total Used to estimate profit.

Promotional
Expense

Average Sell Used to estimate the revenue, which is calculated

Price by the multiplication of the average sell price and
the total sales.

Rev 0 148
Chapter 4

10. Click on the Event Flag Assignment tab.

11. Select either BySequence or ByLabels.

Event Flag Description

Assignment

BySequence The first Event Series is assign to the first series,

and the second event flags is assign to the second
series. If there are not enough Event Series, then
the first event is reused, the second event is
reused thereafter, etc.

Rev 0 149
Chapter 4

Description
Event Flag
Assignment

ByLabels Specify one or more label fields to link the Event

Series and data series. You can select the
hierarchy of the labels for the event flag by
dragging and dropping the ByLabels column
headers.

Note: The number of labels available for this

feature is dependant on the ForecastX
Edition you are running, to verify your
tools capacity see Section 6: Software
Editions 12.

12. Click OK.

13. Click Finish.

On the Audit Report, you see the effect of the events on

Newspaper Sales and how your error measurements improve.
The Audit Trail report provides you with additional statistical data,
including the event lift factors.

Section 13: Gompertz Curve

The Gompertz Curve is one of the three new product models used in
the New Product forecasting method, but it is also a stand-alone
forecasting technique. For additional information, please see Section
18: New Product Forecasting.

To use Gompertz Curve forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Gompertz Curve.

Rev 0 150
Chapter 4

The Gompertz Curve Forecasting technique displays.

3. Select Edit parameters to activate Gompertz Curve’s parameters.

4. In the Value Limit area, select the minimum and maximum

values.

5. Select either Optimized or Fixed.

Option Description

Optimized Optimize the maximum value specified. The

maximum value might be changed to a value lower
than specified but this is done to optimize the
forecast generated.

Fixed Maximum value set is used as the maximum for

the created forecast.

6. Click Finish.

Rev 0 151
Chapter 4

Section 14: Holt-Winters

Holt-Winters is an Exponential Smoothing technique that incorporates

growth and Seasonality into the forecast. Holt-Winters does this by
producing Seasonal lift factors for each seasonal period. The
Seasonal indices are displayed in the Audit Trail report.

If the Historical data is known to change rapidly, large smoothing

constants should be used. For stable, naturally consistent data, the
smoothing constants should be closer to zero. In either case,
ForecastX™ automatically optimizes the parameters for you if you
choose not to specify them.

To use the Holt Winters forecasting method:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Holt Winters method is used. For your company’s purposes,
you will have your own data available.

2. Click on the Holt Winters sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

The Data capture window is displayed.

4. Click on the Forecast Method tab, the following window is

displayed.

Rev 0 152
Chapter 4

5. Select Edit parameters to activate Holt Winter’s parameters.

6. In the Parameters area, select the Seasonal Type and Smoothing

Constants. The following table detail what each parameter
means.

Parameter Description

Seasonal Type Indicate which type of Holt Winters to use.

• Optimize - the optimal Seasonality is
determined for you.
• Multiplicative - Multiplicative Holt Winters.
• Additive - Additive Holt Winters

Smoothing • Level - level of smoothing.

Constants • Seasonal - Seasonal parameter.
• Trend - trend parameter.

Rev 0 153
Chapter 4

7. On the Reports tab, click on Audit Trail report to validate the

generated report’s accuracy.

8. On the Executive tab, check the Upper Limit and Lower Limit
options; this displays the confidence limits of your Forecast.

9. Click Finish.

After running the Forecast, ForecastX creates a graph and a

table with the Seasonal indices and Smoothing Constants. The
graph enables you to quickly assess the Trend and Seasonality
of your data by depicting the peaks and valleys of the sales.

The Audit Trail statistics display an accurate Forecast. In addition, the

Audit Trail report displays Smoothing Constants, Level and Trend,
as well as the Seasonal indices. You can use this detail to learn
about the Trends and Seasonality of your data.

Rev 0 154
Chapter 4

Section 15: Logistic Curve

The Logistic Curve is one of the three new product models used in
the New Product forecasting method, but it is also a stand-alone
forecasting technique. For additional information, please see Section
18: New Product Forecasting.

To use Logistic Curve forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Logistic Curve.

Rev 0 155
Chapter 4

The Logistic Curve Forecasting technique displays.

3. Select Edit parameters to activate Logistic Curve’s parameters.

4. In the Value Limit area, select the minimum and maximum

values.

5. Select either Optimized or Fixed.

Option Description

Optimized Optimize the maximum value specified. The

maximum value might be changed to a value lower
than specified but this is done to optimize the
forecast generates.

Fixed Maximum value set is used as the maximum for

the created forecast.

6. Click Finish.

Rev 0 156
Chapter 4

Section 16: Moving Average

Moving Average is the average of a time series over a specific

number of preceding periods. When a new value is added, the last is
removed from the calculation. Therefore, the specific number of
preceding periods remains constant.

However, two drawbacks are that Moving Averages do not effectively

handle significant trends in data, and all Historical data must be
stored to create the Moving Average. In the case of time series data,
Moving Average forecasts are often used to eliminate unwanted
fluctuations, thereby smoothing the time series. The appropriate
number of preceding periods is determined by selecting the amount
of periods that yields the least amount of error.

The starting value of a Moving Average is where the calculation

begins. The start value uses 0 as its first value. If you would like to
start at the 4th observation, the start value would be 3. The number of
observations to include in the average calculation is called the
History. For example, if you want to start at the 5th observation and
perform a moving average with 3 Historical Observations, the start
value would be 4 because the count starts at 0. The Moving Average
calculation would include the 3rd, 4th, and 5th observations.

To use the Moving Average forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Moving Average.

Rev 0 157
Chapter 4

The Moving Average Forecasting technique displays.

3. Select Edit parameters to activate Moving Average’s parameters.

4. In the Parameters area, type in the number of Average Periods.

5. Click Finish.

Section 17: Multiple Regression

Multiple Regression is similar to Trend (Linear) Regression except

with more Xs, or Independent Variables. Often, the Y or Dependent
Variable can be explained by more than one Independent Variable.
For example, you may have sales data that is dependent on many
factors: time of year, advertising dollars spent, and/or special
promotions implemented. Each of these factors potentially has an
influence over the sales figures. Multiple Regression is one
Forecasting technique that determines the Independent Variable with
the greatest impact. As with Trend (Linear) Regression, Multiple

Rev 0 158
Chapter 4

Regression must be interpreted and evaluated by the individual

performing the Forecast. There are accuracy and descriptive
statistics that help explain what is implied and predicted by the
Multiple Regression Forecast.

Like Trend (Linear) Regression, the slope coefficients, such as m,

needs to be interpreted. The coefficients represent the change in Y, or
the Dependent Variable, from a change in the X, or Independent
Variable, of interest. Multiple Regression determines how more than
one Independent Variable influences a single Dependent Variable.

To use the Multiple Regression Forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Multiple Regression method is used. For your company’s
purposes, you will have your own data available.

2. Click on the Multiple Regression sheet.

The spreadsheet contains 56quarters covering 1993Q1 through

2006Q4. The raw sales numbers act as the dependent variable.
Additionally, the following independent variables are included:

Variable Description

TIME Index of Time

TIME(2) Index of Time Squared (accounts for nonlinearity)

Q2 - Q4 Seasonal Dummy variable to account for

seasonality in Q2, Q3 and Q4.

SP500 Standard & Poor’s 500 Stock Index

3. Click in a cell containing data, and open ForecastX™

by clicking on .

The Data capture window is displays. Notice that ForecastX selects

the relevant data from your spreadsheet and determines it is in
columsn and arranged in quarters.

Rev 0 159
Chapter 4

4. Click on the Forecast Method tab, and from the forecasting

Technique drop down list, select Multiple Regression.

The Multiple Regression Forecasting technique displays.

5. Select Edit parameters to activate Multiple Regression’s

parameters.

ForecastX automatically selects which series in the data is the

dependent series. If you would like to use a different series,
select another series from the dropdown list.

6. On the Reports tab, select Audit Trail report. Ensure that all other
reports are not selected.

7. Select Fitted Values Table from the Audit Trail report tab to
compare the actual values with the Forecasted values.

8. Click Finish.

Rev 0 160
Chapter 4

ForecastX™ has provided you with a detailed report of your

Multiple Regression forecast that includes a graph, a fitted values
table, an ANOVA chart, and a Coefficient Table.

The summary contents of the forecast and the girred values table are
included in the ForecastX Audit Trail Report. The MAPE is 16.45%,
while the R-Squared value is 96.59

These values indicate you can have a high degree of confidence in

the forecast. As you scroll down through the Audit Trail report, you
can view the fitted values compared to the actual values at each data
point.

The Audit Trail report in ForecastX is a powerful feature that enabled

you to view all the relevant statistics when deciding whether to trust
your forecast. In this case, for your Multiple Regression forecast, you

Rev 0 161
Chapter 4

can see that ForecastX included each independent series in the

model. Additionally, ForecastX calculated the F-Test and T-test for
each independent series. This enables you to decide objectively
which series has the greatest influence on you dependent series,
Sales. From the T-Test, you can see that Q4 is the independent
series with the greatest influence on the Sales balance.

Section 18: New Product Forecasting

New Product Forecasting consists of three forecasting techniques.

New Product Forecasting automatically selects which forecasting
technique is optimal.

• Gompertz Curve

• Logistic Curve

• Probit Curve

There are two significant differences within the three forecasting

techniques:

• Amount of time that lapses before a product's growth curve

stabilizes.

• All three forecasting techniques may use different Lower and

Upper Limits for the same Forecast data.

To use the New Product Forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the New Product Forecasting method is used. For your
company’s purposes, you will have your own data available.

2. Click on the New Product sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

Rev 0 162
Chapter 4

The New Product Forecasting technique displays.

4. Select Edit parameters to activate New Product Forecasting’s

parameters.

5. In the Value Limit are, type in the Min and Max values.

6. Select either Optimize or Fixed.

Option Description

Optimize The Max value selected could be changed in order

to maximize results depending on your data.

Fixed The Max value entered is going to be used as the

upper limit for the forecast.

7. On the Reports tab, select Standard and Charts in Table.

8. Click Finish.

Rev 0 163
Chapter 4

ForecastX has produced a Gompertz graph, which displays the

new product's Actual data and Forecast data. The graph displays
the new product launch. ForecastX™ recognizes that after
several periods, the growth will taper off. As the growth rate
decreases, the line plateaus.

Section 19: Poisson

Poisson is the simplest traffic model, and is used to work out how
many lines are required if the traffic figure (in Erlang) during the
busiest hour is known. The Poisson distribution is based on the
following assumptions:

• Calls are served in random order.

• There are an infinite number of sources.

Rev 0 164
Chapter 4

• Blocked calls are held.

• Holding time is exponential or constant.

The usage of Poisson in ForecastX is similar to ErlangB. See

below for information on ErlangB.

To use the Poisson forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Poisson.

The Poisson Forecasting technique displays.

3. Select Edit parameters to activate Poisson’s parameters.

4. In the Compute Type area, select either:

• N to A
• A to N

Rev 0 165
Chapter 4

5. In the Lost Possibility area, select one of the options for the
percentage of loss that is possible.

6. Click Finish.

Section 20: Polynomial Regression

Polynomial Regression models can contain one, two, or even several

Independent Variables similar to that of a Multiple Regression model.
However, Polynomial Regression goes further and treats the
relationship between the Dependent and Independent Variable in
more than a linear way. Polynomial Regression enables the
Independent Variables to be present in various powers.

To use the Polynomial Regression forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Polynomial Regression method is used. For your company’s
purposes, you will have your own data available.

2. Click on the Polynomial Regression sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

The Data Capture window will be displayed.

4. Click on the Forecast Method tab and select Polynomial

Regression as your forecasting Technique.

Rev 0 166
Chapter 4

5. In the Parameters area, configure the options available. The

following table details what each option means.

Parameter Description

Dependent Select the dependent series of the Polynomial

Series Regression.

maxDegree Determine the maximum possible degree of

Independent Variables in the model. It is
recommended that no more than three is used.

FPartial F-value used to determine the highest degree of

the Independent Variables used in the model.

Rev 0 167
Chapter 4

Parameter Description

minFin Determine the minimum F-value of an

Independent Variable that is used in the model.

Note: this is a Stepwise Regression parameter

than can help determine the most significant
Independent Variables in the model.

maxFout Determine the maximum F-value of an

Independent Variable that is used in the model.

Note: this is a Stepwise Regression parameter

than can help determine the most significant
Independent Variables in the model.

6. On the Reports tab, select the Audit Trail report and ensure that
all other reports are not selected.

7. Click Finish.

Section 21: Probit Curve

The Probit Curve is one of the three new product models used in the
New Product forecasting method, but it is also a stand-alone
forecasting technique. For additional information, please see Section
18: New Product Forecasting.

To use Probit Curve forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Probit Curve.

Rev 0 168
Chapter 4

The Probit Curve Forecasting technique displays.

3. Select Edit parameters to activate Probit Curve’s parameters.

4. In the Value Limit area, select the minimum and maximum

values.

5. Select either Optimized or Fixed.

Option Description

Optimized Optimize the maximum value specified. The

maximum value might be changed to a value lower
than specified but this is done to optimize the
forecast generated.

Fixed Maximum value set is used as the maximum for

the created forecast.

6. Click Finish.

Rev 0 169
Chapter 4

Section 22: Simple Exponential Smoothing

Simple Exponential Smoothing uses both Historical data and fitted

data to generate a new forecast. Simple Exponential Smoothing gives
higher importance to recent values than older observations. For
example, in a three-period Moving Average forecast, February, March
and April actual values will evenly determine May's forecast.
Depending on the level given to the data series, a Simple Exponential
Smoothing Method creates a forecast for May that relies more so on
April's observations. Simple Exponential Smoothing is most effective
as a forecasting method when irregular influences comprise the main
effects on the Historical time series values.

The level of smoothing is always a value between zero and one. If the
level value is closer to 1, then the forecasted value is going to more
closely reflect the last actual value. If the values is closer to 0, then
the forecasted value is going to more closely reflect the first actual
value.

To use the Simple Exponential Smoothing forecasting method:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Simple Exponential Smoothing method is used. For your
company’s purposes, you will have your own data available.

2. Click the Simple Exp. Smoothing tab at the bottom of the Excel
sheet to select the example spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .
ForecastX displays.

4. Click on the Forecast Method tab.

5. In the Forecast Technique area, scroll through the list of methods

and select Simple Exponential Smoothing.

Rev 0 170
Chapter 4

The Simple Exponential Smoothing Forecasting technique displays.

6. Select Edit parameters to activate Simple Exponential

Smoothing’s parameters.

7. In the Smoothing Constants area, type in a number for Level.

8. On the Reports tab, select the Audit Trail report. The Audit Trail
report provides a graph of your forecast and analyze the statistics
used to measure the Simple Exponential Smoothing forecast.

9. Click Finish.

ForecastX™ provides two tools to judge the accuracy of your

forecast and the direction your business is heading. The first
thing you see is the Executive Report chart.

Rev 0 171
Chapter 4

One line represents the fitted values and the forecasted values of
the actual values. The other line shows the actual values.

ForecastX™ also produces a detailed table that shows the optimized

parameters for your Exponential Smoothing forecast.

You can see, on the table, that this forecast has an average error of
6.53%. This means that ForecastX™ smoothed your data to a
considerable degree. As before mentioned, a lower smoothing value
represents a high level of smoothing.

Rev 0 172
Chapter 4

Section 23: Slow Moving Demand

Slow Moving Demand is used to calculate the optimal Stock level for
sporadic data. The forecasting method works well with items that
consist of many zeroes and items that have variable Non-seasonal
demand. Croston forecasting method assumes that demand has
normal distribution; Slow Moving Demand assumes that all demand
has some Seasonality.

When this forecast method is used the if the Safety Stock opinion line
is used in any of your reports, including the DRP reports, the Safety
Stock will be taken from the calculations provided by this forecast
method.

To use the Slow Moving Demand forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Slow Moving Demand method is used. For your company’s
purposes, you will have your own data available.

2. Click the SlowMethod tab at the bottom of the Excel sheet to

select the example spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .
ForecastX displays.

4. Click on the Forecast Method tab.

5. In the Forecast Technique area, scroll through the list of methods

and select Slow Moving Demand.

Rev 0 173
Chapter 4

The Slow Moving Demand Forecasting technique displays.

6. Select Edit parameters to activate Slow Moving Demand’s

parameters.

There are two different types of configurations. Each works in the

same way. However, the Step by step configuration walks you
through the process and provides information to help you make
the best selections.

• Step by step configuration

• Express configuration

23-1: Step by Step Configuration

Please see Express Configuration for detailed information on each

selection provided. To use the Step by step configuration:

Rev 0 174
Chapter 4

1. From the Forecast Method tab, click on the Step by step

configuration link.

The Stocking Level window displays.

2. Click the forward arrows to go to the next step.

Rev 0 175
Chapter 4

The Event Index Creation window displays.

3. In the Event Index Creation area, select either Seasonal or Std

Bucket to define the Seasonality. If you selected Std Bucket,
check Exclude 0 if you do not want zeroes in the Event Index
creation.

4. Type in the Start, Increment and End values.

5. Click the forward arrows to go to the next step.

Rev 0 176
Chapter 4

The Number of Simulations window displays.

6. In the Number of simulation area, select either Fixed or % of

threshold. Type in a number for either option selected.

7. In the Event Replaceable area, select Yes or No.

8. Click the forward arrows to go to the next step.

Rev 0 177
Chapter 4

The Best Calculation Method window displays.

9. In the Base Calculation Method area, select either Mean or

Croston. If you chose Mean, check the Exclude 0 checkbox if you
do not want 0 to be part of the calculation.

10. Click the forward arrows to go to the next step.

Rev 0 178
Chapter 4

The Base Forecast Method window displays.

11. In the Base Forecast Method area, select either Same or Other. If
you chose Other, select a forecasting method from the dropdown
list.

12. Click the forward arrows to go to the next step.

Rev 0 179
Chapter 4

The Service Level As window displays.

13. In the Service Level As area, select either % of Total or Std of

Sample.

14. Click the forward arrows to go to the next step.

Rev 0 180
Chapter 4

The Safety Stock window displays.

15. In the Safety Stock area, type in the Lead Time.

16. Select a percentage level in the Level-50 dropdown.

17. Click the forward arrows to go to the next step.

Rev 0 181
Chapter 4

The Service Level in History window displays.

18. In the Service Level in Hist As area, select either % of Amount or

% of Periods.

19. Click Finish.

23-2: Express Configuration

To use the Express Configuration:

1. From the Forecast Method tab, click on the Express configuration

link.

Rev 0 182
Chapter 4

The Slow Moving Demand window displays.

The table below details what each feature means. Please see
Step by Step Configuration if you would like to see how the Slow
Moving Demand walks you through the selections provided.

Feature Description

Event Index Defines the Seasonality.

Creation

Seasonal Identifies the data is cyclical. Enter the number of

periods that define the season cycle.

Rev 0 183
Chapter 4

Feature Description

Std Bucket Creates an Event Index based on the Mean and

the Standard Deviation in the data.

Exclude 0 Excludes the leading zeroes in the calculation.

(Event Index
Creation)

Start Type in the first value in the range for the Event
Index.

Inc Defines the amount of lift from period to period that

should be included in the Event Index.

End Type in the last value in the range for the Event
Index. The value should be at least two points
greater than the Start value.

Base Calculation Determines the method used to calculate the

Method Historical baseline for the Event Model.

Mean Uses the Mean as the Historical baseline.

Croston Calculates the Historical baseline using the

Croston method.

Exclude 0 Excludes zeroes in the calculation.

(Base
Calculation
Method)

Service Level As Defines the Safety Stock levels.

% of total Sorts the simulated values in ascending order and

then selects the value that is tied to the percentage
specified. Uses the percentage that you entered in
the Safety Stock Level value.

Std of Sample Determines if the Standard Deviation of the

simulated data is to be used as the Service Level
for Safety Stock. Assumes that the data follows a
normal distribution.
Rev 0 184
Chapter 4

Feature Description

Service Level in Calculates the Historical service level, assuming

Hist As that the Safety Stock value in the past is the same
as the Forecasted Safety Stock value.

% of Amount Calculates the Historical service level in a

percentage of the total history. Divides the total
history under/out of stock by the total demand in
the History.

% of Periods Calculates the percentage of periods that were

either under/out of stock. Takes the total history
periods that were under/out of stocks and divides it
by the total demand periods in the History.

Number of Determines the number of simulations performed

simulation when generating the Event Index values.

Fixed Indicates that a specific number of simulations are

performed.

% of thresd Sets a threshold for the variance that indicates the

simulation should end. When the variance
percentage between the current number
generated and the previous number does not
exceed the threshold, the simulation process is
stopped.

Event Specifies whether the same period’s value can be

Replaceable used more than once when evaluating the data for
a Seasonality index.

Base Forecast Determines the forecasting technique used to

Method calculate the Forecast values.

Same Identifies that the same forecasting technique

(Mean or Croston) used to calculate the Historical
baseline is also used to calculate the base
forecast.

Safety Stock Defines the parameters used to calculate the

Safety Stock values.
Rev 0 185
Chapter 4

Feature Description

Lead Time Number of periods of future demand that is to be

kept in reserve.

Lead - 50 Percentage of periods of future demand that is to

be kept in reserve.

Section 24: Stepwise

Stepwise Regression is used to determine the relevant Explanatory

Variables from a set of Independent Variables. Explanatory Variables
are the variables most useful to explain the dependent variable.
Explanatory variables are typically included in the final Regression
model. When the number of independent variables is extremely large,
it makes it impossible for common regression models to compute and
determine all of the explanatory variables. In such cases, Stepwise
Regression is used to determine explanatory variables. There are
three different types of Stepwise Regression.

24-1: Stepwise Forward Regression

From all the explanatory or independent variables, this method

selects the one with the highest correlation with Y, the dependent
variable. It then determines the Residuals, or Errors, that result from
this regression and consider these residuals a new set of Y values or
dependent variables. From the remaining independent variables, it
selects the one that correlates most highly with these residuals.
(Remember to consider these residuals as a new set of dependent
values.) Continue this process until no remaining explanatory or
independent variable has a significant relationship with the last set of
variables.

Note: The term Error in forecasting is often interchangeable with Residual.

Both refer to the difference between the forecasted and actual values.
The Error or Residual Opinion Line can be added to a report by
selecting the ‘Error’ option from the Opinions section.

Rev 0 186
Chapter 4

24-2: Stepwise Backward Repression

This method starts with a regression that includes all the variables,
and weeds out the variable that is least significant in the equation, as
measured by the t-test. When this variable is removed, another
regression is performed to determine the next variable to be
removed. This process is repeated until the most significant variable
remains.

24-3: Stepwise Forward with a Backward Look Regression

Combining the advantages of Stepwise Forward Regression and Step

Backward Regression, this method is the most powerful and the most
popular of the three. This method works through an iterative process
to determine the best explanatory variables. When it is no longer
possible to find any new variable that contributes to the R-Squared
value, or if no variable needs to be removed, the iterative process
stops.

To use the Stepwise forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Stepwise method is used. For your company’s purposes, you
will have your own data available.

2. Click the Stepwise Regression tab at the bottom of the Excel

sheet to select the example spreadsheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .
ForecastX displays.

4. Click on the Forecast Method tab.

5. In the Forecast Technique area, scroll through the list of methods

and select Stepwise.

Rev 0 187
Chapter 4

The Stepwise Forecasting technique displays.

6. Select Edit parameters to activate Stepwise’s parameters.

7. In the Parameters area, configure the options available. The table

below details what each option means.

Parameter Description

Dependent Selects the dependent Series of the model.

Series

Stepwise/ Indicates the type of model.

Backward/
Forward

minSeries Indicates the minimum number of the independent

variables that will include in the model.

maxSeries Indicates the maximum number of the independent

variables that will include in the model.
Rev 0 188
Chapter 4

Parameter Description

minFIn Indicates the minimum F-value of an independent

variable that can be allowed into the model.

maxFOut Indicates the maximum F-value of an independent

variable that should be removed from the model.

minLag Indicates the minimum lag used in the model.

maxLag Indicates the maximum possible lag in the model.

LagDep Includes a lagged dependent variable in the

model.

8. On the Reports tab, select Audit Trail report.

9. Click Finish.

The graph you produce with ForecastX™ shows you a smoothed

forecast of sales over the next few periods.

If you needed more detail, you could scroll down in the report to find
the statistics

Rev 0 189
Chapter 4

The summary contents show you the forecast has a Mean Absolute
Percentage Error of 5.26%. As you look further down the report, you
notice the Coefficient Table. Each series is listed in the left-hand
column.

Agter most of the series, there is a Lag amount in parenthesis. This

signifies there is a Lag dependency in the series. For this example,
the Lag represents the days (independent series) it takes for the
advertising method to affect the sales production (dependent series).
For example, Print Media Advertising has Lag 1 following it. This
means that the Print Media Advertising method does not have a
direct effect on the sales production until after 1 day.

One column over, you can see the independent variables ForecastX
included in your regression model: Internet Advertising, Radie
Advertising and the Competition Index. As you notice from the T-test
and F-Test, the most significant independent variable is Internet
advertising. This is easy to notice since the targets T-Test and F-test
represent the most significant independent variable.
You can now confidently assert that Internet Advertising is having the
most effect on your Super Chewy Bar sales.

Rev 0 190
Chapter 4

Section 25: Trend (Linear) Regression

Trend (Linear) Regression analysis uses an equation to analyze the

relationship between two or more quantitative variables in order to
predict one from the other(s). Linear Regression measures the
relationship between two variables: X and Y. X is the independent
variable and Y is the dependent variable. A particular observation of
Y depends on X and an additional random error.

Y = mX + b

ForecastX™ estimates the best-fit regression line for any input data.
However, the user must first make an analysis about the estimated
model and whether it provides meaningful information. A question
that often arises is whether the forecasted relationship between X and
Y exists, and if it does, how good does the regression line fit. There
are several ways to evaluate and judge the goodness-offit of a
regression.

The variance of the dependent variable around the independent

variable can be estimated using sample data. A common statistic
used to judge the accuracy of the regression is called the Standard
Error of the estimation. A large Standard Error indicates a lot of
residual or error variation, which signifies a poor fit. If the Standard
Error is small, there is a minimal amount of residual variation, thus the
fit is good. The Standard Error of the estimation helps determine how
accurate the model is for prediction. The difficulty in using the
Standard Error is making judgments about how large errors can be
before they become unacceptable.

Another common statistic is the Coefficient of Determination. The

Coefficient of Determination, or R-Squared, helps make a better
judgment about the model's goodness-of-fit.

The R-Squared produces a percentage error term rather than a

numerical one. The percentage represents the variation in the Y
variable as explained by the regression. If the R-Squared is large
(98%, 89%, etc.), the model is providing a good fit and you can have
confidence in its ability to predict. However, if the RSquared is small,
the model is not providing a good fit and you should have less
confidence in its ability to predict. Even with the R-Squared, there is
still ambiguity about how large the percentage needs to be in order to

Rev 0 191
Chapter 4

be considered a good fit. Hypothesis tests such as the F-Test and T-

Test enable us to make better judgments about the model as a whole.

The T-Test is used to make a judgment about the variable m, or the

slope coefficient. This test indicates whether m is significantly
different than zero. Again if m is significantly different than zero, there
is a relationship between the dependent and independent variable.
For Linear Regression, the F-Test and the T-Test deduce the same
conclusions. Audit Trail offers users great detail of the corresponding
statistics in ANOVA table and coefficient table. For more information
about these statistics, refer to “Statistics” on page 202

Though in general linear regression, the values in X can be any

number, most useful linear regression model uses predefined values
in the series X. In ForecastX , the values are 1, 2, 3, …. Since the
values are fixed, X is an implicit series, users does not provide this
series in their input data. In some case, a linear regression model
with b=0 is needed, where m is the interception. ForecastX provides
an option to force the value of m to be zero.

To use the Trend (Linear) Regression forecasting technique:

1. Click on and open the ForecastingExamples.xls file.

Note: The ForecastingExamples.xls file is a data example to demonstrate

how the Trend (Linear) Regression method is used. For your
company’s purposes, you will have your own data available.

2. Click on the Linear Regression sheet.

3. Click in a cell containing data, and open ForecastX™

by clicking on .

The Data capture window is displayed.

4. Click on the Forecast Method tab and select Trend (Linear)

Regression from the Forecast Technique drop down list.

Rev 0 192
Chapter 4

5. On the Reports tab, select Audit Trail report.

6. Click Finish.

The Audit Trail graph depicts three aspects of your data:

• Actual observations

• Smoothed values

• Forecasted values

Rev 0 193
Chapter 4

The graph shows the trend of your sales data.

The Audit Trail report calculates the accuracy and descriptive

statistics you need to determine if the Regression model chosen is
optimal.

The Mean Absolute Percentage Error (MAPE) is 10.17%, and

represents the average of the errors in your forecast. The Audit Trail
Coefficient Table displays your T-Test and F-Test values, which
indicate the goodness-of-fit of your regression model. Since the F-
Test value is large, 4.92, you know that the independent variable time
in your example has a significant influence on the dependent variable
sales.

Based on the statistics generated, Trend (Linear) Regression is an

optimal Forecasting method for this particular data set. Trend (Linear)
Regression is best used when one independent variable significantly
influences a dependent variable.

Rev 0 194
Chapter 4

Section 26: Triple Brown

Brown's Triple Exponential Smoothing is implemented on data

showing a quadratic trend over time. It works well for data that has a
very steep growth or decline. The Brown's Triple Exponential
Smoothing method attempts to create a linear equation by performing
three simple exponential smoothing forecasts and then adjusting for
the linear trend in the data.

To use the Triple Brown Forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Triple Brown.

The Triple Exponential Smoothing Brown Forecasting technique

displays.

3. Select Edit parameters to activate Triple Brown’s parameters.

Rev 0 195
Chapter 4

4. In the Smoothing Constants area, type in a number of Levels.

5. Click Finish.

Section 27: Weighted Moving Average

Weighted Moving Averages are moving averages of moving

averages. In other words, rather than replace the oldest observations
within the data, the Weighted Moving Average replaces the oldest
moving average with the most recent moving average. In this way,
Weighted Moving Averages handle significant data trends better than
a plain moving average. Weighted Moving Average forecasts are
used to overcome the strong affect of extreme values within a time
series by assigning current data more weight than older data. The
start and history parameters are the same as those in moving
averages.

To use the Weighted Moving Average forecasting technique:

1. Click on the Forecast Method tab.

2. In the Forecast Technique area, scroll through the list of methods

and select Weighted Moving Average.

Rev 0 196
Chapter 4

The Weighted Moving Average Forecasting technique displays.

3. Select Edit parameters to activate Weighted Moving Average’s

parameters.

4. In the Parameters area, type in a number for the Average

Periods.

5. Click Finish.

Rev 0 197
Chapter 5 Statistics

This chapter includes:

Section 1: Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Section 2: Accuracy Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 204
Section 3: Analysis Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 212
Section 4: Regression Statistics . . . . . . . . . . . . . . . . . . . . . . . 217

Section 1: Overview

The Statistics tab in ForecastX™ is used to retrieve all the statistic

types that can be used for your forecast. The statistics generated
measure the accuracy of the forecast and enable you to
understand your data. The statistics you choose are viewed in
reports. ForecastX™ groups the statistics in three different types:

• Accuracy

• Analysis

• Regression

This chapter details all the statistics and how they can be used
optimally.

Rev 0 198
Chapter 5

From ForecastX, click on the Statistics tab. The Statistics window displays. You
can also view more statistics by clicking on the More Statistics link at the bottom
right corner of the window.

Rev 0 199
Chapter 5

Section 2: Accuracy Statistics

This section details the statistics that can be found under in the
Accuracy measures area of the Statistics tab, as well as the Accuracy
tab after clicking on the More Statistics link.

2-1: AIC

Akaike Information Criterion, AIC is a statistic that aids in the Box-

Jenkins specification stage of model building. A model that minimizes
the AIC is considered to be the most appropriate model. This implies
that when several models may be appropriate, choose the one with
the smallest number of free parameters, thus the smallest AIC.

2-2: BIC

Bayesian Information Criterion (BIC) is an order estimation criterion

used in the Box -Jenkins specification stage of model building. A
model that minimizes the BIC is considered to be the most

Rev 0 200
Chapter 5

appropriate model. When several models may be appropriate,

choose the one with the smallest BIC. The AIC and BIC differ in their
second terms, which are penalty functions for extra parameters.
Often AIC and BIC will lead to the same model choice.

2-3: R-Square

R-Squared is referred to as the coefficient of determination. It tells the

amount of variance in Y that can be explained by variable X. In other
words, Variance is how dispersed the data is, calculated by how far
away the data points are from the average data point. Where Y is the
dependent variable and X is the independent variable, R-Squared
determines the level of influence X has on Y's variance.

2-4: Means Absolute % Error

The MAPE is commonly used in quantitative forecasting methods

because it produces a measure of relative overall fit. The absolute
values of all the percentage errors are summed up and the average is
computed. In comparison to Mean Error (ME), which is determined
simply as the average error value and affected by outliers (large
positive and negative errors can cancel each other out resulting in a
zero error), or Mean Absolute Error (MAE), which de-emphasizes
outliers by their average, the MAPE is a more meaningful
measurement.

For example, if you are forecasting sales that vary greatly from month
to month and the MAPE is +5%, it is a more useful result than a mean
error. The MAPE also de-emphasizes outliers, but produces results
calculated as the average absolute error in percentage terms, which
are easily interpretable.

2-5: Adjusted R-Square

The coefficient of determination establishes that when an explanatory

variable is added to a model, R-Squared decreases, no matter how
useless the additional variable. So, R-Squared is the fraction of
variance explained by the model.

Rev 0 201
Chapter 5

Ideally, the measure of fit would decrease when useless variables are
entered into the model as explanatory variables. In other words, if
the measure of fit decreases every time a useless variable is entered
into the model, then the analysts can measure and determine which
variables to keep and which to expunge. A statistic widely used to
achieve this is the coefficient of determination adjusted for the
number of parameters in the model (Adjusted R-Squared). Adjusted
R-Squared tells you when the negative affect of the variable
outweighs the positive.

2-6: Sum Square Error

Sum Squared Error (SSE) is an accuracy measure where the errors

are squared, then added. It is used to determine the accuracy of the
forecasting model when the data points are similar in magnitude. The
lower the SSE the more accurate the forecast. Understanding this
accuracy statistic will help you choose which forecasting model best
fits your data.

2-7: Safety Stock

Safety Stock is designed to allow you cushion when planning for

demand. If there is demand in your forecast, you can set a specific
number of periods of demand to have. Unlike other statistics, safety
stock is only reported in the forecast table of Executive Report.

Rev 0 202
Chapter 5

1. From the Statistics tab, click on the Change Settings link in the
Safety Stock area.

Rev 0 203
Chapter 5

The Safety Stock Settings window displays.

2. Select either Input parameters or Input Value, Slowing Moving

Output will be automatically selected when the Slow Moving
Demand Forecast method is used.

The table below details the options available.

Input Description
Parameters

Include Stat Fcst Indicate whether to include the Statistic Forecast in

the calculation of the Safety Stock. When set to
Yes this will make will make your Safety Stock a
Stocking Level.

Lead Time Specify the lead time of the Safety Stock. Lead
Time is the periods that the safety stock covers.

Service Level Specify the service level that the Safety Stock
meets.

Safety Stock Select this option to calculate the Safety Stock.

Rev 0 204
Chapter 5

Input Value Description

Safety Stock This field is used when you need to set your Safety
Stock to a default value for all your series or use a
DRP parameter.

As Stocking Enable this option if you wish to make your Safety

Level Stock a Stocking Level. This option is useful for
DRP reports because it will make your
Recommended orders include the Safety Stock.

3. Click OK when finished.

2-8: Chi-Square

Chi-square is a commonly used goodness-of-fit test to compare

actual data with fitted data according to a specific hypothesis, namely,
the null hypothesis. The null hypothesis states that there is no
significant difference between the fitted value and actual data. This
statistics return the p-value of the test instead of the value of Chi-
Square test. The range of p-value is from 0 to 1. The closer to 1, the
better fit.

2-9: Cochran-Orcutt

When there is a strong indication of the presence of auto correlated

errors in a forecasted model (especially, a regression model), it is not
suitable to proceed with ordinary least squares method. But the
alternative estimation to the new data set Yt - r* Y(t-1) will have better
result for appropriated r. Cochrance-Orcutt is the method to estimate r.

2-10: Means Absolute Error

The Mean Absolute Error is similar to the Mean Error, except the
MAE considers the absolute values of the errors. In other words, it will
take negative values and replace them with their absolute (positive)
values. This method de-emphasizes large outliers, because (negative

Rev 0 205
Chapter 5

and positive results aren't able to cancel each other out). A zero MAE
represents a perfect fit.

2-11: Mean Error

The Mean Error (ME) value is calculated as the average error value.
It is less reliable in some cases than other accuracy statistics
because there is the risk that large outliers can cancel each other out,
producing a ME near zero, which would indicate a perfect fit. For
example, positive and negative error values could produce an
average of zero, which may not be an accurate description of the
overall fit of the forecast.

2-12: Mean Square Error

Mean square error is one of the traditional measures of forecast

accuracy. Mean square error is used when all of the errors are similar
in magnitude. If the data contains one or two large errors, calculate
the mean absolute error (MAE), since using sum squares magnifies
these errors.

Also use the MAE or MSE to select the right forecasting model by
choosing the model that results in the smallest MAE or MSE. Keep in
mind that you cannot compare forecast models that used different
data transformations and you cannot compare MSE to MAE.

2-13: Normality Error

A standard assumption of forecasting models is that the error terms

are normally distributed. The Normality Error is the error associated
with the single time series normal distribution.

2-14: Root Mean Square Error

The Root Mean Square Error is the square root of the Mean Square
Error. It is calculated by taking the square root of the arithmetic mean
of squared deviations from the mean. It is also called standard
deviation, when the deviations do not represent errors.

Rev 0 206
Chapter 5

2-15: Standard Deviation Error

The standard deviation error is used in regression analysis to

measure the average variation of the observed values around the
forecast values. It answers the question, "How far away are the
forecasted values from the actual values in the data set?"

2-16: Theils

Theil's Statistic is an accuracy measure that compares the accuracy

of a forecast model to the actual model. Theil's uses the actual value
of the last time period as the forecast. The closer Theil's statistic is to,
zero the more accurate the forecasting model. For instance, if Theil's
Statistic is equal to 1, the forecast is completely inaccurate. If it is
equal to zero then the forecasting model is a perfect fit.

2-17: T-Test

The T-test determines whether or not the parameter or series

contributes to the accuracy of the model. If the T-test statistic is
greater than 1.96, the parameter or series does contribute to the
accuracy of the model.

This test is performed automatically for certain reports and forecast

methods. The T-Test is only included in an Audit report and only when
the chosen forecast method is one of the following:

• Trend (Linear) Regression

• Multiple Regression

• Stepwise

• Polynomial Regression

• Box Jenkins

Rev 0 207
Chapter 5

Section 3: Analysis Statistics

This section details the statistics that can be found under in the
Analysis area of the Statistics tab, as well as the Analysis tab after
clicking on the More Statistics link.

3-1: Durbin Watson

When measuring accuracy, it is necessary to test for auto-correlation

within the errors. Auto-correlation occurs when there is dependence
between the successive error values, which is also called a serial
correlation.

Durbin Watson is the most widely used statistic to determine random

errors. The Durbin Watson statistic always lies between 0 and 4. If it
is closer to zero, it indicates a positive auto-correlation. A value close
to 4 indicates negative auto-correlation. A value close to 2 tends to
reinforce the conclusion that no correlation exists among the error.
Therefore, values below 2 represent positive serial correlation
amongst the errors, and values above 2 represent negative serial
correlation amongst the errors.

Rev 0 208
Chapter 5

3-2: Mean

The mean of a group of numbers is synonymous with the average. It

is calculated by summing up the values of the items and then dividing
the total by the number of items summed. Like the median, it is often
used as a measure of location for a frequency or probability
distribution.

3-3: Median

The median of a sequence of items is the middle item after the entire
sequence is arranged in either ascending or descending order. This is
often used to measure the location for a frequency or probability
distribution.

3-4: Standard Deviation

The standard deviation is a descriptive statistic, which is derived from

taking the square root of the variance. The variance allows us to see
how far each data point is from the average point (mean) in the data
set.

3-5: Variance

The variance is a descriptive statistic that measures the dispersion of

the data. Variance is calculated by finding out how far each data point
is from the Mean.

3-6: Mean Square Deviation

The Sum Squared Deviations (SS) is divided by the degrees of

freedom associated with it to obtain the Mean Squared Deviation
(MS). The sample variance is often called a mean square because a
sum of squares has been divided by the appropriate degrees of
freedom

Rev 0 209
Chapter 5

3-7: Correlation Coefficient

The correlation coefficient tells how strongly two variables are related.
The value of the correlation coefficient is always between -1 and 1.
That means that if the value is 0, then the two variables (X and Y) are
independent. If the correlation coefficient is positive, it means when X
is large, Y is likely to be large also.

If the correlation coefficient is closer to 1, then X and Y are almost

directly related. However, if there is a negative correlation, it means
when X is large, Y is likely to be small. The closer the correlation
coefficient is to -1, the stronger the negative correlation.

3-8: Max

Max (maximum) is the largest value in your data set.

3-9: Min

Min (minimum) is the smallest value in your data set.

3-10: Kurtosis

The Kurtosis of a distribution measures the thickness of the tails of a

distribution. See Skewness on page 121. Kurtosis is a measure of
flatness and tail thickness of a distribution, as compared with the
normal distribution.

Positive kurtosis, or leptokurtic, indicates a relatively peaked

distribution. Negative kurtosis, or platykurtic, indicates a relatively flat
distribution. For a normal distribution, kurtosis is equal to 3. When the
tail of the distribution is thicker than the normal distribution, kurtosis
will be greater than 3. When the tail of the distribution is thinner than
the normal distribution, the kurtosis will be less than 3.

3-11: Skewness

Like mean and variance, which calculate the distribution of a random

variable, the skewness of a distribution measures whether the

Rev 0 210
Chapter 5

distribution is symmetrical. A symmetric distribution has a skewness

of 0. If the tail of the distribution is long and in the positive direction,
then it has positive skewness. If the distribution has a long tail in the
negative direction, then the skewness is negative.

3-12: Range

The range is the difference between the max and the min.

3-13: Auto-Covariance

AutoCovariance is a measure in a single time series of the linear

relationship between two data points separated by a constant time
interval or specified number of lags. AutoCovariance shows whether
time related patterns exist within a given time series.

3-14: Auto Correlation

Auto-correlation describes the dependence between two variables of

the same time series at different time periods. Auto-correlation helps
determine if there is a causal connection between two variables even
though there is a time lag between their occurrences. For example,
auto-correlation can be useful in sales for determining if a special
promotional event occurring at regular intervals is a causal factor in
increased sales after the events happen. Auto-correlation answers
the question, "Even though X happens with Y, is Y dependent on X?"

Not only can auto-correlation be used to determine seasonality, it also

helps describe if a series is stationary. A stationary series is one that
has properties independent of the time period in which they occurred.

3-15: Coefficient of Variation

The relative amount of dispersion in the data can be calculated by

expressing the standard deviation as a percentage of the mean. This
measurement is called the coefficient of variation. This allows
comparison of the dispersions of two separate data sets that are
entirely different.

Rev 0 211
Chapter 5

3-16: Sum of Squared Deviation

Given a single data series, each observation is subtracted from the

estimated mean of the data series, and then squared. All
observations are then summed to produce the Sum Squared
Deviations. The Sum Squared Deviations are used to calculate the
sample variance, alternatively called the Mean Squared Deviation
(MS), of a single data series.

3-17: Root Mean Square

The Root Mean Square error is a measure of the deviation of the

forecasted value from the actual value. In Root Mean Square the
deviations are summed and then divided by the number of time
periods in the time series. Lastly, the square root of this quantity is
evaluated. The Root Mean Square is used to quantitatively measure
how closely the forecasted variable tracks the actual data. The
magnitude of the Root Mean Square error can be evaluated only by
comparing it to the mean of the time series.

3-18: Mean Absolute Deviation

The Mean Absolute Deviation is the mean of the absolute value of the
differences between the supplied values and the arithmetic mean of
all values. The formula for Mean Absolute Deviation is:
n
MAD = (∑t =1 | xt − u |) / n

The variables used in Mean Absolute Deviation are defined as

follows: n=the number of values; x=the supplies values; µ=the
arithmetic mean of all the values.

3-19: Mode

Mode is the value or values that appear most frequently within your
data set.

Rev 0 212
Chapter 5

3-20: Ljung Box

Ljung Box tests for the overall correlation of the fitted errors of a
model. It determines how a variable relates to itself when it has
lagged one or more periods. If the correct ARIMA model is estimated,
the Ljung Box statistic tends to be smaller. Conversely, an erroneous
model inflates the Ljung Box statistic.

Section 4: Regression Statistics

This section details the statistics that can be found under the
Regression tab after clicking on the More Statistics link.

Rev 0 213
Chapter 5

4-1: ANOVA Table

Analysis of variance (ANOVA) is used to uncover the main and

interaction effects of independent variables on a dependent variable.

There are two rows in the output of an ANOVA table. One is for
regression, which represents the part of the dependent variable that
is interpreted by the regression model. The other is for error, which
represents the part not interpreted by the regression model.

„ SS is the sum squared of deviation.

„ df is the degree of freedom.

„ MS is the mean of SS.

„ SEE is the squared root of MS.

The bigger ratio of the MS of the regression and error, the better of
the model. In the above graph, the ratio is 4145489/59359, which is
around 70. This dependent variable is well explained by the
independent variables in the regression model.

Rev 0 214
Chapter 5

4-2: Coefficient Table

This table summarizes the significance of each independent variable

and the overall significance of the regression model.

The table below details what each column means:

Column Description

Series The description of the dependent and independent

Description variable. If the regression uses lag options, this
column also indicates the lagged degree of the
variables.

Included in Indicates whether a variable is included in the

Model regression model or not. It also indicates which
variable is the independent variable.

Coefficient The coefficient of each independent variable

included in the model. For the dependent variable,
it is the value of constant term.

Standard Error The standard error of the corresponding variable.

T-Test The t-test of the corresponding variable. Generally,

if the absolute value of a t-test is greater than two
means the corresponding variable is significantly
important in this regression model.

P-Value Small p-values means the corresponding variable

is significant.

F-Test Generally, if the value of F-test is more than four,

then the variable is significant.

Rev 0 215
Chapter 5

Column Description

Elasticity Indicates the percentage change of dependent

variable when the corresponding variable changes
in one percent. It is one way to tell whether one
independent variable has a large effect on the
dependent variable or not.

Overall F-Test The F-test of the whole regression model. The

bigger value, the better model.

4-3: Coefficient Determination

The coefficient of determination, or R-Squared, tells the proportion of

variance in Y that can be explained by X. X is an independent
variable that has some influence on Y's variance. It can also explain
how dispersed the data points are from the mean data point. This can
be useful in determining if an independent variable affects the
seasonality of your data.

4-4: Covariance

The degree of dependence between two random variables is

measured by covariance. This is useful if you have two random,
dependent variables, X and Y.

Rev 0 216
Index

Covariance 220
A
Croston Intermittent 131
Accuracy Statistics 204
Customize ARIMA Model Input 124
Adaptive Exponential Smoothing 116
Adjust 57
D
Adjusted R-Square 205
Data Analysis 124
AIC 204
Data Capture tab 49
Analysis Statistics 212
Data Cleansing 51
Analyze 58
Data Transformation 124
Append 94
Decomposition 134
Audit Report 90
Differencing 125
Auto Correlation 215
Double Brown 136
Auto-Covariance 215
Double Holt 137
DRP Report 75
B Durbin Watson 212
Bass Model 119
BIC 204
E
Bill of Materials 12
Erlang B 141
Box-Jenkins 121
Erlang C 145
Executive Report 89
C Express Configuration 186
Census X-11 125
Chi-Square 209
F
Cochran-Orcutt 209
Forecast Adjust 36
Coefficient Determination 220
Forecast Method Tab 55
Coefficient of Variation 215
Forecast Relocation 38
Coefficient Table 219
ForecastX Highlights 10
Correlation Coefficient 214

Rev 0 217
Index

Means Absolute Error 209

G
Median 213
Group By Tab 60
Min 214
Mode 216
H
Moving Average 161
Holt-Winters 156
Multiple Regression 162

I
N
Installing ForecastX 20
New Product Forecasting 166
Installing ForecastX™ 20
Normality Error 210

J
P
John Galt Solutions Online Technical Sup-
Performing ARIMA Analysis 124
port 15
Pivot Report 92
Poisson 168
K
Polynomial Regression 170
Kurtosis 214
Preview 60
Probit Curve 172
L
Purchase Order 12
Ljung Box 217
Logistic Curve 159
R
Range 215
M
Report Bursting 100
Max 214
Reports Advanced Options 95
Mean 213
Reports Tab 66
Mean Absolute Deviation 216
required hardware 17
Mean Error 210
required software 18
Mean Square Deviation 213
Revision History 16
Mean Square Error 210
Root Mean Square 216
Means Absolute % Error 205
Root Mean Square Error 210

Rev 0 218
Index

Rough Cut Highlights 10 Triple Brown 199

R-Square 205
U
S User groups 11
Safety Stock 206 Using Advanced Census X-11 with Fore-
castX 129
Scenario 102
Screen Captures 11
Setup.exe 20
V
Variance 213
Simple Exponential Smoothing 174
Skewness 214
Software License 14 W
Standard Deviation 213 Weighted Moving Average 200

Standard Deviation Error 211

Standard Report 67
Statistics Tab 65
Step by Step Configuration 178
Stepwise 190
Stepwise Backward Repression 191
Stepwise Forward Regression 190
Stepwise Forward with a Backward Look
Regression 191
Sum of Squared Deviation 216
Sum Square Error 206
System Administrator 11

T
Theils 211
Tracking Table 99
Transform 56
Trend (Linear) Regression 195

Rev 0 219
johngalt.com

World Headquarters
John Galt Solutions
Chicago, IL 60602

Rev 0