4/27/23, 11:08 AM sales_data - Jupyter Notebook

Implement K-Means clustering.

In [1]: import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
#Importing the required libraries.

In [2]: from sklearn.cluster import KMeans, k_means #For clustering

from sklearn.decomposition import PCA #Linear Dimensionality reduction.

In [3]: df = pd.read_csv("sales_data_sample.csv") #Loading the dataset.

Preprocessing
In [4]: df.head()

Out[4]:
ORDERNUMBER QUANTITYORDERED PRICEEACH ORDERLINENUMBER SALES ORDERDATE STATUS QTR_ID MONTH_ID Y

2/24/2003
0 10107 30 95.70 2 2871.00 Shipped 1 2
0:00

1 10121 34 81.35 5 2765.90 5/7/2003 0:00 Shipped 2 5

2 10134 41 94.74 2 3884.34 7/1/2003 0:00 Shipped 3 7

8/25/2003
3 10145 45 83.26 6 3746.70 Shipped 3 8
0:00

10/10/2003
4 10159 49 100.00 14 5205.27 Shipped 4 10
0:00

5 rows × 25 columns

In [5]: df.shape

Out[5]: (2823, 25)

In [6]: df.describe()

Out[6]:
ORDERNUMBER QUANTITYORDERED PRICEEACH ORDERLINENUMBER SALES QTR_ID MONTH_ID YEAR_I

count 2823.000000 2823.000000 2823.000000 2823.000000 2823.000000 2823.000000 2823.000000 2823.0000

mean 10258.725115 35.092809 83.658544 6.466171 3553.889072 2.717676 7.092455 2003.8150

std 92.085478 9.741443 20.174277 4.225841 1841.865106 1.203878 3.656633 0.6996

min 10100.000000 6.000000 26.880000 1.000000 482.130000 1.000000 1.000000 2003.0000

25% 10180.000000 27.000000 68.860000 3.000000 2203.430000 2.000000 4.000000 2003.0000

50% 10262.000000 35.000000 95.700000 6.000000 3184.800000 3.000000 8.000000 2004.0000

75% 10333.500000 43.000000 100.000000 9.000000 4508.000000 4.000000 11.000000 2004.0000

max 10425.000000 97.000000 100.000000 18.000000 14082.800000 4.000000 12.000000 2005.0000

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 1/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [7]: df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2823 entries, 0 to 2822
Data columns (total 25 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 ORDERNUMBER 2823 non-null int64
1 QUANTITYORDERED 2823 non-null int64
2 PRICEEACH 2823 non-null float64
3 ORDERLINENUMBER 2823 non-null int64
4 SALES 2823 non-null float64
5 ORDERDATE 2823 non-null object
6 STATUS 2823 non-null object
7 QTR_ID 2823 non-null int64
8 MONTH_ID 2823 non-null int64
9 YEAR_ID 2823 non-null int64
10 PRODUCTLINE 2823 non-null object
11 MSRP 2823 non-null int64
12 PRODUCTCODE 2823 non-null object
13 CUSTOMERNAME 2823 non-null object
14 PHONE 2823 non-null object
15 ADDRESSLINE1 2823 non-null object
16 ADDRESSLINE2 302 non-null object
17 CITY 2823 non-null object
18 STATE 1337 non-null object
19 POSTALCODE 2747 non-null object
20 COUNTRY 2823 non-null object
21 TERRITORY 1749 non-null object
22 CONTACTLASTNAME 2823 non-null object
23 CONTACTFIRSTNAME 2823 non-null object
24 DEALSIZE 2823 non-null object
dtypes: float64(2), int64(7), object(16)
memory usage: 375.0+ KB

In [8]: df.isnull().sum()

Out[8]: ORDERNUMBER 0
QUANTITYORDERED 0
PRICEEACH 0
ORDERLINENUMBER 0
SALES 0
ORDERDATE 0
STATUS 0
QTR_ID 0
MONTH_ID 0
YEAR_ID 0
PRODUCTLINE 0
MSRP 0
PRODUCTCODE 0
CUSTOMERNAME 0
PHONE 0
ADDRESSLINE1 0
ADDRESSLINE2 2521
CITY 0
STATE 1486
POSTALCODE 76
COUNTRY 0
TERRITORY 1074
CONTACTLASTNAME 0
CONTACTFIRSTNAME 0
DEALSIZE 0
dtype: int64

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 2/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [9]: df.dtypes

Out[9]: ORDERNUMBER int64

QUANTITYORDERED int64
PRICEEACH float64
ORDERLINENUMBER int64
SALES float64
ORDERDATE object
STATUS object
QTR_ID int64
MONTH_ID int64
YEAR_ID int64
PRODUCTLINE object
MSRP int64
PRODUCTCODE object
CUSTOMERNAME object
PHONE object
ADDRESSLINE1 object
ADDRESSLINE2 object
CITY object
STATE object
POSTALCODE object
COUNTRY object
TERRITORY object
CONTACTLASTNAME object
CONTACTFIRSTNAME object
DEALSIZE object
dtype: object

In [10]: df_drop = ['ADDRESSLINE1', 'ADDRESSLINE2', 'STATUS','POSTALCODE', 'CITY', 'TERRITORY', 'PHONE', 'STATE',

df = df.drop(df_drop, axis=1) #Dropping the categorical uneccessary columns along with columns having null

In [11]: df.isnull().sum()

Out[11]: QUANTITYORDERED 0
PRICEEACH 0
ORDERLINENUMBER 0
SALES 0
ORDERDATE 0
QTR_ID 0
MONTH_ID 0
YEAR_ID 0
PRODUCTLINE 0
MSRP 0
PRODUCTCODE 0
COUNTRY 0
DEALSIZE 0
dtype: int64

In [12]: df.dtypes

Out[12]: QUANTITYORDERED int64

PRICEEACH float64
ORDERLINENUMBER int64
SALES float64
ORDERDATE object
QTR_ID int64
MONTH_ID int64
YEAR_ID int64
PRODUCTLINE object
MSRP int64
PRODUCTCODE object
COUNTRY object
DEALSIZE object
dtype: object

In [13]: # Checking the categorical columns.

In [14]: df['COUNTRY'].unique()

Out[14]: array(['USA', 'France', 'Norway', 'Australia', 'Finland', 'Austria', 'UK',

'Spain', 'Sweden', 'Singapore', 'Canada', 'Japan', 'Italy',
'Denmark', 'Belgium', 'Philippines', 'Germany', 'Switzerland',
'Ireland'], dtype=object)

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 3/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [15]: df['PRODUCTLINE'].unique()

Out[15]: array(['Motorcycles', 'Classic Cars', 'Trucks and Buses', 'Vintage Cars',

'Planes', 'Ships', 'Trains'], dtype=object)

In [16]: df['DEALSIZE'].unique()

Out[16]: array(['Small', 'Medium', 'Large'], dtype=object)

In [17]: productline = pd.get_dummies(df['PRODUCTLINE']) #Converting the categorical columns.

Dealsize = pd.get_dummies(df['DEALSIZE'])

In [18]: df = pd.concat([df,productline,Dealsize], axis = 1)

In [19]: df_drop = ['COUNTRY','PRODUCTLINE','DEALSIZE'] #Dropping Country too as there are alot of countries.
df = df.drop(df_drop, axis=1)

In [20]: df['PRODUCTCODE'] = pd.Categorical(df['PRODUCTCODE']).codes #Converting the datatype.

In [21]: df.drop('ORDERDATE', axis=1, inplace=True) #Dropping the Orderdate as Month is already included.

In [22]: df.dtypes #All the datatypes are converted into numeric

Out[22]: QUANTITYORDERED int64

PRICEEACH float64
ORDERLINENUMBER int64
SALES float64
QTR_ID int64
MONTH_ID int64
YEAR_ID int64
MSRP int64
PRODUCTCODE int8
Classic Cars uint8
Motorcycles uint8
Planes uint8
Ships uint8
Trains uint8
Trucks and Buses uint8
Vintage Cars uint8
Large uint8
Medium uint8
Small uint8
dtype: object

Plotting the Elbow Plot to determine the number of clusters.

In [23]: distortions = [] # Within Cluster Sum of Squares from the centroid
K = range(1,10)
for k in K:
kmeanModel = KMeans(n_clusters=k)
kmeanModel.fit(df)
distortions.append(kmeanModel.inertia_) #Appeding the intertia to the Distortions

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 4/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [24]: plt.figure(figsize=(16,8))
plt.plot(K, distortions, 'bx-')
plt.xlabel('k')
plt.ylabel('Distortion')
plt.title('The Elbow Method showing the optimal k')
plt.show()

As the number of k increases Inertia decreases.

Observations: A Elbow can be observed at 3 and after that the curve

decreases gradually.
In [25]: X_train = df.values #Returns a numpy array.

In [26]: X_train.shape

Out[26]: (2823, 19)

In [27]: model = KMeans(n_clusters=3,random_state=2) #Number of cluster = 3

model = model.fit(X_train) #Fitting the values to create a model.
predictions = model.predict(X_train) #Predicting the cluster values (0,1,or 2)

In [28]: unique,counts = np.unique(predictions,return_counts=True)

In [29]: counts = counts.reshape(1,3)

In [30]: counts_df = pd.DataFrame(counts,columns=['Cluster1','Cluster2','Cluster3'])

In [31]: counts_df.head()

Out[31]:
Cluster1 Cluster2 Cluster3

0 1083 1367 373

Visualization
In [32]: pca = PCA(n_components=2) #Converting all the features into 2 columns to make it easy to visualize using P

In [33]: reduced_X = pd.DataFrame(pca.fit_transform(X_train),columns=['PCA1','PCA2']) #Creating a DataFrame.

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 5/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [34]: reduced_X.head()

Out[34]:
PCA1 PCA2

0 -682.488323 -42.819535

1 -787.665502 -41.694991

2 330.732170 -26.481208

3 193.040232 -26.285766

4 1651.532874 -6.891196

In [35]: #Plotting the normal Scatter Plot

plt.figure(figsize=(14,10))
plt.scatter(reduced_X['PCA1'],reduced_X['PCA2'])

Out[35]: <matplotlib.collections.PathCollection at 0x24749910>

In [36]: model.cluster_centers_ #Finding the centriods. (3 Centriods in total. Each Array contains a centroids for

Out[36]: array([[ 3.72031394e+01, 9.52120960e+01, 6.44967682e+00,

4.13868425e+03, 2.72022161e+00, 7.09879963e+00,
2.00379409e+03, 1.13248384e+02, 5.04469067e+01,
3.74884580e-01, 1.15420129e-01, 9.41828255e-02,
8.21791320e-02, 1.84672207e-02, 1.16343490e-01,
1.98522622e-01, 2.08166817e-17, 1.00000000e+00,
-6.66133815e-16],
[ 3.08302853e+01, 7.00755230e+01, 6.67300658e+00,
2.12409474e+03, 2.71762985e+00, 7.09509876e+00,
2.00381127e+03, 7.84784199e+01, 6.24871982e+01,
2.64813460e-01, 1.21433797e-01, 1.29480614e-01,
1.00219459e-01, 3.87710315e-02, 9.21726408e-02,
2.53108998e-01, 2.08166817e-17, 6.21799561e-02,
9.37820044e-01],
[ 4.45871314e+01, 9.98931099e+01, 5.75603217e+00,
7.09596863e+03, 2.71045576e+00, 7.06434316e+00,
2.00389008e+03, 1.45823056e+02, 3.14959786e+01,
5.33512064e-01, 1.07238606e-01, 7.23860590e-02,
2.14477212e-02, 1.07238606e-02, 1.31367292e-01,
1.23324397e-01, 4.20911528e-01, 5.79088472e-01,
1.66533454e-16]])

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 6/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [37]: reduced_centers = pca.transform(model.cluster_centers_) #Transforming the centroids into 3 in x and y coor

In [38]: reduced_centers

Out[38]: array([[ 5.84994044e+02, -4.36786931e+00],

[-1.43005891e+03, 2.60041009e+00],
[ 3.54247180e+03, 3.15185487e+00]])

In [39]: plt.figure(figsize=(14,10))
plt.scatter(reduced_X['PCA1'],reduced_X['PCA2'])
plt.scatter(reduced_centers[:,0],reduced_centers[:,1],color='black',marker='x',s=300) #Plotting the centri

Out[39]: <matplotlib.collections.PathCollection at 0x53aa880>

In [40]: reduced_X['Clusters'] = predictions #Adding the Clusters to the reduced dataframe.

In [41]: reduced_X.head()

Out[41]:
PCA1 PCA2 Clusters

0 -682.488323 -42.819535 1

1 -787.665502 -41.694991 1

2 330.732170 -26.481208 0

3 193.040232 -26.285766 0

4 1651.532874 -6.891196 0

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 7/8
4/27/23, 11:08 AM sales_data - Jupyter Notebook

In [42]: #Plotting the clusters

plt.figure(figsize=(14,10))
# taking the cluster number and first column taking the same cluster number
plt.scatter(reduced_X[reduced_X['Clusters'] == 0].loc[:,'PCA1'],reduced_X[reduced_X['Clusters'] == 0].loc[
plt.scatter(reduced_X[reduced_X['Clusters'] == 1].loc[:,'PCA1'],reduced_X[reduced_X['Clusters'] == 1].loc[
plt.scatter(reduced_X[reduced_X['Clusters'] == 2].loc[:,'PCA1'],reduced_X[reduced_X['Clusters'] == 2].loc[
​
​
plt.scatter(reduced_centers[:,0],reduced_centers[:,1],color='black',marker='x',s=300)

Out[42]: <matplotlib.collections.PathCollection at 0x2430c448>

In [ ]: ​

In [ ]: ​

localhost:8888/notebooks/Downloads/sales_data.ipynb#Implement-K-Means-clustering/-hierarchical-clustering-on-sales_data_sample.csv-datase… 8/8