Analyze the internet data of

www.datadb.com
The goals of this project are:

 To analyze each variable of the data collected through data summarization to get a
basic understanding of the dataset and to prepare for further analysis.

 To know whether the unique page view value depends on visits.

 Find out the probable factors from the dataset, which could affect the exits.

 Find the variables which possibly have an effect on the time on page.

 To determining the factors that are impacting the bounce.

# To analyze each variable of the data collected through data summarization to get a basic
understanding of the dataset and to prepare for further analysis.

library(readxl)
InternetDataset <- read_excel("C:/Users/Tanya/Desktop/simplilearn test/Project/Projects for
Submission/Internet/Internet/Internet_Dataset.xlsm")
View(InternetDataset)

str(InternetDataset)

#from this we can see structure InternetDataset, we see continent and source group are character string
#to convert character to factor
internet<-as.data.frame(unclass(InternetDataset))
str(internet)

# Now the characters are converted to factors

summary(internet)

#From the result of summarized dataset, it is observed that the numerical data includes
#information related to the maximum, minimum, and mean data.
#We can see that there is a maximum value of 30 bounces for the website.
#This site was accessed maximum number of times by visitors from N.America
##to know whether the unique page view value depends on visits.

cor(internet$Uniquepageviews,internet$Visits)
annova1<-aov(Uniquepageviews~Visits, data=internet)
summary(annova1)

#We can infer from the results that the visits variable has a significant impact on
#Unique.Pageviews. So the team can conclude that unique page values depend on visits.

#Find out the probable factors from the dataset, which could affect the exits.

annova2<-aov(Exits~.,data = internet)
summary(annova2)
#From the result of ANOVA given here, we can see that source.group, bounces,
#and unique.pageviews have more significance. Visits have comparatively less significance.
#Hence we can say that exit from the site is affected by the factors of source group,
#bounces, and unique.pageviews

#Find the variables which possibly have an effect on the time on page.

annova3<-aov(Timeinpage~.,data = internet)
summary(annova3)

# from the result of annova3 we can say except sourcegroup all affecting time on page.
# to determine the factors that are impacting the bounce
#this bounce rate is having variables,data for the variable bounces has to be between 0 and 1,

internet$Bounces=internet$Bounces*0.01
impactfactors<-glm(Bounces~Timeinpage+Continent+Exits+Sourcegroup+Uniquepageviews+Visits,data =
internet,family = "binomial")
summary(impactfactors)

#As can be inferred from the result shown, the BouncesNew, Unique.Pageviews and visits are the variables that
#impact the target variable bounces it has greater significance.
Codes
library(readxl)
InternetDataset <- read_excel("C:/Users/Tanya/Desktop/simplilearn test/Project/Projects for
Submission/Internet/Internet/Internet_Dataset.xlsm")
View(InternetDataset)
str(InternetDataset)
#from this we can see structure InternetDataset, we see continent and source group are character string
#to convert character to factor
internet<-as.data.frame(unclass(InternetDataset))
str(internet)
# Now the characters are converted to factors
#To analyze each variable of the data collected through data summarization
summary(internet)
#From the result of summarized dataset, it is observed that the numerical data includes
#information related to the maximum, minimum, and mean data.
#We can see that there is a maximum value of 30 bounces for the website.
#This site was accessed maximum number of times by visitors from N.America

##to know whether the unique page view value depends on visits.
cor(internet$Uniquepageviews,internet$Visits)
annova1<-aov(Uniquepageviews~Visits, data=internet)
summary(annova1)
#We can infer from the results that the visits variable has a significant impact on
#Unique.Pageviews. So the team can conclude that unique page values depend on visits.
#Find out the probable factors from the dataset, which could affect the exits.
annova2<-aov(Exits~.,data = internet)
summary(annova2)
#From the result of ANOVA given here, we can see that source.group, bounces,
#and unique.pageviews have more significance. Visits have comparatively less significance.
#Hence we can say that exit from the site is affected by the factors of source group,
#bounces, and unique.pageviews
#Find the variables which possibly have an effect on the time on page.
annova3<-aov(Timeinpage~.,data = internet)
summary(annova3)
# from the result of annova3 we can say except sourcegroup all affecting time on page
# to determine the factors that are impacting the bounce
#this bounce rate is having variables,data for the variable bounces has to be between 0 and 1,
internet$Bounces=internet$Bounces*0.01
impactfactors<-glm(Bounces~Timeinpage+Continent+Exits+Sourcegroup+Uniquepageviews+Visits,data =
internet,family = "binomial")
summary(impactfactors)
#As can be inferred from the result shown, the BouncesNew, Unique.Pageviews and visits are the variables that
#impact the target variable bounces it has greater significance.

Output of Codes

> library(readxl)
> InternetDataset <- read_excel("C:/Users/Tanya/Desktop/simplilearn test/Project/Projects for
Submission/Internet/Internet/Internet_Dataset.xlsm")
> View(InternetDataset)
> str(InternetDataset)
tibble [32,109 x 8] (S3: tbl_df/tbl/data.frame)
$ Bounces : num [1:32109] 0 0 0 0 0 0 0 0 0 0 ...
$ Exits : num [1:32109] 0 0 0 0 0 0 0 0 0 0 ...
$ Continent : chr [1:32109] "OC" "N.America" "N.America" "N.America" ...
$ Sourcegroup : chr [1:32109] "(direct)" "(direct)" "Others" "public.tableausoftware.com"
$ Timeinpage : num [1:32109] 18 4 35 70 81 75 186 710 712 344 ...
$ Uniquepageviews: num [1:32109] 1 1 1 1 1 1 1 1 1 1 ...
$ Visits : num [1:32109] 0 0 0 0 0 0 0 0 1 1 ...
$ BouncesNew : num [1:32109] 0 0 0 0 0 0 0 0 0 0 ...
> internet<-as.data.frame(unclass(InternetDataset))
> str(internet)
'data.frame': 32109 obs. of 8 variables:
$ Bounces : num 0 0 0 0 0 0 0 0 0 0 ...
$ Exits : num 0 0 0 0 0 0 0 0 0 0 ...
$ Continent : Factor w/ 6 levels "AF","AS","EU",..: 5 4 4 4 4 4 4 4 5 2 ...
$ Sourcegroup : Factor w/ 9 levels "(direct)","facebook",..: 1 1 4 5 5 5 5 1 1 4 ...
$ Timeinpage : num 18 4 35 70 81 75 186 710 712 344 ...
$ Uniquepageviews: num 1 1 1 1 1 1 1 1 1 1 ...
$ Visits : num 0 0 0 0 0 0 0 0 1 1 ...
$ BouncesNew : num 0 0 0 0 0 0 0 0 0 0 ...
> summary(internet)
Bounces Exits Continent Sourcegroup
Min. : 0.000 Min. : 0.000 AF : 321 google :11542
1st Qu.: 0.000 1st Qu.: 1.000 AS : 3171 (direct) : 7532
Median : 1.000 Median : 1.000 EU : 6470 Others : 5360
Mean : 0.713 Mean : 0.906 N.America:20043 tableausoftware.com : 2388
3rd Qu.: 1.000 3rd Qu.: 1.000 OC : 1356 t.co : 2249
Max. :30.000 Max. :36.000 SA : 748 public.tableausoftware.com: 1354
(Other) : 1684
Timeinpage Uniquepageviews Visits BouncesNew
Min. : 0.00 Min. : 1.000 Min. : 0.000 Min. :0.00000
1st Qu.: 0.00 1st Qu.: 1.000 1st Qu.: 1.000 1st Qu.:0.00000
Median : 0.00 Median : 1.000 Median : 1.000 Median :0.01000
Mean : 73.18 Mean : 1.114 Mean : 0.906 Mean :0.00713
3rd Qu.: 10.00 3rd Qu.: 1.000 3rd Qu.: 1.000 3rd Qu.:0.01000
Max. :46745.00 Max. :45.000 Max. :45.000 Max. :0.30000

> #From the result of summarized dataset, it is observed that the numerical data includes
> #information related to the maximum, minimum, and mean data.
> #We can see that there is a maximum value of 30 bounces for the website.
> #This site was accessed maximum number of times by visitors from N.America
> ##to know whether the unique page view value depends on visits.
> cor(internet$Uniquepageviews,internet$Visits)
[1] 0.8144457
> annova1<-aov(Uniquepageviews~Visits, data=internet)
> summary(annova1)
Df Sum Sq Mean Sq F value Pr(>F)
Visits 1 8052 8052 63257 <2e-16 ***
Residuals 32107 4087 0
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> #We can infer from the results that the visits variable has a significant impact on
> #Unique.Pageviews. So the team can conclude that unique page values depend on visits.
> #Find out the probable factors from the dataset, which could affect the exits.
> annova2<-aov(Exits~.,data = internet)
> summary(annova2)
Df Sum Sq Mean Sq F value Pr(>F)
Bounces 1 10578 10578 1.043e+05 < 2e-16 ***
Continent 5 3 1 5.960e+00 1.62e-05 ***
Sourcegroup 8 7 1 8.760e+00 4.89e-12 ***
Timeinpage 1 130 130 1.279e+03 < 2e-16 ***
Uniquepageviews 1 1573 1573 1.552e+04 < 2e-16 ***
Visits 1 1 1 5.014e+00 0.0251 *
Residuals 32091 3254 0
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> #From the result of ANOVA given here, we can see that source.group, bounces,
> #and unique.pageviews have more significance. Visits have comparatively less significance.
> #Hence we can say that exit from the site is affected by the factors of source group,
> #bounces, and unique.pageviews
> #Find the variables which possibly have an effect on the time on page.
> annova3<-aov(Timeinpage~.,data = internet)
> summary(annova3)
Df Sum Sq Mean Sq F value Pr(>F)
Bounces 1 5.947e+07 59466495 422.868 < 2e-16 ***
Exits 1 1.304e+08 130400662 927.283 < 2e-16 ***
Continent 5 4.767e+06 953431 6.780 2.51e-06 ***
Sourcegroup 8 1.545e+06 193153 1.374 0.202
Uniquepageviews 1 1.791e+08 179133934 1273.826 < 2e-16 ***
Visits 1 1.073e+08 107321113 763.163 < 2e-16 ***
Residuals 32091 4.513e+09 140627
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> internet$Bounces=internet$Bounces*0.01
> impactfactors<-glm(Bounces~Timeinpage+Continent+Exits+Sourcegroup+Uniquepageviews+Visits,da
internet,family = "binomial")
Warning messages:
1: In eval(family$initialize) : non-integer #successes in a binomial glm!
2: glm.fit: fitted probabilities numerically 0 or 1 occurred
> summary(impactfactors)

Call:
glm(formula = Bounces ~ Timeinpage + Continent + Exits + Sourcegroup +
Uniquepageviews + Visits, family = "binomial", data = internet)

Deviance Residuals:
Min 1Q Median 3Q Max
-2.26149 -0.02406 0.00206 0.00895 1.81288

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -4.9667681 0.6784678 -7.321 2.47e-13 ***
Timeinpage -0.0010294 0.0005774 -1.783 0.0746 .
ContinentAS 0.0022768 0.6932044 0.003 0.9974
ContinentEU -0.0069240 0.6786600 -0.010 0.9919
ContinentN.America 0.0101334 0.6674188 0.015 0.9879
ContinentOC 0.0201123 0.7333671 0.027 0.9781
ContinentSA 0.0237507 0.7914250 0.030 0.9761
Exits 1.3907608 0.3356504 4.143 3.42e-05 ***
Sourcegroupfacebook -0.0241949 1.1045171 -0.022 0.9825
Sourcegroupgoogle -0.0783631 0.1720157 -0.456 0.6487
SourcegroupOthers -0.0767919 0.2182692 -0.352 0.7250
Sourcegrouppublic.tableausoftware.com -0.2528285 0.4923123 -0.514 0.6076
Sourcegroupreddit.com -0.0092792 0.4709304 -0.020 0.9843
Sourcegroupt.co 0.0148690 0.2760157 0.054 0.9570
Sourcegrouptableausoftware.com -0.1129305 0.3190762 -0.354 0.7234
Sourcegroupvisualisingdata.com -0.0822525 0.4614866 -0.178 0.8585
Uniquepageviews -3.2363108 0.5791664 -5.588 2.30e-08 ***
Visits 2.1941121 0.5202216 4.218 2.47e-05 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

(Dispersion parameter for binomial family taken to be 1)

Null deviance: 234.937 on 32108 degrees of freedom

Number of Fisher Scoring iterations: 11

> #As can be inferred from the result shown, the BouncesNew, Unique.Pageviews and visits are
that
> #impact the target variable bounces it has greater significance.
>