Social Network Ads

Madhurish Gupta

1. Introduction

Social network advertising, also social media targeting, is a group of terms that are used to describe forms of online advertising that focus on social networking services. One of the major benefits of this type of advertising is that advertisers can take advantage of the users’ demographic information and target their ads appropriately.Advantages are advertisers can reach users who are interested in their products, allows for detailed analysis and reporting, information gathered is real, not from statistical projections, does not access IP-addresses of the users.

2. Overview of the Study

Our field study concerns the dependence of purchase of product on gender , age , estimated salary of a person with given userID.

3. An empirical field study of Social Network ads

3.1 Overview

The specific objective of this Study was to investigate the advertising strategy employed by company to which group of people they must advertise more. Our goal was to compare purchasing of the product by person based on sex , age and estimated salary.

Accordingly, we construct the following hypothesis:

Hypothesis H1: The number of males who purchased the product are more than number of females who purchased the product

3.2 Data

For this study, we collected data from trusted website.

Data contains 5 columns.

1. UserID – Each person has a unique ID from which we can identify the person uniquely.

2. Gender – Person can male or female. This field is very important for our hypothesis.

3. Age – Age of the person. Because our product can be useful to some ages only.

4. EstimatedSalary – This column contains salary of a person as salary can affect the shopping of a person.

5. Purchased – Contains two numbers ‘0’ or ‘1’. ‘0’ means not purchased and ‘1’ means purchased.This variable is our dependent variable.

3.3 Model

In order to test Hypothesis , we proposed the following model:

\[Purchased = \alpha_0 + \alpha_1 Gender + \alpha_2 Age + \alpha_3 EstimatedSalary + \epsilon\]

# Read the data
dataset <- read.csv('Social_Network_Ads.csv')
attach(dataset)
# OLS Model
# Replace the 'sex' columns as follows: 1 = Male, 2 = Female
# Convert them both into factors
dataset$Gender[dataset$Gender == 1] <- 'Male'
dataset$Gender[dataset$Gender == 2] <- 'Female'
dataset$Gender <- factor(dataset$Gender)
linearMod <- lm(Purchased ~ Gender + Age + EstimatedSalary, data=dataset)
summary(linearMod)

## 
## Call:
## lm(formula = Purchased ~ Gender + Age + EstimatedSalary, data = dataset)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.00533 -0.28014 -0.03037  0.25799  0.86629 
## 
## Coefficients:
##                   Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     -9.352e-01  6.788e-02 -13.778  < 2e-16 ***
## GenderMale       3.232e-02  3.166e-02   1.021    0.308    
## Age              2.850e-02  1.530e-03  18.627  < 2e-16 ***
## EstimatedSalary  3.118e-06  4.622e-07   6.746 4.12e-11 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.3563 on 511 degrees of freedom
## Multiple R-squared:  0.4565, Adjusted R-squared:  0.4533 
## F-statistic: 143.1 on 3 and 511 DF,  p-value: < 2.2e-16

We established the effect of age , gender , estimated salary on the purchase of a product with the model. We regressed Purchase on Age , Gender , EstimatedSalary. We estimated model, using linear least squares.

3.4 Results

We found empirical support for Hypothesis. The number of males who purchased the product are more than number of females who purchased the product.

4. Conclusion

This paper was motivated by the need for research that could improve our understanding of how social advertising can affect the purchase of product based on age, gender and estimated salary of a person. The unique contribution of this paper is that we investigated the number of males and females who purchased the product through social network advertisements. We observed the number of males are larger than females who purchased the product.

Appendix 1

Descriptive statistics

# Summarize the Data
library(psych)
describe(dataset)

                vars   n        mean       sd   median     trimmed
User.ID            1 515 15689466.94 71282.69 15692819 15688984.03
Gender*            2 515        1.50     0.50        1        1.50
Age                3 515       37.60    10.38       37       37.36
EstimatedSalary    4 515    68100.97 34416.90    65000    65464.89
Purchased          5 515        0.37     0.48        0        0.33
                     mad      min      max  range skew kurtosis      se
User.ID         91417.12 15566689 15815236 248547 0.01    -1.18 3141.09
Gender*             0.00        1        2      1 0.01    -2.00    0.02
Age                11.86       18       60     42 0.17    -0.68    0.46
EstimatedSalary 32617.20    15000   150000 135000 0.54    -0.42 1516.59
Purchased           0.00        0        1      1 0.56    -1.69    0.02

One Way Contingency Table

mytable <- xtabs(~ Gender+Purchased, data=dataset)
mytable

##         Purchased
## Gender     0   1
##   Female 159 100
##   Male   168  88

BoxPlot

boxplot(Purchased ~ EstimatedSalary , data=dataset)

boxplot(Purchased ~ Age , data=dataset)

Histograms

a <- dataset$Purchased
hist( a , data = dataset,
           main = "Distrution of Purcahsed", xlab="Purchased or Not", col='grey' )

## Warning in plot.window(xlim, ylim, "", ...): "data" is not a graphical
## parameter

## Warning in title(main = main, sub = sub, xlab = xlab, ylab = ylab, ...):
## "data" is not a graphical parameter

## Warning in axis(1, ...): "data" is not a graphical parameter

## Warning in axis(2, ...): "data" is not a graphical parameter

b <- dataset$Age
hist( b , data = dataset, main = "Distrution of Age", xlab="Different Ages", col='blue' )

## Warning in plot.window(xlim, ylim, "", ...): "data" is not a graphical
## parameter

## Warning in title(main = main, sub = sub, xlab = xlab, ylab = ylab, ...):
## "data" is not a graphical parameter

## Warning in axis(1, ...): "data" is not a graphical parameter

## Warning in axis(2, ...): "data" is not a graphical parameter

c <- dataset$EstimatedSalary
hist( c , data = dataset,
           main = "Distrution of EstimatedSalaries", xlab="Different Salaries ", col='green' )

## Warning in plot.window(xlim, ylim, "", ...): "data" is not a graphical
## parameter

## Warning in title(main = main, sub = sub, xlab = xlab, ylab = ylab, ...):
## "data" is not a graphical parameter

## Warning in axis(1, ...): "data" is not a graphical parameter

## Warning in axis(2, ...): "data" is not a graphical parameter

Correlation matrix

library(corrgram)

## Warning: package 'corrgram' was built under R version 3.4.3

corrgram(dataset, order=TRUE, lower.panel=panel.shade,
         upper.panel=panel.pie, text.panel=panel.txt,
         main="MBA Starting Salaries")

Correlation matrix using corrgram

library(corpcor)
library(tseries)

## Warning: package 'tseries' was built under R version 3.4.3

data_mat <- as.matrix(dataset[,3:5])
covmat = cov(data_mat)
cov2cor(covmat)

##                       Age EstimatedSalary Purchased
## Age             1.0000000       0.1241452 0.6387109
## EstimatedSalary 0.1241452       1.0000000 0.2954804
## Purchased       0.6387109       0.2954804 1.0000000

ScatterPlot Matrix

library(car)

## Warning: package 'car' was built under R version 3.4.3

## 
## Attaching package: 'car'

## The following object is masked from 'package:psych':
## 
##     logit

scatterplotMatrix( formula = ~ Gender   + Age   + EstimatedSalary   + Purchased , cex = 0.6 , data = dataset)