URKUND

Urkund Analysis Result

Analysed Document: 15_Priyanshu_Roul_SIRP.pdf (D76571759)
Submitted: 7/19/2020 9:15:00 PM
Submitted By: [email protected]
Significance: 5%

Sources included in the report:

https://www.researchgate.net/ publication/220219695_Measuring_advertisement_effectiveness_-
_A_neural_network_approach https://github.com/ROVAN1220/Predict-an-Ad-s-Success-
Hackerearth-Machine-Learning- Challenge

Instances where selected sources appear:

3
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

S. P. MANDALI’s PRIN. L. N.

WELINGKAR INSTITUTE OF MANAGEMENT DEVELOPMENT & RESEARCH SUMMER

INTERNSHIP RESEARCH PROJECT REPORT ON “

Factors affecting the success of an advertisement” BY MS. PRIYANSHU ROUL PGDM -

RESEARCH & BUSINESS ANALYTICS (2019 – 21) Specialization: Marketing Roll No: 15
PROJECT FACULTY GUIDE PROF. P.V. CHANDRIKA

ii LETTER OF AUTHORIZATION

CERTIFICATE TO WHOSOEVER IT MAY CONCERN 1. This is to certify that Ms. Priyanshu Roul,
a student of

S.P. Mandali’s Prin. L. N. Welingkar Institute of Management Development & Research and

pursuing two years full time Post Graduate Diploma in Management – (PGDM- RBA),
underwent two months of SIRP with me from 27/04/2020 to 20/07/2020. 2. During the SIRP, Ms.
Priyanshu Roul has successfully completed the project titled “

Factors affecting the success of an advertisement” under my

guidance of Prof. P.V. Chandrika, Assistant Professor. 3.

The students’ performance during the SIRP and comments on his projects are as under: ………
………………………………………………………….. (Signature of Internal faculty member) Name:
Designation: Date
iii

NO PLAGIARISM DECLARATION BY THE STUDENT I, the undersigned, hereby declare that the
project titled “Factors affecting the success of advertisement” a) Has been prepared by me
towards the partial fulfilment

for the award of Post- Graduation Diploma in Management –

Research & Business Analytics under the

guidance of Prof. P.V. Chandrika,

S.P. Mandali’s Prin. L.N. Welingkar Institute of Management Development & Research,

Mumbai. b) This work is original and has not been submitted for any degree/diploma in this or any
other Institute/Organization. c) The information furnished is this dissertation is genuine and original
to the best of my knowledge and belief. d) I have not indulged in plagiarism. The project report has
been checked plagiarism and output report has been attached.

2
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

Priyanshu Roul Student Name Signature Place: Mumbai

Date:

iv ACKNOWLEDGEMENT

The successful completion of this project was possible due to tremendous research, extensive work
and dedication towards the project. This wouldn’t have been possible without the constant support
received from peers & surroundings directly or indirectly. I would like to express a sincere thanking
note and gratitude to all of them for their support. I

firstly thank my parents for all the support and motivation they have shown towards me. Any project
is incomplete without the support of one’s family. I owe my deep gratitude to my mentor Prof. P.V.
Chandrika for mentoring me and guiding me. She has been my mentor for the project motivating me
with very useful Interpretations and suggestions despite the busy schedule.

I would also like to thank all the Research and Business Analytics faculty members for their guidance.

Also thanking the Dean of our department, Prof. C. Y. Nimkar for all the Interpretations and guidance.

Lastly, a thanking note to all my family members, friends, individuals around & mother nature for the
support. Thanking all of them,

Ms. Priyanshu Roul

v TABLE OF CONTENTS CHAPTER NO TITLE PAGE NO.

ACKNOWLEDGEMENT LIST OF TABLES LIST OF GRAPHS EXECUTIVE SUMMARY 1.

INTRODUCTION 1.1 Overview 1.2 Problem Statement 1 1 1 2. REVIEW OF LITERATURE 2 3.
OBJECTIVE OF STUDY 3 4. RESEARCH DESIGN 1.1 Data source and variable structure 1.2
Tools and technique 1.3 Limitations 3 3 3 3 5. DATA DESCRIPTION AND PREPARATION 1.1
Data set and variable considerations 4 6. DATA ANALYSIS 1.1 Data Interpretations and
Exploratory Analysis 1.2 Data Preparation 5 5 13 7. MODELING 1.1 Analysis and predictions 14 8.
RESULTS
AND CONCLUSION 19 9. REFERENCES 20

vi LIST OF TABLES TABLE NUMBER NAME PAGE NUMBER 5.1 Data Dictionary 4 7.1 Matrix on
training data 16 7.2 Matrix on testing data 16 7.3 Comparing matrices (Logistic Regression) 18

vii LIST OF GRAPHS GRAPH NUMBER NAME PAGE NUMBER 6.1 Country wise advertisement
air location 5 6.2 Correlation graph 6 6.3 Genre wise ratings graph 6 6.4 Relationship status wise
ratings graph 7 6.5 Expense wise average air time graph 7 6.6 Average run time of relationship
status 8 6.7 Average run time of relationship status and netgain 8 6.8 Industry count 9 6.9 Targeted
sex (gender) count with netgain 9 6.10 Genre count 9 6.11 Expense count
10 6.12 Genre count with netgain 10 6.13 Expensive wise netgain score 11 6.14 Average run
time graph 11 6.15 Average run time based on airtime 12 6.16 Average run time based on

3
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

industry 12 6.17 Rating wise expensive graph 13 7.1 Sigmoid Function 15 7.2 Confusion matrix 15
7.3 ROC-AUC Curve of logistic regression-training set 17 7.4 ROC-AUC Curve of logistic
regression-testing set 18

viii EXECUTIVE SUMMARY

The purpose of this study is to identify the factors that

are affecting the effectiveness of an advertisement and to investigate the effect of these features on
the success of the advertisement. A sample of 26000 customers used to study the identified
features and its effect on the success of the advertisement. The factors which are included for the
study are customer’s relationship status, advertisement air location, advertising industry, genre,
advertising airtime, customer’s gender, average run time, ratings of the advertisement and the
expense of the advertisement. Exploratory data analysis is performed for the initial investigations on
data to discover the patterns also to spot the anomalies and missing data. Logistic Regression is
used in order to test the success of the advertisement. Findings from the exploratory and predictive
analysis shows that effectiveness of individual features over the advertisement. The advertisements
made on pharmacy has more success than the others also it tells the most success advertisement
genre is comedy among the customers. The demographic details shows that United States has
number of advertisement aired which indicates the majority of the customers are from that location.
Predictive analysis has been done using classification models that predict the class membership.
Logistic regression is used as a classification model for the prediction of the success of the
advertisements. The data divided in to train and test data set in which for training the model train
data set is used while test data to predict the outcomes based on the trained model. Upon training
the data various parameter kept in to considerations. Logistic regression classifier given the
accuracy of all i.e. 80.2% which makes it an efficient algorithm to predict the outcomes. These
analysis and predictions will help the advertising

and
0: preferences
https://www.researchgate.net/
publication/220219695_Measuring_advertisement_effectiveness_-
_A_neural_network_approach 100%

organizations to design their promotional campaigns according to the characteristics

0: https://www.researchgate.net/
publication/220219695_Measuring_advertisement_effectiveness_-
_A_neural_network_approach 75%

of their target customers which will contribute to the success of their product
advertisements.

1 1.

4
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

INTRODUCTIONS 1.1 OVERVIEW: As a recent changes in people’s lifestyle and with the new
products introduction in the markets. The various brand companies are encouraged to advertise
their products and services to the customers for better reach. Advertising is one the most important
form of communication with the target customers which encourage them to purchase the products
and services provided by the companies. An essential part of part of advertising is the promotion of
brands as the companies are constantly involved creation and promotion of their brands, they are
constantly looking for the advertising agencies to promote their products in the market. As with the
technological advancement now the companies have the data of their rival companies which gives
them both competitive advancement as well the threat from the others. Therefore the companies
needs to have better brands image in order to sustain in the market. It is often difficult to attract the
customer to a brand as they try to ignore them, sometimes the advertising message is not
understandable for the target customers as a result requires lot of content and creativity in the
advertising message. India is major exporter of minerals and refined petroleum around the world,
Indian firms have created many brands and advertised it international and domestic markets. 1.2
PROBLEM STATEMENT: A research study on the various factors which affecting the success of
an advertisement. An Advertisement has many factors which can influence it success in the
markets factors like demographics, targeted customers, demands, expense etc. In this study all the
factors will be considered and its interdependency on each other with the application of predictive
models to predict the future outcomes.

2 2. LITERATURE REVIEW To accomplish the objectives at the lower level is easier than at the top
of it, the prospective of the customer also decline as the objective increases, Burton (2011). The
advertiser must generate the interest among the target customers to lead them to the final step of
purchasing the goods and services. Advertising requires series of steps to move through from
unawareness to the purchasing of goods. Creativity plays an important role in the success of an
advertisement, it is a primary criteria for creating an advertisement. A creative and entertaining ad is
more likely to be remembered by consumers. That is why many firms are investing a lot of money to
make advertisements creative and humorous Mandan, Hossein & Furuzandeh (2013). Studies
shows that creativity and entertainment can generate customer satisfaction. A research by Duncan
and Nelson (1985) where out of 157 sample, 152 participants voted entertainment has a positive
influence on customers. Familiarity with the goods and services related experience gathered by the
customers. Familiarity of the items or the services can be direct or indirect through a salesperson or
advertisement over the internet. Zajonc & Markus (1982) suggest that familiarity with the items may
lead to creation of positive impression on customer. When confronted with a familiar services the
consumer may feel a glow of warmth because of its familiarity. Demographic influence on
advertisement to be precise online advertisement where the data generated over the internet can
help the advertiser to study the customer behavioral pattern to endorse their product as per the
targeted customer Vaclav Stritesky (2016). Spending is another factor which influence the buying
behavior of the customers as more spending means quality and more frequency advertisement
which can have lasting impression on the customers. Three main characteristics which an
advertisement should have is meaningful, convincing and different.
Meaningful means they should point out the benefits which makes it desirable. The product must
be convincing enough for the customers to purchase it and lastly different, some

5
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

distinguishable features must be there in the products which are different from the
competitors for the survival in the market place.

3 3. OBJECTIVE OF THE STUDY The objective of this study is to understand the relationship
between the different features or factors affecting the success of advertisements and to predict the
success of an advertisement based on the multiple features. This will help in analyzing and in the
selection of features for the success of the advertisements in the future.
4. RESEARCH DESIGN 4.1 DATA SOURCES AND VARIABLE STRUCTURE ? I have used the
data set from the kaggle: Machine learning and data science community. ? The dataset consist of
data points of the features associated with the advertisement and data if an advertisement will be
success or not which is in the column netgain. 4.2 TOOLS AND TECHNIQUES ? I have conducted
exploratory and predictive analysis using python. ? Data visualization is done on python and
tableau. Following steps were executed after the data extracted: ? Data integration in python- This
included basis imports of data and libraries and exploratory of the data. ? Data Profiling and
treatment in python-This included outlier treatment, missing values, and removing redundant
variables. 4.3 LIMITATIONS ? United States has the maximum number of advertisements aired in
the location which made the results demographically biased. ? Number of features for the prediction
were less to get more accurate results more number of features are required.

4 5. DATA DESCRIPTION & PREPARATION 5.1 DATA SET AND VARIABLE CONSIDERATION I
have
worked on a single dataset that is Advertisement success dataset which includes the products,
advertising services and target customers details. It also includes the information regarding the net
gain which indicates whether the advertisement gain the profit. 5.2 DATA DICTIONARY SR.NO
VARIABLE

0: https://github.com/ROVAN1220/Predict-an-Ad-s-Success-Hackerearth-Machine-Learning-
Challenge 97%

DESCRIPTION 1. id Unique id for each row 2. Ratings Metric out of 1 which represents how much
of the targeted demographic watched the advertisement 3. Air location Country of origin 4. Airtime
Time when the advertisement was aired 5. Average_runtime (minutes_per_week) Minutes per
week the advertisement was aired 6. targeted_sex Sex that was mainly targeted for the
advertisement 7. genre The type of advertisement 8. industry The industry to which the product
belonged
Table 9. Economic_status The economic health during which the show aired 10.
No. 5.1
Relationship_status The relationship status of the most responsive customers to the
5advertisement
6. DATA ANALYSIS 6.1 DATA
11. Expensive INTERPRETATIONS
A general AND
measure of how EXPLORATORY
expensive ANALYSIS
the product or serviceInisthis
that
category we will look at some of the visuals and graphs which give some Interpretations about
the ad is discussing 12. Money_back_guarantee Whether or not the product offers a refund in the
case of customer dissatisfaction 13 Netgain Whether the ad will incur a gain or loss when sold

6
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

the demographics and nature of the customer as per the parameters/variable defined above.
Following figure shows the country wise count of advertisement location refer fig.6.1 Fig. 6.1
Interpretations: Fig. 6.2 shows the correlation among the features, the correlation is from low to no
correlation. Fig. 6.3 show the ratings of advertisings based on different genre. It is further divided
into netgain value 0 (nonprofit) and 1 (profit) where the ratings are higher for the profited
advertisements. Among all the genre “Drama” got the highest ratings.

6 Fig. 6.2 Fig. 6.3

7 Fig. 6.4 Interpretations: Fig. 6.4 present the ratings based on the relationship status of the
customers, which further divided on netgain value 0 (nonprofit) and 1 (profit). Fig. 6.5

8 Fig. 6.6 Fig. 6.7 Interpretations: Fig.6.5 shows the three different graphs of average run time v/s
ratings, netgain value 1 has more constant rating than at value 0 throughout the expenses. Fig. 6.6
shows average run time based on the relationship status and in fig. 6.7 the marital status is further
divided in to netgain value at 1 and 0.

9 Fig. 6.8 Fig. 6.9 Fig. 6.10 Interpretations: From fig. 6.8 it can be seen that pharmacy industry
provided the most profit to the advertisers whereas the advertisement related to auto provided
least profit. Fig. 6.9 shows the gender wise division of customers where the male participants are
more than female. Fig. 6.10 shows the most number of adverting was of comedy genre.

10 Fig. 6.11 Fig. 6.12 Interpretations: Fig. 6.11 genre count based on the expenses where
majority of the advertisement are of low expense, further genre comedy have the highest
number among all in the genre category. Fig. 6.12 shows the genre wise count of
advertisement which is further subdivided in to net gain value at 1 and 0.

11 Fig. 6.13 Fig. 6.14

12 Interpretations: Fig. 6.13 shows the expense division based on netgain (0 and 1) where low
expense incurred most no profit from the advertisements. Fig.6.14 shows the average run time of
advertisements that is 40 minutes per week. Fig. 6.15 Fig. 6.16 Interpretations: Fig.6.15 show a
violin plot based on the airtime and average airtime, primetime advertisement has more in number
than others. Fig.6.16 shows the average run time based on the genre of advertisement where
pharmacy ads most run time.

13 Fig. 6.17 Interpretations: Fig.6.17 show a violin plot where all expense have same average
ratings irrespective of the category. 6.2 DATA PREPARATION 6.2.1 Variable Transformation In the
data set which consist details of customer and advertisement has many absolute numeric as well as
float variable. Those variables were transformed into Range variable i.e. those variables were
partitioned into data ranges. In line with the objective of the data set the target variable was coded 0
and 1. 0 being the advertisement did not provide the profit and 1 being coded as to provide the
profit. 6.2.2 Missing Values and Outliers The variables had missing values. To negate the effect of
minimum and maximum value of and that of outliers the missing value were imputed by median
instead of mode.

7
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

14 7. MODELING 7.1 ANALYSIS AND PREDICTIONS I have used a single dataset to develop
models for the objectives. As already mentioned the dataset consist of details of customers,
services, demography of the advertisement. For analysis I have taken the advertisement providing
the profit (Valued as 1) and the advertisement which do not provide profit (Valued as 0) The data is
divided into two partitions as mentioned below Type of Data File Name Used in Coding Percentage
Breakup Development Data Train 70% Holdout Data Test 30% Logical Regression Logistic
regression is a statistical model that uses a logistic function to model a binary dependent variable,
In regression analysis, logistic regression (or logit regression) is calculating the parameters of a
logistic model (a form of binary regression).A binary logistic model has a dependent variable with
two possible values, such as Yes/No which is represented by an indicator variable, where the two
values are labeled "0" and "1". In the logistic model, the log-odds for the value labeled "1" is a
linear combination of one or more independent variables or values ("predictors"); the independent
variables or values can each be a binary variable or a continuous variable. The corresponding
probability of the value labeled "1" can range between 0 and 1. Formula to calculate logistic
regression: Y= e (b0+b1x)
/1+e (b0+b1x) Where b0 is the slope of intercept and b1 is b1 is the coefficient of x variable. Y is
dependent (target) variable and x is independent variable. Sigmoid Function: In Logistic regression
it uses sigmoid function, it takes any real input and outputs whose value is between 0 and 1. In
sigmoid function x axis consist of independent variables and y axis consist of dependent variable.
Trained logistic regression model will produce the probability that the inputs belong to the class [0,
1].

15 Fig. 7.1 Confusion Matrix A confusion matrix is used to present the performance of a classifier
that is classification model. It is applied on the test data set for the values are known. Fig. 7.2 TN-
True Negative FP- False Positive FN- False Negative TN-True Negative

16 Matrix on training data Table No. 7.1 True Negatives: 12970 True Positives: 1640 False
Negatives: 2743 False Positives: 879 Sensitivity: 0.374 Specificity: 0.936 Matrix on test data
Predict Class Target 0 1 0 5267 375 1 1167 645 Table No. 7.2 True Negatives: 5267 True
Positives: 645 False Negatives: 1167 False Positives: 375 Sensitivity: 0.355 Specificity: 0.933
Predict Class Target 0 1 0 12970 879 1 2743 1640

17 ROC-AUC CURVE Roc-auc curve is a performance measurement for the classification

problems at different thresholds. It tells how perfectly the model can distinguish between the
different classes. Higher the value better the model in classification of 0 and 1.When we need to
check or visualize the performance of the classification problem, roc-auc curve is used.
Plotting ROC Curve and AUC Fig. 7.3

18 Fig. 7.4 Interpretations: Fig.7.1 and fig.7.2 show the ROC-AUC of train and test data for the
logistic regression algorithm at different thresholds, about 1% of difference can be seen which
indicates a good prediction. Comparing matrices (Logistic Regression) Matrices Train Test Accuracy
0.802 0.803 AUC 0.655 0.646 Precision 0.39 0.37 Recall 0.374 0.355 F1 Score 0.475
0.455 Table No. 7.3

19 8. RESULTS & CONCLUSIONS At the end of the modelling and analysis, I would like to point
the following: 1. From exploratory data analysis ? From the data set the advertisement aired in

8
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

40 countries out of which it can be seen as United States has the highest number of advertisement
aired followed by Mexico. ? The overall ratings of the advertisement which provided the profits to the
advertisers ranging from (0 to 1) where genre drama has the highest ratings. ? Relationship status
“Married but spouse absent” accounts for the maximum number of customer based on the ratings. ?
The rating for the advertisement provided profit is more and constant for the expenses (high, low
and medium). ? During the exploration it can be seen that the industry of advertisement provided the
most profit is pharmacy and the least is political same way auto industry has maximum number of
advertisement which does not provide any form of success. ? The advertisement made of low
expense has more number of counts than others and further if we see genre wise comedy has the
most number of advertisement under it. 2. From the predictive analysis ? Considering the accuracy
as the measuring parameter, Logistic regression algorithm provides the 80.2% accuracy. ?
Visualizing the model fit from ROC-AUC curve gives 66 % for training data set and 65% for test data
set.
Conclusion: Advertisement aims to influence the way customers views themselves and can change
the perspective towards the product by showing the how certain products can be beneficial to
them. The message conveyed through the advertisement influence the customer’s buying
decisions. The features mentioned above affecting the effectiveness of the advertisement which
leading to the success of an advertisement. With the predictive analysis of above features it also
give us an insight regarding the algorithm which can help in predicting the future prosperity of
advertisements. This will help the producers and competitors to boost their products in the
markets.

20 REFERENCES ? Ansari, Azarnoush. (2016).

An investigation of factors affecting brand advertising success and effectiveness.

International Business Research. 9. 10.5539/ibr.v9n4p20. ? Dawn Burton & Mary Klemm (2011)
Whiteness, ethnic minorities and advertising in travel brochures, The Service Industries Journal,
31:5, 679- 693, DOI: 10.1080/02642060902822083 ? Hossein, Safarzadeh & Furuzandeh, Askary.
(2013). Investigating the Impact of Advertising on Customers’ Behavioral Intentions A Case of
Agriculture Bank. Business and Economic Research. 3. 10.5296/ ber.v3i1.2799. ?

Robert B. Zajonc, Hazel Markus, Affective and Cognitive Factors in Preferences, Journal of
Consumer Research, Volume 9, Issue 2, September 1982, Pages 123–131, https://
doi.org/10.1086/208905 ?

Stříteský, Václav & Stritesky, Marek & Quigley, Martin & Říha, David. (2016). Significance of
Demographic Variables for Targeting of Internet Advertisements. Business, Management and
Education. 14. 1-18. 10.3846/bme.2016.277. ? https://www.kaggle.com/rohanchreddy/
advertsuccess ? Article- https://www.theatlantic.com/magazine/archive/1904/01/the- psychology-of-
advertising/303465/ ? Wikipedia

9
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

Hit and source - focused comparison, Side by Side:

Left side: As student entered the text in the submitted document.
Right side: As the text appears in the source.

_Measuring_advertisement_effectiveness_-
Instances from: https://www.researchgate.net/publication/2202196 95
_A_neural_network_approach

1 100% 1: https://www.researchgate.net/
publication/220219695_Measuring_advertisement_effectiveness_-
_A_neural_network_approach 100%
organizations to design their promotional campaigns according to
the characteristics
organizations to design their promotional campaigns according to
the characteristics

2 75% 2: https://www.researchgate.net/
publication/220219695_Measuring_advertisement_effectiveness_-
_A_neural_network_approach 75%
of their target customers which will contribute to the success of
their product advertisements.
of their target customers which will contribute to the cumulative
1 1.
effectiveness of their brand advertisements.

10
URKUND 15_Priyanshu_Roul_SIRP.pdf (D76571759)

Instances from: https://github.com/ROVAN1220/Predict-an-Ad-s-Suc ess-Hackerearth-Machine-Learning-Challenge

c

3 97% 3: https://github.com/ROVAN1220/Predict-an-Ad-s-Success-
Hackerearth-Machine-Learning-Challenge 97%
DESCRIPTION 1. id Unique id for each row 2. Ratings Metric out
of 1 which represents how much of the targeted demographic Description id :: Unique id for each row ratings :: Metric out of 1
watched the advertisement 3. Air location Country of origin 4. which represents how much of the targeted demographic watched
Airtime Time when the advertisement was aired 5. the advertisement airlocation :: Country of origin airtime :: Time
Average_runtime(minutes_per_week) Minutes per week the when the advertisement was aired
advertisement was aired 6. targeted_sex Sex that was mainly average_runtime(minutes_per_week) :: Minutes per week the
targeted for the advertisement 7. genre The type of advertisement 8. advertisement was aired targeted_sex :: Sex that was mainly
industry The industry to which the product belonged 9. targeted for the advertisement genre :: The type of advertisement
Economic_status The economic health during which the show aired industry :: The industry to which the product belonged
10. Relationship_status The relationship status of the most economic_status :: The economic health during which the show
responsive customers to the advertisement 11. aired relationship_status :: The relationship status of the most
Expensive A general measure of how expensive the product or responsive customers to the advertisement expensive
service is that the ad is discussing 12. Money_back_guarantee :: A general measure of how expensive the product or service is
Whether or not the product offers a refund in the case of customer that the ad is discussing. money_back_guarantee :: Whether or not
dissatisfaction 13 Netgain Whether the ad will incur a gain or loss the product offers a refund in the case of customer dissatisfaction.
when sold netgain [target] :: Whether the ad will incur a gain or loss when sold

11