Augmentation of Customer’s Profile Dataset Using Genetic Algorithm

International Journal of Research and Scientific Innovation (IJRSI) | Volume IV, Issue VIS, June 2017 | ISSN 2321–2705
www.rsisinternational.org Page 33
Augmentation of Customer’s Profile Dataset Using
Genetic Algorithm
Nethravathi P.S1
, K.Karibasappa2
1
Department of Master of Computer Applications, Shree Devi Institute of Technology, Mangaluru, Karnataka, India
2
Department of Computer Science and Engineering, Dayanand Sagar College of Engineering, Bengaluru, Karnataka, India
Abstract: - Data is the lifeblood of all type of business. Clean,
accurate and complete data is the prerequisite for the decision-
making in business process. Data is one of the most valuable
assets for any organization. It is immensely important that the
business focus on the quality of their data as it can help in
increasing the business performance by improving efficiencies,
streamlining operations and consolidating data sources. Good
quality data helps to improve and simplify processes, eliminate
time-consuming rework and externally to enhance a user’s
experience, further translating it to significant financial and
operational benefits [1] [2]. All organizations/ businesses strive to
retain their existing customers and gain new ones. Accurate data
enables the business to improve the customer experience. Data
augmentation adds value to base data by enhancing information
derived from the existing source. Data augmentation can help
reduce the manual intervention required to develop meaningful
information and insight of business data, as well as significantly
enhance data quality. Hence the business can provide unique
customer experience and deliver above and beyond their
expectations. The Data Augmentation is immensely important as
it helps in improving the overall productivity of the business. It
is also important in making the most accurate and relevant
information available quickly for decision making.
This work focuses on augmentation of the customer
dataset using Genetic Algorithm(GA). These augmented data are
used for the purpose of customer behavioral analysis. The data
set consists of the different factors inherent in each situation of
the customer to understand the market strategy. This behavioral
data is used in the earlier work of analyzing the data [13]. It is
found that collecting a very large amount of such data manually
is a very cumbersome process. It is inferred from the earlier
work [13] that the more number of data may give accurate
result. Hence it is decided to enrich the dataset by using Genetic
Algorithm.
I. INTRODUCTION
n today’s competitive business environment it is much
tougher to understand the opinion of the customer towards
the purchase of a product. People are more mobile oriented
and better informed. The personalized, individualized, and
relevant information of the customers are required for
business intelligence appraisal. In the previous work [13] the
data is collected manually for the customer behaviour
analysis. The experiment summarizes that the purchase
behaviour of a person is purely related to his/her credentials
(e.g. Hobby). From the result and analysis, it is observed
that, with the huge dataset it is still possible to improve the
advocacy level of the customer. As the manual data
collection is tedious and time consuming, it is decided to
generate the data by data augmentation using Genetic
Algorithm.
Introduction to Genetic Algorithm (GA): GA is inspired by
the process of that belongs to the larger class of Evolutionary
Algorithm (EA). Genetic Algorithms are commonly used to
generate high-quality solutions to and by relying on bio-
inspired operators such as mutation, crossover and selection
[4].
Functionality of GAs: Three basic operators responsible for
GA are (a) selection, (b) crossover and (c) mutation [8].
Crossover performs combination of different
solutions to ensure that the genetic information of a child life
is made up of the genes from each parent. Figure 1 Illustrates
the process of generating a dataset using GA. The reason
behind selecting GA for the augmentation of data is due to its
benefits [6]. (1) Generality and Versatility [6]: GA applied in
a wide variety of settings and can be easily moulded to
particular problems. (2) Robust and Online Problem Solving
[6]: The decisions will be made automatically in run-time to
cater to dynamic channel parameters indicating it is a faster
process. (3) Support for Global Optimization [6] GAs is
suited to ﬁnd the global optima due to a number of properties
• Search by means of a population of individuals.
• Work with an encoding of the multiple parameters.
• Use a ﬁtness function that does not require the
calculation of derivatives.
• Search probabilistically.
(4) GA is computationally simpler compared to other
complementary Artificial Intelligence techniques [6]. (5)
GAs use evolutionary techniques to test and improve the
solutions by using techniques such as mutation, crossover,
selection, and recombination [8]. The important benefit of
enhancing data in this paper are: (1) The data collection
efforts are reduced by enhancing datasets (2) Easy to generate
any number of data in future as the enhanced data is accurate.
(3) To achieve accurate results.
For the previous work [13] the data collected from
the respondent directly to understand the market strategy and
the different factors inherent in each situation of the
I

customers. It is inferred from the earlier work that the more
number of data may give accurate result. Hence it is decided
to enrich the dataset by using Genetic Algorithm.
Mehboob, Junaid Qadir, Salman Ali, and Athanasios
Vasilakos [6] provided a detailed survey of applications of
GA using different kinds of GA techniques in wireless
networking. They have also highlighted pitfalls and
challenges in successfully implementing GAs in wireless
networks and open issues of GA.
Moheb R. Girgis [14] presents an automatic test data
generation technique using Genetic Algorithm. The GA
technique presented in this paper is guided by the data flow
dependencies in the program to search for test data to fulfil
the all-uses criterion. The algorithm produces a set of test
cases, the set of def-use paths covered by each test case, and a
list of uncovered def-use paths. Experiments have been
carried out to evaluate the effectiveness of the proposed GA
compared to the random testing technique, and to compare the
proposed random selection method to the roulette wheel
method. The results of these experiments showed that the GA
technique outperformed the random testing technique in 12
out of the 15 programs used in the experiment. The
experiments also showed that the proposed selection method
produced better results than the roulette wheel method [15].
M. Anbarasi et. al. [16] attempt to predict the
presence of heart disease with reduced number of attributes
using Genetic Algorithm. The algorithm determines the
attribute contribute more towards the diagnosis of heart
ailments which indirectly reduces the number of tests which
are needed to be taken by a patient. Naive Bayes, clustering
classification and decision tree classifiers are used to predict
the diagnosis of patients. The accuracy is measured before
and after reduction of number of attributes. The
observations exhibit that the decision tree outperforms other
two data mining techniques after incorporating feature subset
selection with relatively high model construction time. Naïve
Bayes performs consistently before and after reduction of
attributes with the same model construction time.
Classification via clustering performs poor compared to other
two methods.
Amit Kumar Sharma [17] proposes a GA-based
software test data generator to demonstrate its feasibility. GAs
show good results in searching the input domain for the
required test sets. Genetic Algorithms may not be the answer
to the approach of software testing, but do provide an
effective strategy.
Silvia TRIF [18] demonstrates the use of genetic
algorithms for training neural networks used in secured
Business Intelligence Mobile Applications. He assesses the
use of genetic algorithm by the comparison between classic
back-propagation method and a genetic algorithm based
training. A comparative study is realized for determining the
better way of training neural networks, from the point of view
of time and memory usage. His study reveals that genetic
algorithms are a solution that can be used on mobile devices
to solve optimization problems like training a neural network.
The obtained solutions are good and the resources used to
obtain the solution are reasonable compared to classic training
methods.
Nidhi Bhatla and Kiran Jyoti [19] aims at analyzing
the various data mining techniques for heart disease
prediction. Various techniques and data mining classifiers are
defined for efficient and effective heart disease diagnosis. The
analysis shows that Neural Network with 15 attributes has
shown the highest accuracy compared to Decision Tree and
Genetic Algorithm.
II. METHODOLOGY
For this work the practical data was collected from various
public and private sectors from various regions to achieve the
uniformity and consistency in data. Following block diagram
Figure 1 illustrates this Methodology: This has got three
processes mentioned as follows
1. Data acquisition
2. Data cleansing
3. Data Transforming and moulding
These processes are explained in the next section
Figure 1: Schematic block diagram of the generation of a new data set using genetic algorithm.

Data Acquisition: The reason behind selecting behavioural for
the purpose of this work is to understand the market strategy
and the different factors inherent in each situation. These
factors play a major role on habits, profession and opinion of
the customers. Decisions and buying behaviour are obviously
also influenced by the characteristics of each customer. A
consumer does not buy the same product or service at 20 or 70
years. His lifestyle, values, environment, activities, hobbies,
age group and consumer habits evolve throughout his life.
The factors influencing the buying decision process may also
change. The lifestyle of a person will influence on his
behaviour and purchasing decisions. For example, a human
with a healthy and balanced lifestyle will prefer to eat organic
products and go to specific grocery stores, will do some
jogging regularly (and therefore will buy shoes, clothes and
specific products), etc. The occupation and economic situation
of a person also has significant impact on his buying
behaviour. For example, a marketing manager of an
organization will try to purchase business suits, whereas a
student tries to purchase books or stationary and a housewife
try to purchase household items. Hobbies reflect the inner
most desires of people, help them fulfil their needs. So, it is
obvious that all these factors influence the purchase pattern.
By identifying and understanding these factors, purchasing of
the product can be predicted for a new customer.
Based on the following data given in Table1 from
different customers, the whole analysis is prepared. Table 1
List the credentials of the customers used for the analysis of
the Business Intelligence. This study requires large set of live
data to obtain the accurate results. Since, online customer
websites like Amazon, Flipkart have large amount of such
data however they are not available for research and other
purposes due to confidentiality reason. Moreover, those data
may not have all the parameter that we are planning to capture
as mentioned above. This work takes the data from different
sources required for evaluation of the prediction. Since, the
type of data required depends purely on behavioural aspects
of the respondents, further proceedings of data collection were
done by collecting the live data from various public and
private sectors. The typical data set contains gender, age
group, hobbies, profession, product’s usage and opinion on
used products. The data collection was done from various
sources to obtain various categories like different age group,
different occupation, etc.
The data collected by the Google forms are shown in
the following table below.
Table 1. A Typical dataset generated by the Google form
Gender Occupa
tion
Age
gro up
Income
gro up
Hobby1 Hobby2
Product
category
Elec-
tronics
Satisfact-ion
Level
Advo-cacy
Level
Feeling
Male Engi
neer
40-49
yrs
5 Lakh
To
8 Lakh
Read
ing
Photogra
phy
Electronic Laptop
6-very
satisfied
3- Neutral Neutral
Female Doctor
20-29
yrs
2.5 L to
5 L
Reading Music Books Books
6-very
satisfied
4-Likely Happy
Male Athlete
30-39
yrs
2.5 L to
5 L
Sports
Adv
sports
Sports
Men
sports
cloth
5- satisfied 4-Likely Happy
Female House
maker
20-29
yrs
---
gardenin
g
Singing Household
Refriger
ator
7-Ex.
Satisfied
4-Likely Happy
Male Student
20-29
yrs
---
Arts &
crafts
Playing
with pet
household
Painting
kit
6-very
satisfied
5-Most
likely
Happy
The issues while collecting data manually by interacting with
the respondents are:
 Manual data collection consumed more time because
challenges faced by the respondents due to poor
knowledge of English.
 Time constrains to respond to the survey during
office hours. They were unreachable in their post
during office hours.
 Entry related access restrictions to certain offices;
need for prior permission in such cases.
 Issues related to geographical spread of people in
reaching them.
 Manual data collection includes inconsistencies.
Hence some percentage of data will be invalid and
become waste.
To avoid these problems the questionnaire is distributed
through Google forms. Only 20 percent of people responded
for the Google forms.
Hence this work is taken up to generate a large dataset for
further improvements and for the better and accurate results.
As this work required more number of dataset, GA is
proposed to enrich the dataset to improve the accuracy of the
result.
Data Cleansing: Data Cleansing is the process of identifying
and correcting inaccurate data from a data set. With reference
to customer profile data, data cleansing is the process of
maintaining consistent and accurate customer data through
identification and removal of incorrect, incomplete, out-of-
date data. Data cleansing help in the creation of a clean
customer datasets which offers multiple benefits across

functions and serves as a critical factor in the growth of
business.
Real-world datasets are highly susceptible to missing
and inconsistent data, lacking certain attributes of interest.
Low-quality data or un-cleansed will lead to low-quality
results. As this paper collected the real world data, which has
a mixture of raw data with the datasets. This has to be filtered
by manually as well as by machine in order to improve the
quality of the data. This stage includes filling missing values,
identify or remove outliers, and resolve inconsistencies.
Missing values are filled and resolved the inconsistencies
from the original dataset. It has been rejected records of these
kinds as the information is very less and inconsistent for the
processing.
The purpose of this work is to explore the views,
experiences of customers with different hobbies, professions,
gender etc., on specific products which they have used. For
this study, sufficient large number of datasets with different
verities are collected. The data is collected by distributing the
questionnaire to the respondents directly and also collected
through the Google Forms from various locations. As this
work is related to behavioural data, more importance is given
to parameters such as hobby and profession. These
parameters play a vital role in purchase behaviour of a
customer.
III. FLOW OF A TYPICAL GENETIC ALGORITHM
Three basic operators responsible for GA are (a)
selection, (b) crossover & (c) mutation. Crossover performs
recombination of different solutions to ensure that the genetic
information of a child life is made up of the genes from each
parent. The Figure1 above illustrates the process of generating
a dataset using Genetic Algorithm.
As a first step the existing data sets are randomly
populated. Out of these n record sets, a record R1 is selected
and all the chromosomes of R1 that is P0, P1…Pt will be
copied to the New record Rng1. The record set Rng1, is taken
for further process of cross over and mutation. In this Rng1
one or a few chromosomes (Pi, Pj, Pl) are selected, for
crossover and mutation. This process of cross over is
explained in the next section.
Cross over:
In this work, crossover process uses random
operation to generate new record from two parent records. As
explained in the previous section after copying the record into
new record Rng1, one more record R2 is selected from n
record set and taken for crossover. Since it is preferable to
carry out the crossover for the chromosomes, randomly some
chromosomes Pi have been selected from the new record set
for the process of crossover. In the selected (Pi)sets of
chromosomes random bits are selected for the crossover.
These selected bits from the chromosomes (Pi’s) of Rng1 are
replaced from those of R2 chromosome.
The principle behind Genetic algorithm is that they
create and maintain a population of individuals represented by
chromosomes (essentially a character string analogous to the
chromosomes appearing in DNA). These chromosomes are
typically encoded solutions to a problem. The chromosomes
then undergo a process of evolution according to rules of
selection, mutation and crossover. Each individual in the
environment (represented by a chromosome) receives a
measure of its fitness in the environment. Reproduction
selects individuals with high fitness values in the population,
and through crossover and mutation of such individuals, a
new population is derived in which individuals may be even
better fitted to their environment. The process of crossover
involves two chromosomes swapping chunks of data.
Mutation introduces slight changes into a small proportion of
the population and is representative.
Example: A record Rng1 is selected which is having
chromosomes P0, P1…Pt. Let for example one chromosome
Pi is selected randomly that contains the bits as shown in the
Figure2 bellow. Similarly, same set of chromosomes is
selected from R2 as shown in Figure 3.
Y0 Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10
Figure 2. Bit patterns of pith
Chromosome of Rng1
X0 X1 X2 X3 X4 X5 X6 X7 X8 X9 X10
Figure 3. Bit patterns of pith
Chromosome of R2
In the above chromosome of R1 any random bits are selected,
for example 1st
4th
7th
and 10th
bits these bits are replaced by
the bits of the same place from the chromosome selected from
R2, which forms a new chromosome of category Pi as shown
in Figure 4.
Y0 X1 Y2 Y3 X4 Y5 Y6 X7 Y8 Y9 X10
Figure 4. Bit patterns of pi the newly generated Chromosome after crossover
This crossover chromosome is copied in to a new data set or
else the same chromosome is processed further for mutation.
The Mutation:
The above crossover chromosome shown in Figure 4
is further taken for the mutation. The process of mutation is as
shown in Figure 4. In this process of mutation, the most
significant bit is written as it is as Y0. Zeroth bit and first bit
are XORed the resultant is pleased in the first bit of the
resultant chromosome that is resultant of Y0 X1. First and
second bits are XORed to get the result as second bit, that is
X1 Y2 leads to the resultant of the third bit. Similarly, all
the bits are calculated as shown in the Figure 5, for the
process of mutation. The advantage of this method is there is
no over flow and data loss.

Figure 5. Example of Method of Mutation for a given datasets.
Example for Mutation :
This will be demonstrated as the new chromosome
after mutation of 11010110 is illustrated next. To maintain
the trace earlier chromosome, x1 will be written as it is in the
new chromosome. x1 is XORed with x2 written in the position
of x2. X2 is XORed with x3 written in the position of x3.
Similarly, it will continue till the end that is x8. The new
chromosome generated in this example is 10111101. After
this process of mutation, the resultant chromosome is
introduced in the data set and the dataset is tested for the
validation. The next section explains about the process of
validation.
Validation rules for the enhanced dataset
To ensure the resultant data set is in line with the
existing data set, which is nearer to the real life and
correctness of the dataset is validate during the association
rule. This validation procedure is applied to the enhanced data
set. Data validation is intended to provide certain well-defined
guarantees for fitness, accuracy, and consistency for any of
various kinds of user input. It also confirms that the following
rules that have been established for the applications to
validate data prior to sending updates the underlying database
of the data sets. The resultant data set is ‘rejected’ in case if it
satisfies any of the following association rules:
1. Gender=male and Occupation=Home maker.
2. Occupation=student and Age group between 30 to 39
or above 50 years
3. Occupation=student and Income group between 2.5
to 5 Lakhs or above 12 lakhs
4. Occupation=office assistant and Income group above
12 lakhs
5. Occupation=Home maker and Age group is under 19
6. Age group=under 19 and Occupation ≠ student.
7. Age group=above 50 years and
Hobby1=Adventurous sports.
8. Age group=above 50 years and
Hobby2=Adventurous sports
9. Age group= under 19 and Occupation other than
student
10. Satisfaction level=Extremely dissatisfied and
advocacy level=Likely/ Most Likely.
11. Satisfaction level=Extremely satisfied and advocacy
level=Unlikely/ Most Unlikely.
12. Satisfaction level=very Dissatisfied and advocacy
level=Likely or Most Likely.
13. Satisfaction level=very satisfied and advocacy
level=Unlikely/Most unlikely
14. Satisfaction level=Satisfied and advocacy
level=Most Unlikely
15. Satisfaction level=Neutral and advocacy level=Most
likely
16. Satisfaction level=Dissatisfied and advocacy
level=Most likely
IV. RESULTS AND ANALYSIS
Every new record is generated by picking up two
random records from the existing dataset. The hobby
parameters of these two records along with age group,
occupation, income group and satisfaction are crossed over
correspondingly to get the new record. Random 3, 4, 5, 6, and
7 bits are used for crossing over of parameters. The newly
generated record is further mutated to modify its
characteristics as per the genetic algorithm procedure. This
final new record is vetted through a validation procedure to
validate its authenticity to make it look like a real survey
response. The validation procedure involves checking the new
record parameters for their valid values as well as checking
their valid combination. For example, a record with age group
under 19 years with a high-income group is considered
invalid. Similarly, records with conflicting values for
Satisfaction and Advocacy, for e.g. Very Dissatisfied with the
product but Most Likely Advocating it to others.
The Figure 6 below shows the rejection rate of newly
generated records in the dataset of 5000 records. As can be
seen in the graph below in Figure 6, the rejection rate dips
when the crossover between two records happens with higher
number of bits (6 or 7) with a combination of parameters.
However, when the crossover is made using hobby and one
other parameter, the dip is highest in 5 bits.
It is obvious from the graph that for these attempted data the
drop is high. However, by the observation of the following

graph shown below in Figure 6, it is evident that in case of
multiple attribute crossovers the drop of the generated data is
very less compared to the data of the attribute alone.
Figure 6. Graph of Percentage drop for 5000 generated crossovers for
different attributes by mentioned number of bits and mutated data.
The negative peak in the graph shows the drop in the
generated genome. This is very much helpful to decide about
the attributes of the genome for the maximum generation
during the new generation. It is possible to find well in
advance the rejection and acceptance at the earlier stage, so as
to improve the rejection rate.
In the second case, 5 bits are taken for the crossover
operation. The percentage drop versus number of bits are
plotted in the graph shown below in Figure 7. The
observation is made on the basis of the following Figure 7.
This graph takes percentage drop in the Percentage Drop axis
and number of bits in the case of number of bits taken for the
crossover axis. It is evident that the graph drops during the
selection of 5bits for the crossover. This indicates clearly that
Figure 7. Graph of Percentage drop for 5000 generated crossovers
for different attributes by mentioned number of bits and mutated
data.
for the better performance, it is better to take 5 bits for the
crossover. However, in this case the crossover is done for only
one parameter of the genome. When two or more parameters
are considered the result is shown as in the Figure 8 below.
Figure 8: Graph of number of bit crossover vs percentage of drop in the
generation for multiple parameters of the genome
In the Figure 8 it is obvious that more than 5 bits gives a good
value of percentage drop. Considering the Figure 2 and Figure
4 it is apparent that more than 5 bits of crossover and mutation
will leads to a good combination for the generation of the new
genome.
V. CONCLUSION
It is observed that, it is possible to generate a valid
new record by picking up two random records from the
existing dataset. It is also observed from the result and
analysis that in case of multiple attribute crossovers the drop
in the generated data is very less compared to, the crossover
when only one data of the attribute is considered. For the
dataset of 5000, the rejection rate dips with 6 or 7 bits of
crossover operation, and will get about 5% of datasets
accepted. However, from the result of multiple attribute
crossovers the drop of the generated data is very less
compared to the data of the attribute alone and more than 5
bits of crossover and mutation will lead to a good combination
for the generation of the new genome. However, it is also
possible to enhance this percentage by evaluation of the data
set for acceptance or rejection in advance.
REFERENCES
[1]. Erhard Rahm, Hong Hai Do, “Data Cleaning: Problems and
Current Approaches”, University of Leipzig,Germany.
http://dbs.uni-leipzig.de.
[2]. Heiko Müller, Johann-Christoph Freytag, “Problems, Methods,
and Challenges in Comprehensive DataCleansing”, Humboldt-
Universität zu Berlin zu Berlin, 10099 Berlin, Germany.
[3]. Grefenstette, J.J. and Baker J.E. How Genetic Algorithms Work:
A Critical Look at Implicit Parallelism. In Schaffer, J.D. (ed.),
92.00%
93.00%
94.00%
95.00%
96.00%
97.00%
98.00%
99.00%
100.00%
3 4 5 6 7
%drop
Number of bits crosover
Percentage drop for 5000 generated crossover for
differentattributes by mentioned number of bits
and mutated data
Occupatio
n
Age
Group
Income
Group
Satisfactio
n Level
Occ+Age
Group
0.96
0.97
0.98
0.99
1
1.01
1.02
3 4 5 6 7
%drop
number of bits taken for the crossover
Graph of number of bit crossover vs %ge of drop in the
generation for single parameter of the genome
occupation
agegroup
Income group
0.92
0.93
0.94
0.95
0.96
0.97
0.98
3 4 5 6 7
%drop
Number of bits crosseover
Graph of number of bit crossover vs %ge of drop
in the generation for multiple parameter of the
genome
Occ+Aga Gr
Occ + Age Gr. +
incom Gr + Satis

Proceedings of the Third International Conference on Genetic
Algorithms. Morgan Kaufmann, San Mateo, CA,1989, pp. 20–27.
[4]. Genetic algorithm [Online] Available:
http://en.wikipedia.org/wiki/Genetic_algorithm
[5]. D. E. Goldberg and J. H. Holland, “Genetic algorithms and
machine learning,” Machine learning, vol. 3, no. 2, 1988, pp. 95–
99.
[6]. Mehboob, Junaid Qadir, Salman Ali, and Athanasios Vasilakos,
“Genetic Algorithms in Wireless Networking: Techniques,
Applications, and Issues” arXiv :1411, CS.NI
[7]. Li, L Weinberg., Darden T.A. and Pedersen,L.G. (2001) Gene
selection for sample classiﬁcation based on gene expression data:
study of sensitivity to choice of parameters of the GA/KNN
method. Bioinformatics, 17, 2001, pp 1131–1142
[8]. Melanie Mitchell Santa Fe L. D. Davis, “Handbook of Genetic
Algorithms. New York: Van Nostrand Reinhold”, Santa Fe., 1991.
[9]. Shruthi Rathnakar, K. Rajeswari, Rose Jacob, “Prediction of Heart
Disease Using Genetic Algorithm For Selection of Optimal
Reduced Set of Attributes “International Journal of Advanced
Computational Engineering and Networking, Volume-1, Issue-2,
April 2013.
[10]. K. F. Man, Member, IEEE, K. S. Tang, and S. Kwong, “Genetic
Algorithms: Concepts and Applications”, IEEE Transactions on
Industrial electronics, VOL. 43, NO. 5, October 1996
[11]. Mitchell Melanie, “An Introduction to Genetic Algorithms A
Bradford Book “, The MIT Press Cambridge, Massachusetts,
England, Fifth printing, 1999.
[12]. Nuwan I. Senaratna, “Genetic Algorithms: The Crossover-
Mutation Debate”, A literature survey submitted in partial
fulfilment of the requirements for the Degree of Bachelor of
Computer Science(Special) of the University of Colombo, 2005.
[13]. Nethravathi P. S, K.Karibasappa, “Business Intelligence Appraisal
of the Customer Dataset Based on Weighted Correlation Index”,
International Journal of Emerging Technology and Research.
2016.
[14]. Moheb R. Girgis, “Automatic Test Data Generation for Data Flow
Testing Using a Genetic Algorithm”, Journal of Universal
Computer Science, vol. 11, no. 6 (2005), 898-915
[15]. D. E. Goldberg and J. H. Holland, “Genetic algorithms and
machine learning,” Machine learning, vol. 3, no. 2, 1988, pp. 95–
99.
[16]. M. Anbarasi et. al., “Enhanced Prediction of Heart Disease with
Feature Selection Method by using Genetic Algorithm”,
International Journal of Engineering Science and Technology,
Vol. 2(10), 2010, 5370-5376.
[17]. A.K. Sharma, “Text Book of Correlations and Regression”,
Discovery Publishing House, 2005.
[18]. Silvia TRIF, “Using Genetic Algorithms in Secured Business
Intelligence Mobile Applications”, Informatica Economica,
Volume 15, Romania.2011.
[19]. Nidhi Bhatla Kiran Jyoti , “International Journal of Engineering
Research & Technology IJERT Vol. 1 Issue 8, October 2012.

Augmentation of Customer’s Profile Dataset Using Genetic Algorithm

More Related Content

What's hot (20)

Similar to Augmentation of Customer’s Profile Dataset Using Genetic Algorithm (20)

More from RSIS International (20)

Recently uploaded (20)

Augmentation of Customer’s Profile Dataset Using Genetic Algorithm