International Journal of Innovative Research in Computer Science & Technology (IJIRCST)

ISSN: 2347-5552, Volume-2, Issue-1, January 2014

Analyzing Target Customer Behavior Using

Data Mining Techniques for E- Commerce Data
Dattatray V. Bhate , M. Yaseen Pasha

 At present, internet technologies are widely used by

Abstract— In this scenario Customer satisfaction is no longer customer for purchase different types of goods for those
satisfied with a simple listing of marketing contacts, but Internet technologies have seamlessly automated interface
wants detailed information about Customers past purchase as processes between customers and retailers, retailers and
well as prediction of future purchases. From long period of distributors, distributors and factories, and factories and
ago, information of customer is most important for every
their suppliers. E-commerce is changing the face of most
business. With the help of advance information technology,
firms are able to collect and store mountains of data
business functions in competitive enterprises. In the context
describing infinite offering and different type of customer of e-commerce, it is not that much of easy to generating
profile. With the help of this information we are able to find large-scale real-time data. For that reason there are
customer needs and wants. Electronic commerce or numerous opportunities for gathering customer
ecommerce is a term for any type of business, or commercial information in electronic form. By using Data Mining it
transaction that involves the transfer of information across generate non obvious yet useful information for decision
the Internet. It covers a range of different types of businesses, makers from very huge databases. The various mechanisms
from consumer based retail sites, through auction or music of this generation include abstractions, aggregations,
sites, to business exchanges trading goods and services
summarizations, and characterizations of data. These
between corporations. It is currently one of the most
important aspects of the Internet to emerge. Ecommerce
forms, in turn, are the result of applying sophisticated
allows consumers to electronically exchange goods and modeling techniques from the diverse fields of statistics,
services with no barriers of time or distance. Electronic database management, artificial intelligence, and computer
commerce has expanded rapidly over the past five years and graphics.
is predicted to continue at this rate, or even accelerate. Even after gathering consumer information, but still how
Traditional forecasting methods are no longer suitable for to analyze these data effectively is of interest to marketers
these business situations. For that type of business we are and researchers. The traditional methods for analyzing and
able to use the principles of data mining concept. With the predicting customer demands have found a wide range of
help of data mining principle huge amount of customer
applications. They are mainly used for predicting the total
information into cluster customer segments by using
K-Means algorithm which is used to cluster observations into
quantity of products that belong to the same family rather
groups of related observations without any prior knowledge than the relationship between the different customer groups
of those relationships and data from web log of various and associated product groups. By using K-Means
ecommerce websites. algorithm it generate clusters customer segments and data
from web log of various ecommerce websites. It is clear
Keywords: Data mining , K-Means algorithm, E-commerce that, the results showed that there was a clear distinction
between the segments in terms of customer behavior. It is
seen that this data mining model can used as an efficient
vehicle for firms not only to predict the products or services
I. INTRODUCTION
that should be provided or improved for their target
customer groups, but also to identify the right customers for
It may be observed that the revenue of any organization is a specific product family or service.
customer satisfaction. Every industry or organization tries
to increase the satisfaction of customer and also even tries
to increase the expectation of different types of customer. In II. LITERATURE REVIEW
addition to increase new customers, predicting potential
buyers, firms are devoting a great deal of resources to Transacting or facilitating business on the Internet is
delighting and retaining existing target customers to called ecommerce. Ecommerce is short for "electronic
cultivate a long-term, close relationship with them. commerce." Popular examples of ecommerce revolve
around buying and selling online. But the ecommerce
universe contains other types of activities as well. Any form
of business transaction conducted electronically is
Manuscript received January 23, 2014 ecommerce.
Dattatray V. Bhate, M.Tech CSE Student, Spoorthy Engineering College,
Examples of E commerce:
JNTU Hyderabad, Hyderabad, Andhra Pradesh, India (e-mail:
dattatraybhate12@gamil.com).
M. Yaseen Pasha, Assistant Professor, Department of CSE , Spoorthy
Engineering College, Hyderabad, Andhra Pradesh , India

16
 Analyzing Target Customer Behavior Using Data Mining Techniques for E- Commerce Data

A. Online shopping these companies to determine relationships among

"internal" factors such as price, product positioning, or staff
Buying and selling goods on the Internet is one of the most skills, and "external" factors such as economic indicators,
popular examples of ecommerce. Sellers create storefronts competition, and customer demographics. And, it enables
that are the online equivalents of retail outlets. Buyers them to determine the impact on sales, customer
browse and purchase products with mouse clicks. Though satisfaction, and corporate profits. Finally, it enables them
Amazon.com is not the pioneer of online shopping, it is to "drill down" into summary information to view detail
arguably the most famous online shopping destination. transactional data. The term ‘data mining’ is used to
describe the process of analyzing a company’s internal data
B. Electronic Payments
for customer profiling and targeting. In e commerce
When you are buying goods online, there needs to be a application, the end goal of data mining is to improve
mechanism to pay online too. That is where payment processes that contribute to delivering value to the end
processors and payment gateways come into the picture. customer.
Electronic payments reduce the inefficiency associated with At the basic level, the information available in web log
writing and mailing checks. It also does away with many of files can illuminate what prospective customers are seeking
the safety issues that arise due to payment made in currency from a site. Are they purposefully shopping or just
notes. browsing? Buying something they’re familiar with or
something they know little about? Are they shopping from
C. Online Auctions home, from work, or from a hotel dialup? The information
available in log files is often used to determine what
When you think online auction, you think eBay. Physical profiling can be dynamically processed in the background
auctions predate online auctions, but the Internet made and indexed into the dynamic generation of HTML, and
auctions accessible to a large number of buyers and sellers. what performance can be expected from the servers and
Online auctions are an efficient mechanism for price network to support customer service and make e-business
discovery. Many buyers find the auction shopping interaction productive.
mechanism much interesting than regular storefront E-commerce data are classified as usage data, content
shopping. data, structure data, and user data. Usage data provide
details information about user sessions and page views. The
D. Internet Banking
content data in a site are the collection of objects and
Today it is possible for you to perform the entire gamut of relationships that are conveyed to the user. Structure data
banking operations without visiting a physical bank branch. represent the designer’s view of the content organization
Interfacing of websites with bank accounts, and by within the site. Structure data also include the intra-page
extension credit cards, was the biggest driver of structure of the content represented in the arrangement of
ecommerce. HTML or XML tags within a page. The user data may
include demographic or other identifying information on
E. Online Ticketing registered users, user ratings on various objects such as
pages, products, or movies, past purchase or visit histories
Air tickets, movie tickets, train tickets, play tickets, tickets of users, as well as other explicit or implicit representations
to sporting events, and just about any kind of tickets can be of users’ interests. Once the data types are clear, data
booked online. Online ticketing does away with the need to preparation is easily achieved. The author then proposes
queue up at ticket counters. association rules, sequential and navigational patterns, and
clustering approaches for personalization of transactions as
E-commerce can change the face of business. It provides well as web pages. Liuying Shen and Jana Hawley describe
better customer management, an expanded range of an approach to predict user behavior in e-commerce sites.
product, new strategies for marketing, and more efficient The core of their approach involves extracting knowledge
operations. A key enabler of this change is the widespread from integrated data of purchase and path traversal patterns
use of increasingly sophisticated data mining tools. of past users (obtainable from web server logs) to predict
Generally, data mining (sometimes called data or the purchase and traversal behavior of future users.
knowledge discovery) is the process of analyzing data from Most of the algorithms presented in the literature to deal
different perspectives and summarizing it into useful with clustering web sessions treat sessions as sets of visited
information - information that can be used to increase pages within a time period and do not consider the sequence
revenue, costs, cuts or both. of the click-stream visitation. This has a significant
Data mining software is one of a number of analytical consequence when comparing similarities between web
tools for analyzing data. It allows users to analyze data from sessions. Quinlan, J.R proposes an algorithm based on
many different dimensions or angles, categorize it, and sequence alignment to measure similarities between web
summarize the relationships identified. Technically, data sessions where sessions are chronologically ordered
mining is the process of finding correlations or patterns sequences of page accesses. Nonetheless, reviews and
among dozens of fields in large relational databases. research in this area are handicapped by the proprietary
Primarily Data mining is technique used today by nature of the data and algorithms. A great deal of effort is
companies with a strong consumer focus - retail, financial, being expended in this area, but most of it is secret.
communication, and marketing organizations. It enables Certainly Amazon, Google, and Microsoft are deeply

17
 International Journal of Innovative Research in Computer Science & Technology (IJIRCST)
ISSN: 2347-5552, Volume-2, Issue-1, January 2014

engaged in statistical research, and in time the broader

research community may learn more about their findings. 4. Online Address
But for now, all this paper can really be done is to lay out the Address 1 - At work
main strategies in the relevant areas. Address 2 - At home

5. Language
III. DATA MINING MODEL Language1 - China
Language2 – English
A systematic method was used to collect e-commerce
transactions from web log file of e-commerce site. There 6. Target customer behavior
were 2518 transactions collected from 1st to 31st of Behavior 1 -Buying computer products
December 2013. Each session gave detail information about Behavior 2 -Buying cloth
web accessed from each user such as the web sites, Behavior 3 -Buying gifts and flower
requested web pages and their order, and the period of time Behavior 4 -Buying books
pages were viewed. This data is very important to analyze Behavior 5 -Buying CDs and DVDs
the customer. Behavior 6 -Buying toys and children product
Behavior 7 -Buying airline tickets and other tickets
Behavior 8 -Online trading
A. Preprocessing of Data Behavior 9 - Online banking
Data preprocessing techniques are used to increase the
quality of collected data, this can help to increase the
accuracy and efficiency of the subsequent mining processes. C. Result analysis
It is clear to see that data preprocessing is an important step K-Means algorithm used to transfer user data into
in the knowledge discovery process, as high-quality cluster. Results were segmented in five clusters. Based on
decisions must be based on high-quality data. Detecting the results of customer usage, e banking transactions can be
data anomalies, rectifying them early, and reducing the classified into 5 clusters.
data amount to be analyzed can result in substantial benefits
and advantages for the decision makers. For usage profiles, Cluster 1: The male customers who are 21 to 30 years old
a session file from data preparation stage was used as input used personal computers to purchase computer products
in data mining. While K-means algorithm was applied for and books. They accessed ecommerce site via personal
the purpose of clustering some incomplete data were computers between 6.00 hours to 17.59 hours at home. The
deleted. The usable data were 2363 transactions. language was Chinese.

B. Data Analysis Cluster 2: This was the second smallest cluster. The male
Data mining software, Mineset was used in data analysis. Customers were online between 6.00 hours to 17.59 hours.
The usable customer data of 4263 ecommerce transactions They accessed e-commerce site to buy books, airline tickets
were divided into two groups. Group 1 was about 70% of and other tickets at work. The language used was Chinese
the total transactions and was used as training data. Group and their age was from 26 to 35.
2 was about 30% of the total transactions and was used as
testing data. Five factors used in data segmentation Cluster 3: This cluster gained a majority of the ecommerce
included: age, gender, online in time, address, language, application since it had the largest population (26%). By
and target customer behavior type. using personal computers in Chinese language at home or
at work, the female and male Customers accessed e
1. Age was divided in 6 kinds commerce site to purchase books, CDs and DVDs, airline
Age 1 - 11 years old to 15 years tickets and other tickets. The online time was between
Age 2 - 16 years old to 20 years 12.00 hours to 23.59 hours and their age was from 21 to 30.
Age 3 - 21 years old to 25 years
Age 4 - 26 years old to 30 years Cluster 4: The male customers who are 11 to 20 years old
Age 5 - 31 years old to 35 years used personal computers at home. They accessed
Age 6 - 36 years old to 40 years e-commerce to purchase gifts (including Jewelry) and
flowers, toys and children products between 0.00 hours to
2. Gender was divided in 2 kinds 05.59 hours. The language was Chinese.
Gender 1 - man.
Gender 2 - woman. Cluster 5: This was the second largest cluster, where female
customers used personal computers for ecommerce between
3. How long online was divided in 4 groups 6.00 hours to 17.59 hours. The language used was Chinese
Group 1 -0.00 hours to 05.59 hours and transactions were bill payments. Their age was from 21
Group 2 -06.00 hours to 11.59 hours to 30 and they purchase cloth, toys and children products.
Group 3 -12.00 hours to 17.59 hours
Group 4 -18.00 hours to 23.59 hours

18
 Analyzing Target Customer Behavior Using Data Mining Techniques for E- Commerce Data

Cluster 6: Male customers who are 31 to 40 years old used

personal computers for e commerce site between 12.00
hours to 23.59 hours at work. This was the only cluster
where customers used English. The major behaviors were
online trade.

IV. CONCLUSION
Data mining is technique primarily used today by
companies with a strong Customer focus - retail,
communication, financial and marketing organizations.
Data mining is having lot of importance because of its huge
applicability. It is being used increasingly in business
applications for understanding and then predicting
valuable data, like Customer buying actions and buying
tendency, profiles of Customers, industry analysis, etc. Data
Mining is used in several applications like market research,
Customer behavior.
E-commerce companies are shifting from the old world
of mass production where standardized products,
homogeneous markets, and long product life and
development cycles were the rule to the new world where
variety and customization supplant standardized products.
Instead of tens of thousands of products in a superstore,
consumers may choose among millions of ones in an online
store to satisfy the personalization demands. It is clear that
target customers marketing can be effective when a
e-commerce company is able to collect rich information
about buyers behavior on ecommerce site. According to this
study, the majority of the customers in e-commerce were
male and online period was between 6 hours to 17.59 hours.
The age is also an important factor that affects customer
behavior. This can cause a market segment in the
e-commerce. Personal computers at home are more popular
for ecommerce than at work that can be attributed to the
convenience of online purchasing at home. Chinese is more
popular than English because Cluster 6 which used English
was about 16%. It shows that some customers begin to
access English e-commerce site to engage in international
trade.

V. REFERENCES
[1] Liuying Shen, Jana Hawley, etc. “E-commerce Adoption For Aupply
Chain Management in U.S. Apparel manufacturers”. Journal of Textile
and apparel, technology and management, vol.4, no.1, 2004, pp.1-10.
[2] N.R. Srinivasa Raghavan. “Data Mining in E-commerce: A Servey”.
Sadhana, vol,30, no.2, 2005, pp.275-289.
[3] Mobasher. B. Web Usage Mining and Personalization. Practical
Handbook of Internet Computing (ed.) M P Singh (CRC Press), 2004.
[4] Randall S. Sexton, Richard A. and Michael A. “Predicting
Internet/e-commerce use”, Internet Research, vol.5,2002, pp. 402-410.
[5] Strader, T.J., Shaw, M.J. Electronic Markets: Impact and implications,
in: Shaw, M., Blanning, R., Strader, T., Whinston, A. Handbook on
Electronic Commerce, Springer, Berlin, 2000, pp.77-98.
[6] Quinlan, J.R., Induction of Decision Trees, Machine Learning, Morgan
Kaufmann Publishers, Inc, CA, 1990.
[7] John A. Rodgers, David C. Yen and David C. Chou. Developing
E-business: a Strategic Approach, Information Management &
Computer Security, no.4, 2002, pp.184-192.

Dattatray V. Bhate received his B.E. in Computer Engineering from

university of pune in 2009, pursuing M. Tech in CSE from JNTU,
Hyderabad. Presently he is working in ZES’s Dnyanganga College of
Engineering and Research, Pune from 2 years. Areas of research interest in
Data Mining, Text Mining.

19