IJRRAS 10 (1) ● January 2012 www.arpapress.com/Volumes/Vol10Issue1/IJRRAS_10_1_14.

pdf

REVIEW OF LITERATURE ON DATA MINING

Mrs. Tejaswini Abhijit Hilage1 & R. V. Kulkarni 2
1
Assistant Professor, Deshbhakt Ratnappa Kumbhar College of Commerce, Kolhapur, India
Email : Tejaswini.Hilage@Gmail.Com
2
Professor and HOD, Chh. Shahu Institute of Business Education And Research Centre, Kolhapur, India
Email : Drrvkulkarni@Siberindia.Co.In

ABSTRACT
Data mining is used for mining data from databases and finding out meaningful patterns from the database. Many
organizations are now using these data mining techniques. In this paper authors has reviewed the literature of data
mining techniques such as Association Rules, Rule Induction Technique, Apriori Algorithm, Decision tree and
Neural network. This review of literature focuses on how data mining techniques are used for different application
areas for finding out meaningful pattern from the database.

KEYWORDS: Association Rules, Rule Induction Technique, Apriori Algorithm, Neural Network, Decision Tree.

REVIEW OF LITERATURE
Data mining techniques provide a popular & powerful tool set to generate various data driven classification systems.
Leonid Churilov, Adyl Bagirov, Daniel Schwartz, Kate Smith and Michael Dally had already studied about
combined use of self organizing maps & nonsmooth, nonconvex optimization techniques in order to produce a
working case of a data driven risk classification system. The optimization approach strengthens the validity of self
organizing map results. This study is applied to cancer patients. Cancer patients are partitioned into homogenous
groups to support future clinical treatment decisions.
Most of the different approaches to the problem of clustering analysis are mainly based on statistical, neural
network, machine learning techniques. Bagirov et al. [4] propose the global optimization approach to clustering and
demonstrate how the supervised data classification problem can be solved via clustering. The objective function in
this problem is both nonsmooth and nonconvex and has a large number of local minimizers. Due to a large number
of variables and the complexity of the objective function, general purpose global optimization techniques, as a rule
fail to solve such problem. It is very important therefore, to develop optimization algorithm that allow the decision
maker to find “deep” local minimizers of the objective function. Such deep mininizers provide a good enough
description of the data set under consideration as far as clustering is concerned. Some automated rule generation
methods such as classification and regression trees are available to find rules describing different subsets of the data.
When the data sample size is limited, such approaches tend to find very accurate rules that apply to only a small
number of patients. In Schwarz et al. [16] it was demonstrated that data mining techniques can play an important
role in rule refinement even if the sample size is limited. For that at first stage methodology is used for exploring
and identifying inconsistencies in the existing rules, rather than generating a completely new set of rules. K-mean
algorithm lies in the improved visualization capabilities resulting from the two dimensional map of the cluster.
Kohonen developed self organizing maps as a way of automatically detecting strong features in large data sets. Self
organizing map finds a mapping from the high dimensional input space to low dimensional feature space, so the
clusters that form become visible in this reduced dimensionability. The software used to generate the self organizing
maps is Viscovery SOMine (www.eudaptics.com), which provides a colorful cluster visualization tool, & the ability
to inspect the distribution of different variables across the map.
The subject of cluster analysis is the unsupervised classification of data & discovery of relationship within the data
set without any guidance. The basic principle of identifying this hidden relationship is that if input patterns are
similar, they should be grouped together. Two inputs are regarded as similar if the distance between these two inputs
is small.
This study demonstrates that data mining techniques can play an important role in rule refinement, even if the
sample size is limited. Leonid Churilov, Adyl Bagirov, Daniel Schwartz, Kate Smith and Michael Dally
demonstrated that both self organizing maps & optimization based clustering algorithms can be used to explore
existing classification rules, developed by experts and identify inconsistencies with a patient database. As the
proposed optimization algorithm calculate clusters step by step and the form of the objective function allow the user
to significantly reduce the number of instances in a data set. A rule based classification system is important for the
clinicians to feel comfortable with the decision. Decision tree can be used to generate data driven rules but for small
sample size these rules tend to describe outliers that do not necessarily generalize to larger data sets.

107
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

Anthony D Anna & Oscar H. Gandy develop a more comprehensive understanding of data mining by examining the
application of this technology in the marketplace. As more firms shift more of their business activities to the web,
increasingly more information about consumers and potential customers is being captured in web server logs.
Anthony D Anna & Oscar H. Gandy examine issues related to social policy that arise as the result of convergent
developments in e_business technology and corporate marketing strategies. About consumers and potential
customers is being captured in web server logs. Sophisticated analytic and data mining software tools enable firms
to use the data contained in these logs, to develop & implement a complex relationship management strategy.
Individuals whose profile suggest that they are likely to provide a high lifetime value to the firm will be provided
opportunities that will differ from those that are offered to consumers with less attractive profiles. Analytic software
allows marketers to combine through data collected from multiple customers touch points to find patterns that can be
used to segment their customer base. Web generated data includes information collected from forms, transactions as
well as from clickstream records.
Artificial neural networks are designed to model human brain functioning through the use of mathematics. Like
neural network data mining through the use of decision tree algorithms discerns patterns in the data without being
directed. According to Linoff “decision trees work like a game of 20 questions”, by automatically segmenting data
into groups based on the model generated when the algorithms were run on a sample of the data (1998, p. 44).
Decision tree models are commonly used to segment customers into “statistically significant” groups that are used as
a point of reference to make predictions (Vaneko and Russo, 1999). Both neural networks & decision trees require
that one knows where to look in the data for patterns, as a sample of data is used as a training device. The use of
market basket analysis & clustering techniques does not require any knowledge about relationships in the data,
knowledge is discovered when these techniques are applied to the data. Market basket analysis tools sift through
data to let retailers know what products are being purchased together. Clusters prove to be most useful when they
are integrated into a marketing strategy.
The software companies that market personalization products that use data mining techniques for knowledge
discovery, speaks to their potential clients in a language that make the benefits of these systems unmistakable.
However customer relationship management appears to be the philosophy that will drive marketing strategies in the
21th century. Customer relationship management focuses not on share market, but on share of customer. Marketing
strategists have been able to demonstrate that a firm’s profitability can increase substantially by focusing marketing
resources on increasing a firm’s share of its customers business rather than increasing its number of customers
(Peppers & Rogers, 1993).
One of the basic tenets behind customer relationship management is the Pareto Principle, the notion that 80% of any
firms profit is derived from 20% of its customers. Engaging in a dialogue with that 20% in order to ascertain what
their needs are & offering goods & services to meet those needs are said to be what customer relationship
management is all about. Data mining technologies have allowed firms to discover and predict whom their most
profitable customers will be by analyzing customer information aggregated from previously disparate database. The
web has created a forum for firms to engage in a one to one dialogue with particular segments of their customer base
in order to ascertain what the needs of those segments are.
Huda Akil, Maryann E. Martone, David C Van Essen made a study about understanding the brain requires a broad
range of approaches and methods from the domains of biology, psychology, chemistry, physics & mathematics. The
fundamental challenge is to decipher the “neural choreography” associated with complex behaviors and functions
including thoughts, memory, actions and emotions. National Institute Of Health recently launched the Human
Connectome Project and awarded grants to two consortia. The consortium led by Washington University of
Minnesota aims to characterize whole brain circuitry & its variability across individuals in 1200 healthy adults.
Neuroimaging & behavioral data from the HCP will be made freely available to the neuroscience community via a
database & a platform for visualization & user friendly data mining. The informatics effort involves major challenge
owing to the large amounts of data, the diversity of data types & the many possible types of data mining. Some
investigators will drill deeply by analyzing high resolution connectivity maps between all gray matter locations. It is
inefficient for individual researcher to sequentially visit and explore thousands of databases. To promote the
discovery and use of online databases, the NIF created a portal through which users can search not only the NIF
registry but also the content of multiple databases simultaneously. The current NIF federation includes more than 65
databases accessing 30 million records in major domain of relevance to neuroscience. Beside every large genomic
collection, there are nearly 1 million antibody records, 23000 brain connectivity records and more than 50,000 brain
activation coordinates. Many of these areas are converted by multiple databases, which the NIF knits together into a
coherent view. The NIF users should be able not only to locate answers that are known but to mine available data in
ways that spur new hypothesis regarding what is not known. Perhaps the single biggest roadblock to this higher
order data mining is the lack of standardized frameworks for organizing neuroscience data. Individual investigators
often use terminology or spatial coordinate systems customized for their own particular analysis approach. This

108
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

customization is a substantial barrier to data integration, requiring considerable human effort to access each
resource, understand the context and content of the data and determine the conditions under which they can be
compared to other data sets of interest.
Neurolex terms are accessible through a wiki that allows users to view, augment and modify these concepts. The
goal is to provide clear definitions of each concept that can be used not only by human but by automated agents,
such as NIF, to navigate the complexities of human neuroscience knowledge. A key feature is the assignment of
unique resource identifier to make it easier for search algorithms to distinguish among concepts that share the same
label.
So in future good planning and future investment are needed to broaden & harden the overall framework for
housing, analyzing, and integrating future neuroscience knowledge. The international neuroinformatics coordinating
facility plays an important role in coordinating and promoting this framework at a global level.

Following suggestions are given by Huda Akil, Maryann E. Martone, David C Van Essen:

Neuroscientists should as much as feasible, share their data in a form that is machine accessible, such as through a
web based database or some other structured form that benefits from increasingly powerful search tools. Database
spanning a growing portion of the neuroscience realm need to be created, populated and sustained. This effort needs
adequate support from federal & other funding mechanism. Because databases become more useful as they are more
densely populated, adding to existing database may be preferable to creating customized new one. Some type of
published data should be reported in standardized table formats that facilitate data mining. Cultural changes are
needed to promote wide spread participation in this endeavor.
Chandrika Kamath in her study said that the size & the complexity of the data from scientific simulations,
observations and experiments becoming a major impediment to their analysis. To enable scientists to address this
problem of data overload and benefit from their improved data collecting abilities, the Sapphire project team has
been involved in the research, development & application of scientific data mining techniques for nearly a decade.
According to her team the raw data available for analysis was in the form of images, structured or unstructured mesh
data with physical variables at each mesh point or time series data collected by different sensors. They designed and
built a software toolkit with separate modules for different tasks such as denoising, background subtraction to
identify moving objects in video, dimension reduction to identify key characteristics of objects, pattern recognition
for clustering and classification. Recognizing that for each task different methods were likely to be appropriate
based on the data; we used object oriented techniques to provide a uniform interface to the algorithms. Data from
first survey are available in two forms of image maps & catalog. Goal of this scientific application is not to build a
predictive model but to discover a set of features that may provide insights into the phenomena of interest.
Study made by Fadi Thabtah about associative classification mining said associative classification integrates two
known data mining tasks, association rule discovery and classification to build a model for the purpose of prediction.
Classification and association rule discovery are similar tasks in data mining, with the exception that the main aim of
classification is the prediction of class labels, while association rule discovery describes correlations between items
in a transactional database. Rule induction approach such as IREP (Furnkranz & Widmer, 1994) and RIPPER
(Cohen, 1995) derive local sets of rules in a greedy manner. The derived rules are local because when a rule is
discovered, all training data objects associated with it are discarded & the process continues until the rule found has
an acceptable error rate. This means rules are discovered from participations of the training data set & not from the
whole training data set once. The search process for the rules is greedy as most rule induction algorithms normally
look for the rule that maximizes a statistical measure.

Example:-

The IREP rule induction algorithm constructs the rules based on first order inductive learner, gain measure (Quinlan
& Cameron Jones, 1993). This means that the attribute value in the training data set with the best FOIL gain is
chosen first as a member of the current rule left hand side (antecedent) and the process is repeated until a stopping
condition is met.

109
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

Associative classification steps:

Training data

Training data

Frequent rule items

Set of class association rule

Classifier

Test data

In the Apriori association rule discovery algorithm (Agrawal and Shrikant, 1994) the discovery of frequent itemsets
is accomplished in levels, where in each level Apriori uses itemsets found to be frequent in the previous level to
produce new candidate itemsets. Apriori utilizes the downward closure property with aim of speeding up the search
process by reducing the number of candidate itemsets at any level. The downward closure property ensures that all
subsets of a frequent itemset must be frequent as well. If an itemset is infrequent at any level, it will be removed
because any addition of items to it will not make it frequent. Apriori uses this property to prune candidate itemsets
that have infrequent subsets before counting their support at any level. This should reduce the time to produce and
compute the support for all items combinations in the transactional database.

In Perkowitz & Etzioni (1998), the problem is defined in terms of automatic constructions of index pages based on
logfile data, with the goal of constructing index pages that provide users access to information that they are likely to
view. The algorithm proposed has four steps :

1. Processing the logfiles into user visits.

2. Computing co-occurrence frequencies between pages and creating a similarity matrix.
3. Creating a graph from this matrix & then finding cliques in this graph.
4. Creating index pages corresponding to each clique in the graph.

The algorithm is evaluated based on how often users simultaneously visit the pages in the index page. Shrikant &
Yang (2001) propose another interesting approach to the problem of automatically synthesizing web pages. They
address the issue of comparing the “expected location” of various pages with the actual location where the page is,
and they propose an algorithm that automatically finds such pairs of web pages. The main idea is examine user
session in the logfiles and identify backtracking points where a user backtracks to a previous page & continues the
session. An objective of website design can be to maximize navigational simplicity by minimizing the average
number of clicks required to get to any page. Further there may be several constraints on the design. Yet another
constraint may require the sports & finance pages to be explicitly linked to each other based on input from the
marketing department. However it is difficult to provide user specific constraint a priori. Data Mining can help to
address this problem. First the website is structured as per the solution to the optimization problem. As the user
navigates this site, data on user’s access pattern is gathered. Data Mining can thus identify additional constraint
using patterns identified by this data. Data Mining can find most user access the finance & sports pages together.
This information can then be used as a constraint. The optimization problem is then solved incorporating the
additional constraint. This process can be iteratively done until some user defined stopping criteria. This example
demonstrates the opportunity for new research to identify similar problems that can be formulated in the context of

110
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

website design. In this mode, problem in website design will be mainly formulated as optimization problems.
However Data Mining will play a key role in the specification of the optimization step and will also play a part in
the specification of the constraints.
Another interesting customer interaction issue is the provision of personalized services to customers, including
recommendations. Most of the approaches developed for solving this type of recommendation problem use
statistical & data mining approaches and usually tries to determine “good” products to recommend to the customer.
Various existing recommendation method were classified by Balabanovic & Shoham (1997) into content based,
collaborative & hybrid approaches and have been reviewed in survey (Pazzani 1999, Schafer et al. 2001,
Adomavicius and Tuzhilin 2003)

Example :-
Amazon.com may want to determine which best 10 books to put on the customers welcome page. The challenge for
this problem is that the rating function is usually partially specified. Once the optimization problem is defined, data
mining can contribute to its solution by learning additional constraints with data mining methods, thereby
significantly reducing the search space. For example, in the case of online wine store, we can learn that a customer
usually prefers to buy red inexpensive wines, however on certain occasions, such as his wife’s birthday, he usually
buys midrange white Chablis. By storing this information in the customers profile and invoking it on these special
occasions, much tighter constraints on recommendations can be specified. In this section, we reviewed some of the
existing approaches to solving three main analytical e_CRM problems. First is specifying effective performance
metrics and finding optimal solutions based on them. Second is developing effective customer analysis & third is a
customer interaction method.
Another study is made by Tae Kyung Sung, Namsik Chang and Gunhee Lee about how data mining approach
develop bankruptcy prediction model suitable for normal & crisis economic condition. It observes the dynamics of
model change from normal to crisis condition and provides interpretation of bankruptcy classification. The
bankruptcy prediction model revealed that the major variables in predicting bankruptcy were “cash flow to total
assets” and “productivity of capital” under normal conditions and “cash flow to liabilities”, “Fixed assets to
stockholders equity and long term liabilities” under crisis conditions. The accuracy rate of final prediction models in
normal conditions & in crisis conditions were found to be 83.3 % and 81.0 % respectively. When the normal model
was applied in crisis situations, prediction accuracy dropped significantly in the case of bankruptcy classification
(from 66.7 percent to 36.7).
As we are in the 21th century, corporate bankruptcy in the world, especially in East Asia, has reached an
unpredictable level. Corporate bankruptcy brings with it economic losses to management, stockholders, employees,
customers and others together with great social and economical cost to the nation. Thus accurate prediction of
bankruptcy has become an important issue in finance. Since the seminal study of Altman on bankruptcy prediction,
numerous follow up studies have tried to further develop appropriate models, by applying data mining techniques
including multivariate discriminate analysis, logistical regression analysis, profit analysis, genetic algorithms, neural
networks, decision trees and other statistical & computational methods. It is worth noting however that all of these
bankruptcy models assume “normal” economic conditions.
The detection of corporate failures is a subject that has been particularly amenable to financial ratio analysis.
According to Altman, the first study was done in 1935 by smith & winakor during the great depression era, then in
1942, Merwin showed that failing firms exhibit significantly different ratio than do successful firms. Beaver then
applied univariate analysis of financial ratios to predict corporate bankruptcy, while others strongly recommended
multivariate analysis.
The breakthrough bankruptcy prediction model, the Z – score model developed by Altman came in the late 1960s.
The five variables Z – score model using multiple descrimenant analysis showed by very strong predictive power.
Most of the studies including that of Altman used relatively small firms in their samples; generalization of research
results was hard to accept. Altman, Haldeman and Narayanan therefore developed the ZETA model to be applied to
larger firms, not limited to specific industries.
The study made by HE Zengyou, XU Xiafei and DENG Shengchun said that clustering is an important KDD
technique with numerous applications, such as marketing & customer segmentation. Clustering typically groups data
into sets in a way that intra – cluster similarity is maximized, while the inter cluster similarity is minimized. Many
efficient clustering algorithms such as ROCK, C2P, DBSCAN, BIRTH, CURE,CHAMELEON, wavecluster &
CLIQUE have been proposed by the database research community.
Most previous clustering algorithm focus on numerical data whose inherent geometric properties can be exploited
naturally to define distance function between data points. However many of the data in databases are categorical,
where attribute values cannot be naturally ordered as numerical values. An example of categorical attribute is shape
whose values including circle, rectangle, ellipse etc. Due to the special properties of categorical attributes, the

111
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

clustering of categorical data seems more complicated than that of numerical data. They present Squeezer, a new
clustering algorithm for categorical data. The basic idea of Squeezer is simple. Squeezer repeatedly read tuples from
a dataset one by one. When the first tuple arrive, it forms a cluster alone. The consequent tuples are either put into
existing clusters or rejected by all existing clusters to form a new cluster by given a similarity function defined
between a tuple and a cluster. The Squeezer algorithm only makes one scan over the dataset, thus is highly efficient
for disk resident datasets where the I/O cost becomes the bottleneck of efficiency. The main objective of Squeezer is
the combination of efficiency and scalability. Experimental result shows that Squeezer achieves both high quality
clustering results & scalability. Main contribution of the study are the algorithm is suitable for clustering data
streams, where given a sequence of points, the objective is to maintain consistently good clustering of the sequence
so far, using a small amount memory & time, outliers can be handled efficiently and directly, the algorithm does not
require the number of desired clusters as an input parameter. This is very important for the user who usually does
not know this number in advance. The only parameter to be pre –specified is the value of similarity between the
tuple and the cluster, which incorporates the users expectation that how close the tuples in a cluster should be.
Swift (2002) describes analytical eCRM as a four step iterative process consisting of collecting & integrating online
customer data, analyzing this data, building interactions with customers based on this analysis such that certain
performance metrics such as LTV are optimized. A typical performance metric used in many CRM applications is
the LTV of a customer. One of the key questions in CRM is how to develop proactive customer interaction
strategies that maximize LTV. This key problem is viewed by some as the “holy grail” of CRM.
Traditionally the problem of estimating LTV is divided into two components. One is estimating how long a
customer will stay& second is estimating the flow of revenue from the customer during this period.
Estimating the flow of revenue during customer’s lifetime was done with parametric models in the marketing
literature. With respect to estimating customer tenure, Schmittlein et al. (1987) provide analytical models that
determined whether a customer at any point in time is “active”, by identifying a set of qualitative criteria that
capture when a customer is more likely to be active. The concept of customer retention is a natural extension of this
approach, because it deals with trying to prevent a customer from becoming inactive and hence it is a proactive
method of increasing LTV. Blattberg and Deighton (1991) present a more general framework for inactive marketing
for LTV, suggesting the key notion that customers are “addressable” and can be engaged in interaction. This notion
of customer interaction is further explored in Dreze and Bonfrer (2002) in which the problem of optimal
communication to maximize LTV is addressed. They show that both too little and too much communication can
result in a firm’s failure to capture adequate value from its customers. This analysis provides useful insights into the
LTV optimization problem. However this work considers only optimal communication frequency with customers &
the model does not take into account the existence of data on customer behavior.
In contrast the Data Mining community studied the LTV problem in the presence of large volumes of customer data
(Mani et al.1999, Roset et al. 2002). In particular Mani et al. (1999) predict customer tenure using classical survival
analysis methods by building a neural network & training it on past customer data. However Mani et al. (1999) do
not address the problem of computing optimal parameters for LTV models. Roset et al. (2002) compute LTV based
on large volumes of customer data by focusing on using Data Mining to estimate customer churn & future revenues.
They use statistical and data mining methods to estimate future revenues. They point out that it is difficult to solve
an LTV optimization problem in the presence of large volumes of data and Rosset et al. (2002) stopped short of
doing this. Similar Dreze and Bonfrer (2002) studied only an optimization problem & did not deal with large
volumes of data. Therefore developing new algorithms computing optimal LTVs and utilizing optimization & data
mining methods in the presence of large volumes of data constitutes an important & challenging problem for
operations research or management science researchers.
One way to deal with the complexity of the general LTV optimization problem is to reduce it to simpler types of
problems. One such reduction may consider various heuristics producing higher LTV values, rather than attempting
to find an optimal solution.
For example, we can study customer attention problems & determine policies that will result in lower attrition rates
and therefore higher customer LTVs. Another way to deal with this complexity is to seek to optimize simpler
performance measurers that can serve as proxies for LTV. We can use customer satisfaction rates with various
offerings as one such measure.
Customer analysis includes two main steps in the eCRM context. First is preprocessing data that tracks various
online activities of the customers – this involves staring with individual user clicks on a site and constructing logical
users “session” and summary variables. Second is building customer profiles from this and other data. At a general
level, a profile is a set of patterns that describe a user. Data Mining is used to learn these patterns from data.
Customer profiles are then built from these results.
Current literature considers heuristic methods for analyzing click stream data generated by websites. One of the
most important problem is the session identifies problems, which determines how to group consecutive clicks into

112
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

session. This is an important business problem because most of the users tracking system developed by such
companies such as Epiphany and Blue Martini provided only ad hoc solutions to the session identification problem.
Some popular session identification methods include session level characterization (Srivastava et al. 2000,
Theusinger and Huber 2000) that aggregates user clicks into session, a fixed length sliding window method (Cooley
and Mobasher 1999) that breaks a session into several slinding windows, and different types of clipping methods
that break a session into windows of different sizes using various splitting methods. (Brodley and Kohavi 2000,
Vandermeer et al. 2000, Padmanabhan et al. 2001). Zheng et al. (2003) demonstrate that various session
identification methods can produce radically different conclusions derived from the same data.
The main reason for preprocessing click stream data is to build a model, such as one that would predict the
likelihood that a current user’s session would result in a purchase. Accurate model are crucial for such problems,
requiring optimal preprocessing of the click stream data. To partition the click stream data into session, one needs to
specify optimization criteria. This can be done by first identifying “similar” groups of consecutive pages in the click
stream, which can then be partitioned into sessions to maximize intrasession similarities and intersession
differences. One way to specify these similarities & difference is to identify the variance of some browsing measure,
such as the time spent viewing a page.
Building rich & accurate profiles of customer based on their transactional histories is also crucial in many CRM
applications, including recommendation applications, one to one marketing & personalized web content delivery.
These user profiles can contain factual information about the users such as demographic & psychological data, a set
of rules capturing behavior of the user.
Because many of the discovered rules can be superiors, irrelevant or trivial, one of the main problem is how to select
an optimal set of rules for each customer from the set of rules previously discovered with Data Mining methods. In
such applications personalized content needs to be delivered in real time, while the customer is waiting online.
Therefore content delivery discussion should be driven by only a few rules to guarantee real time delivery and
relevant content for the customer (Davis 1998). Consequently optimization can help customer profiling and other
applications to select a small number of the most important patterns from the set of previously discovered patterns.
The aim of the study made by T. W. Rennie and W. Roberts was to demonstrate the epidemiological use of multiple
correspondence analyses as applied to tuberculosis (TB) data from North East Landon primary care trusts between
the years 2002 & 2007 was used. TB notification data were entered for Multiple Corresponding Analysis allowing
display of graphical data output.
There has been rise in tuberculosis notifications in the UK since 1987. However excluding TB in Landon, rates of
TB in the UK are relatively low & stable. In the context of North East Landon, high rates of TB are observed in
some primary care trust areas whilst in others rates are relatively low. This demonstrates the complexity of TB
epidemiology in the UK and Landon & is suggestive of a range of factors that give rise to high rates of TB in
specific geographical areas. The enhanced tuberculosis surveillance (ETS) system was introduced in 1999 to aid
notification. These collected data show the different demographic and clinical profiles of patients observed in NE
London & may account for variations in TB rates. This requires appropriate statistical support & effective
communication to decision makers. However analysis of large volume of data with a large proportion of categorical
or nominal data that can display multiple associations may prove to be difficult to interpret if bivariate comparison is
made. Factor analysis & Principle Component Analysis are inappropriate methods of analysis for these data which
include a mix of continuous & categorical data. Multiple correspondence analyses are an analytical method that
allows analysis of multiple categorical variables. We demonstrate the use of MCA as a tool for performing
epidemiological mapping of TB patient variables. This may prove to be useful in identifying commissioning
priorities in NE London.
Multiple Correspondences Analysis is a multivariate extension of Correspondence Analysis that allows explanation
of relationships between two or more variables. By including two or more variables in this type of analysis the
complexity is increased, relationships between variables are described in terms of the variance of data.
The study made by Shakil Ahmed, Frans Coenen and Paul Leng consider strategies for partitioning the data to deal
effectively. Partitioning approach organizes the data into tree structure that can be processed independently. The
performance of these methods, depend on the size of original database.
The study made by Balaji Padmanabhan & Alexander Tuzhilin said previous work on the solution to analytical
electronic customer relationship management problem has used either data mining or optimization methods, but has
not combined the two approaches. By leveraging the strength of both approaches, the eCRM problems of customer
analysis, customer interaction & the optimization of performance metrics can be better analyzed. In particular many
eCRM problems have been traditionally addressed using DM methods. There are opportunities for optimization to
improve these methods .Balaji Padmanabhan & Alexander Tuzhilin describes these opportunities. Prior research has
used optimization methods for solving data mining problems (Mangasarian et al. 1990, Vapnik 1995, Fu et al. 2003)
and used DM methods for solving optimization problems (Brijis et al.1999, Campbell et al. 2001). In this survey we

113
IJRRAS 10 (1) ● January 2012 Hilage & Kulkarni ● Review of Literature on Data Mining

systematically explore how optimization and DM can help one another for certain customer relationship
management (CRM) applications in e_commerce, termed analytical eCRM (Swift 2002). Analytical eCRM includes
customer analysis, customer interactions and optimization of various performance metrics such as customer lifetime
value in web based e_commerce.
To illustrate how optimization and DM interact in eCRM settings, consider the following two important eCRM
problems. The first deal with finding the optimal lifetime value of a customer by determining proactive customer
interaction strategies resulting in maximal lifetime profits from that customer. These eCRM applications include
Max. of customer LTV, Customer analysis, including preprocessing click stream data & building profiles, Customer
interaction methods including website design & personalization.
By considering previous studies authors find out the scope to go for research in market basket analysis using three
different algorithms namely Association Rule Mining, Rule Induction Technique and Apriori Algorithm. Authors
will make a comparative study of three techniques and adopt the best conclusion.
In Association Rule Mining, we will generate association rules and calculate support and confidence. Assume
minimum support and minimum confidence. The rules satisfying both the criteria of minimum support & minimum
confidence is true otherwise false. Rule induction technique retrieves all interesting patterns from the database. In
rule induction systems the rule itself is of the simple form of “if this and this and this then this”. In some cases
accuracy is called the confidence and coverage is called the support. Accuracy refers to the probability that if the
antecedent is true that the precedent will be true. High accuracy means that this is a rule that is highly dependable.
Coverage refers to the number of records in the database that the rule applies to. High coverage means that the rule
can be used very often and also that it is less likely to be a spurious artifact of the sampling technique or
idiosyncrasies of the database. Assume minimum accuracy and minimum coverage. The rules satisfying both the
criteria of minimum accuracy & minimum coverage is true otherwise false.
The theory of Apriori algorithm is that “All nonempty subsets of a frequent itemset must also be frequent.” This
property prune the candidate which is not in any of the category & thus to reduce number of candidates.
Authors will collect the data from shopping mall and will apply the data mining algorithms to find out the
association between the products.

REFERENCES :-
[1] Leonid Churilov. Adyl Bagirov, Daniel Schwarta, Kate Smith, Michael Dally , Journal of management
information system : 2005, Data mining with combined use of optimization techniques and self organizing
maps for improving risk grouping rules : application to prostate cancer patients
[2] Anthony Danna, Oscar H. Gandy, Journal of business ethics : 2002, All that glitters is not gold : Digging
Beneath the surface of data mining.
[3] AC Yeo, KA Smith, RJ Willis and M Brooks, Journal of the operation research society : 2002 , A
mathematical programming approach to optimize insurance premium pricing within a data minning
framework.
[4] Shakil Ahmed, Frans Coenen, Paul Leng, Knowledge Information System : 2006, Tree based partitioning
of data for association rule mining
[5] Timothy T. Rogers, James L. Mcclelland, Behavioral & Brain Sciences : 2008, Precis of Semantic
Cognition : A Parallel Distributed Processing Approach.
[6] Ana Cristina, Bicharra Garcia, Inhauma Ferraz and Adriana S. Vivacqua, Arificial Intelligence for
engineering design, analysis and manufacturing : 2009, From data to Knowledge Mining
[7] Rachid Anane, Computer and the humanities : 2001, Data mining and serial documents.
[8] Balaji Padmanabhan and Alexander Tuzhilin, Institute for Operation Research and Management Science :
2011, On the use of optimisation for data mining : Theorotical Interaction and eCRM opportunities.

114