Application Of Data Mining For Customer Segmentation Of Bank

Purpose of Research

Discuss about the Application of Data Mining for Customer Segmentation of Bank.

In the banking domain, the consumers are subjected to computerization of financial operations, by and large. The process entails over operation of E-banking along with mechanical cellular machines have facilitated alteration in the basic conception of banking operations and the manner in which banking function is being performed. Since the year of 1990, the whole concept of banking experienced a complete overhaul in terms of operations (Thomas, 2000). The core operation got switched to databases, ATMs, and online transactions. In the banking sector, data is one of the important possessions and with the aid of data mining, important data may be recovered which is embedded in the database of the banks. The service of data mining in any financial domain is indispensible nowadays since it helps in optimization of business decisions by augmentation of value of consumers through the means of customer satisfaction and communication. In the sector, there exists bulk of data wherein the data is derived from numerous sources like, bank account transactions, along with loan repayments, transactions, credit card payments to name a few (Tsai and M.-L. Chen, 2010). Hence, it is deemed to be a valuable information with regard to the financial profile of consumer base which is contained in the huge operational database the same in further may be applied for betterment of the banking operation. At the outset, computerization or TBC are being applied in different branches for serving daily transactions along with design of new MIS or for that matter reorganization of existing system which may not be possible by the means if mere exchange of present Total Branch Computerization packages.

The purpose of this study is to engage in an in-depth analysis of the products and services offered by the bank with regard to their consumers for attracting other latent customers. This study mainly focuses on the impact or influence of consumer satisfaction to the efficacy in the market. The following account recognizes the diverse range of strategies offered by the bank for attainment of consumer satisfaction along with customer’s reaction to the bank offered products and services. It strives to analyze customer satisfaction which in turn contributes in the success of the bank overall. Furthermore, the detailed study reflects over frauds and other shortcomings of the banking sector (Van Gestel and B. Baesens, 2012).

Acquisition of loyal consumers is deemed to be a good prospect for different entities and the enterprise system projects like that of banking sectors. Hence, the organization’s primary attempt is to locate and obtain the loyal base of consumers for satisfying their needs and demands. As a matter of fact, the entities should also strive to devise patterns for fulfillment of customer needs.

Literature Review

Various experts opine that consumers exert various thoughts and emotions in keeping with personal requirements for performing and fulfillment of every consumer needs and desires. For that the banking sector must obtain a plethora of functions from various sources and thereby create modern policies and initiatives. Hence, consumer requirements and the profile may be performed on the basis of the lifestyle and consumer transaction history of the same (Vivek Bhambri, 2012). On the other hand, it is observed that consumer actions and behavior exert influence on the profitability of the company which is construed to improve the consumer profitability and customer revenue by retention of the consumer for a substantial period of time. Hence, data mining is being utilized for continuation of previously mentioned behavioral pattern which may solve the employment predicaments by association laws which are maintained in the database with utmost secrecy and confidentiality. Based on the research, document of data mining, measures that may be utilized by usage of various techniques which provides support for decision-making procedures in best possible methods when numerous users are striving for creating analogous alterations at uniform time zones and also transactions from the same branch. The Data Mining ploy is clinical in detection of frauds of various degrees like that of multiple transactions from the same branch.

Different experts opines about largely consumer profiling in the research report since it accumulates various duties and responsibilities for creation of connection in the divisions of the banking system for performing monetary transactions. The banking domain uses data mining technique for accessing data through various networks for continuation of relieving application of the particular set of banks (West S. Dellana, & J. Qian, 2005).

Kazi Imran and Qazi Baser Ahmed speak out that more or less every banks award loans to the users or consumers for verification of consumers’ relevant and key information, the likes of income of the customer, credit history rate of lending and payment period to name a few. The factual reality is even though most of the banks are most likely to be the recipient of loan, chances are there hat they might fall victim to the probability of frauds by the users. Hence, in order to ascertain the type, or nature of frauds in relation to loans, data mining is used quite efficiently (Xun Liang, 2006).

In the event of gathering information in the report, the aforesaid data mining scheme plays an important role for detection of consumer loans and credit card also. It also provides support to discover consumer needs. Hence, it may be deduced that we may contribute the necessity of the bank users by being observant of their transactions and profile. For identification of frauds, in the present context, banks are implementing Falcon’s approach since it is one of the easy methods which comes with lesser cost in comparison to data mining techniques. The data warehouse modus operandi is being utilized for creation of link between the divisions of the networks. In addition to that, it is deemed to be a proper functioning device which is purely secure in terms of efficient functioning or so (West S. Dellana, & J. Qian, 2005).

Methodology

The research topic is data mining in the banking domain and this account explains the effectiveness of data mining procedure for uncovering of frauds and other canvassing occuences in the same parlance. Various research paper has been used to decipher the diverse ways and techniques of data mining approach in the banking sector.

Sources of information: The information source has been the research papers and multifarious articles available pertaining to Data Mining in the banking Sector. In the papers, data mining ascends supremacy which is related to the banking operations, overall.

Observation: A host of articles, papers and articles which necessitated for the research paper. It has been observed that various procedures and techniques have been utilized for recognition of frauds and strategies needed to fulfill then consumer needs. In simple terms, it is used as a tool for measurement of different data and attributes which are inconsistent in the overall report. The concept of inconsistency deserves special mention in the light of other necessary issues pertaining to Data Mining (Xun Liang, 2006).

Analysis of this topic can be effectively undertaken by breaking down the topic into different aspects of the topic that can be analyzed individually and then topic or concept can be restructured in a manner to have a holistic understanding of the topic being discussed. In analytical research it is important to undertake research individually so that the topic being research is completely understood by the researcher and develop expert knowledge about the topic to the extent that the researcher have reconstruct and present the parts of the topic according to his own perspective or view point. This approach of research is used to set and employ a model that would enable the researcher to predict unknown entities among different classes in the banking system. Some of the most commonly used classification techniques are “neural nets”, “the Naïve Bayes techniques”, “decision trees”, and “support vector machines”. These classification methods are employed to detection of credit cards, health and vehicle insurance, and to detect corporate frauds, among other types of frauds. With the help of arrangements, models can be developed   that would be utilized to anticipate all the obscure objects or items in order to identify object belonging to distinctive classes. For basic order strategies neural systems, the Gullible Bayes procedure, choice trees, and help vector machines can be used. The purpose of using these methods are used to identify MasterCard, medicinal services, accident coverage, and corporate extortion, apart from that its can also be utilized various types of misrepresentation, and also in data mining in fraud detection.

Analysis

For the present scenario, it aims only to detect unwanted frauds or transactions in the particular location in the proposed i.e. falcon software to detect all the default transaction. Further, to employ the falcon software, in the proposed banking system it is important to use data miming. In addition, quantitative method of data collection, i.e. survey instruments employed to gather the required information

From the analysis of the collected through surveys, scholar articles and journals, following techniques is proposed to overcome the fraud as it is presented in the researches below:

Only one journal article from Korea presented on small and medium enterprise credit scoring as the Korean Government focuses on values knowledge based economy

Though there are few articles and literature on SME credit scoring, the field is expected to increase research on credit scoring in the future as the SME organizations are rapidly increasing in the economy that would attract researchers

In addition, it is observed that the most of the articles focuses on real non UCI datasets and has employed variable selection in the process of their model building

Further, it is noted that the trait for which the word being forecasted is derived by request rather than stated in a straight out fashion. This characteristic alluded as the anticipated characteristic. For this, neural systems and logistic model expectation are employed as the most commonly utilized models of forecasting methods.

The neural systems can be understood as non direct information measuring and displaying instruments that influenced by utility of the human mind and that utilizes a set of interconnected hubs. It is observed that neural systems are connected in arrangement and bunching and is focused on the following parameters. At first it can be seen as versatile method and secondly it can be used to create hearty models and lastly it can be seen as a model that allows arrangement procedures to be changed as per the requirements of the researchers such as if the preparing weights are situated. Neural systems are mostly linked with charge cards, accident coverage, and corporate frauds or misrepresentations. Further, neural systems can be employed to as tool to identify monetary extortion apparatus. Previous researchers have established the feasibility of neural systems, choice trees and Bayesian conviction to identify fraudulent transaction related expressions and to recognize elements connected with FFS. 

A decision tree (DT) is a decision support tool using a tree-like graph that helps in decision making and their possible outcomes or consequences. This graph connects every hub to the events, utility or costs. The connection lines reflect the prescient model that helps in perceiving the gap into several unrelated groups. The decision-three also helps in utilizing information and machine learning errands.  The choice trees are prescient choices that help the users in mapping the link between perceptions and conceivable results. The decision-trees can be used for several machine-learning based calculations such as C4.5, Id3 and Truck. The leaves symbolize expectations and the extensions symbolize conjunctions of peculiarities. The choice trees can also be put to utility for accident protection, MasterCard and corporate extortion. The relapse tree calculation is presented in systems for recognizing and anticipating the consequences of fake budgetary arrangement and proclamations. The system can be presented through various models such as choice trees, neural systems, logistic models and the Bayesian conviction system.

Hereditary calculations utilize an interactive tool that helps in calculating certain traits that displays evaluator choice conduct in extortion setting. In case, there is indecent recognition of issues of charge card extortion, the hereditary calculation and double help vector framework (BSVS) prove useful. These tools are focused around help vector machines (SVM).

Fluffy Rationale is a scientific strategy that helps in assigning an inferior rank or position to a specific group or gathering. The strategy groups subjective thinking and uses probability as a base for relegating information to a specific gathering. The master fluffy processes are critical as it empowers an individual to perform inexact thinking. The inexact thinking can be executed in three ways. Firstly, the execution can be modified with the numerical representation of ambiguous terms. However, it must be certain that the fluffy innovation can be demonstrated numerically. Secondly, execution can be upgraded through expansion of scope. This is beneficial when the situations are not characterized and fluffy philosophy can prove of utility through fractional participation demonstration. This demonstration of information may not be simply characterized in customary investigation.

Researchers have investigated piece of Expert Systems to analyse the limit of customers who supervise and enunciate clear beliefs. The expert system also helps in finding abilities for accounting coercion under several circumstances and settings. This expert system helps in providing a much strong proposal through careful examination of strategy. The investigation further confirms that the expert structure enhances commentators’ execution. The circumstances can be assorted and the evaluators divide work in a better manner with the usage or backing up from expert systems. The expert structure is critical for decision making and adapts to audit exercises.

Several data mining strategies such as neural frameworks, decision trees, K-closest neighbours and Bayesian conviction framework are endeavoured by a couple of researchers. The primary aim is to apply a combination of decision using multiple methodologies that can be used in recognizing misleading cash related clarifications. A few models are also utilized in recognizing budgetary tricks in countries such as New Zealand.

Conclusion:

Data mining is a process to concentrate learning from existing information. It is critical as it helps in managing accounts, searching valuable data and enables better decision making. Several methods such as database innovation, conversion of measurements, data science, visualization and machine learning are beneficial. The steps include determining information, combination, modification, mining, assessment and presentation of information. The information can be used by banks in different categories such as showcasing, recognition of misrepresentation, hazard management and speculation. The examples demonstrated can help the bank in estimating the future occasions helping in making appropriate choices. The banks are increasingly putting resources into information mining.

References:

Ding Sheng Wan, Chong Liu, Yuan Sheng Liu, Application of Data Mining in the Customer Segmentation of Bank, Microcomputer Applications,Vol 25,pp 31—34,No.2,2009.

Dr. K. Chitra, B. Subashini “Data Mining Techniques and its Applications in Banking Sector” International Journal of Emerging Technology and Advanced Engineering Website: www.ijetae.com (ISSN 2250-2459, ISO 9001:2008 Certified Journal, Volume 3, Issue 8, August 2013)

Dr. Madan Lal Bhasin, “Data Mining: A Competitive Tool in the Banking and Retail Industries”, the Chartered Accountant October 2006.

Hand, D.J. and W.E. Henley. (1997). Statistical classification methods in consumer credit scoring: a review. Journal of the Royal Statistical Society: Series A (Statistics in Society). 160(3), 523-541.

Mahesh Kharote, V. P. Kshirsagar “Data Mining Model for Money Laundering Detection in Financial Domain” International Journal of Computer Applications (0975 – 8887) Volume 85 – No 16, January 2014.

Ngai, E.W.T., H. Yong, Y.H. Wong, C. Yijun and S.Xin, 2011. The application of data mining techniques in financial fraud detection literature. Decision Support Syst., 50.

Patnaik, Debprakash, Sastry P. S., Unnikrishnan K. P, Inferring neuronal network connectivity from spike data: A temporal data mining approach, Scientific Programming, 2008, Vol. 16 Issue 1, p49-77.

Tan, P.N., M. Steinbach, and V. Kumar. (2006). Introduction to data mining. Pearson Addison Wesley Boston.

Thomas, L.C. (2000). A survey of credit and behavioral scoring: forecasting financial risk of lending to consumers. International Journal of Forecasting. 16(2), 149-172.

Tsai, C.F. and M.-L. Chen. (2010). Credit rating by hybrid machine learning techniques. Applied Soft Computing. 10(2), 374-380.

Van Gestel, T. and B. Baesens. Credit Risk Management: Oxford University Press.

Vivek Bhambri “Role of Data Mining in Banking Sector”, International Indexed & Referred Research Journal Vol. 3 Issue-33, 2012.

West, D., S. Dellana, and J. Qian. (2005). Neural network ensemble strategies for financial decision applications. Computers & Operations Research. 32(10), 2543-2559.

Xun Liang, Data mining algorithms and application, Peking university press, April .2006.