داده کاوی یکپارچه و مدل امتیاط دهی رفتاری برای تجزیه و تحلیل مشتریان بانک

Analyzing bank databases for customer behavior management is difficult since bank databases are multi-dimensional, comprised of monthly account records and daily transaction records. This study proposes an integrated data mining and behavioral scoring model to manage existing credit card customers in a bank. A self-organizing map neural network was used to identify groups of customers based on repayment behavior and recency, frequency, monetary behavioral scoring predicators. It also classified bank customers into three major profitable groups of customers. The resulting groups of customers were then profiled by customer's feature attributes determined using an Apriori association rule inducer. This study demonstrates that identifying customers by a behavioral scoring model is helpful characteristics of customer and facilitates marketing strategy development.

Contemporary marketing strategies perceive customers as important resources to an enterprise. Therefore, it is essential to enterprises to successfully acquire new customers and retain high value customers. To achieve these aims, many enterprises have gathered significant numbers of large databases, which then can be analyzed and applied to develop new business strategies and opportunities. However, instead of targeting all customers equally or providing the same incentive offers to all customers, enterprises can select only those customers who meet certain profitability criteria based on their individual needs or purchasing behaviors (Dyche & Dych, 2001). Credit scoring and behavioral scoring are techniques that help decision makers to realize their customers. Credit scoring models help to decide whether to grant credit to new applicants by customer's characteristics such as age, income and martial status (Chen & Huang, 2003). Behavioral scoring models help to analyze purchasing behavior of existing customers (Setiono, Thong, & Yap, 1998). These two scoring models are highly related to the field of classification analysis by statistical analysis (Hand, 1981 and Johnson and Wichern, 1998), especially classification analysis by neural networks in the field of data mining (Lancher, Coats, Shanker, & Fant, 1995). Until now, most existing data mining approaches have been discovering general rules (Agrawal et al., 1993, Bult and Wansbeek, 1995 and Setiono et al., 1998), predicting personal bankruptcy (Dasgupta et al., 1994, Desai et al., 1996 and Zhang et al., 1999) and credit scoring (Kim and Sohn, 2004, Lancher et al., 1995 and Sharda and Wilson, 1996) in bank databases. Few works have studied the mining of bank databases from the viewpoint of customer behavioral scoring (Sharda & Wilson, 1996). More specifically, we wanted to look at both the account data of the customers and their credit card transactions. With these data, the aim was to discover interesting patterns in the data that could provide clues about what incentives a company could offer as better marketing strategies to its customers. As shown in Fig. 1, this study presents a two-stage approach for behavioral scoring analysis of implicit knowledge using bank customer account and transaction data. Topics discussed include data preprocessing, customer behavior scoring modelling, sensitivity analysis of relative importance attributes contributing to the customer profiling, and the two stages of the behavioral scoring model itself.The key feature of the two-stage behavioral scoring model is a cascade involving self-organizing map (SOM) and an Apriori association rule inducer. An SOM (Kim and Sohn, 2004 and Kohonen, 1995) is an unsupervised learning algorithm that relates multi-dimensional data as similar input vectors to the same region of a neuron map, and Apriori (Agrawal et al., 1993) is mainly used to find out the potential relationships between items or features that occur synchronously in the database. In the first stage of the approach presented here, a conceptual customer behavioral scoring model was established to predict profitable groups of customers based on previous repayment behavior and RFM (Bult & Wansbeek, 1995) behavioral scoring predicators. This SOM was employed to classify customers into three major profitable groups of customer: revolver user, transactor user, and convenience user. Once the SOM identified the profitable groups of customers, an Apriori profiled each group of customers focusing on demographic and geographic characteristics for building and maintaining the most profitable customer base. The customer profile then was used to describe a representative case in each group of customers, and served as a tool for establishing better bank marketing strategies. After analyzing the bank database, this study demonstrates that customer behavior scoring models are an effective method for banks to realize their most profitable customers. We conclude by analyzing target groups of customers using the proposed two-stage behavioral scoring model. For a better understanding of our solutions, this study is organized as follows. Section 2 makes a description of the analyses methodology. An integrated data mining and behavioral scoring model was presented. Section 3 assesses neural networks as a tool for customer segmentation while using past repayment behavior and RFM scoring variables to build behavioral scoring models. Section 3 also presents the processes of creating customer profiles according to their feature attributes as determined by an Apriori association rule inducer. Finally, conclusions are made in Section 4.

Credit and behavioral scoring have become useful tools to model financial problems. However, most studies have concentrated on building an accurate credit scoring model to decide whether or not to grant credit to new applicants. In order to strengthen customer behavior management for existing credit card customers, we created a behavioral scoring model using neural networks and an association rule inducer. The existing customers were divided into three profitable groups of customers according to their shared behavior and characteristics. Marketers then can infer the profiles of customers in each group and propose management strategies appropriate to the characteristics of each group. This study provides a good method of analyzing bank databases. Beyond simply understanding customer value, the bank gains the opportunities to establish better customer relationships while increasing customer loyalty and revenue. Additionally, this two-stage behavioral scoring model also can be applied to predicate personal bankruptcy among bank customers to the account database. Further research may aim at time-series behavioral scoring models that could include the change of credit status in every period. Credit card customers could be segmented into more subgroups according to newly developed predicators and so on. Thanks to this paper and many others, more detailed management and marketing strategies can be implemented according to more detailed customer sub-groups.