Volume III, Issue III, March 2016 IJRSI ISSN 2321 - 2705 www.rsisinternational.org Page 51 Selection of Favourable Customers by Discovering Trends in Buying Pattern Rohan Kishan, Prashant Shinde, Sachin Mane, Sandhya Pati Department of Computer Engineering. Fr. Conceicao Rodrigues Institute of Technology, Vashi Abstract—Every organization needs to ensure that it has a strong support from its customers so that it can survive in this world of cut-throat competition. Every day organizations are coming up with new strategies to retain its customer base .One such strategy that is widely employed by organizations is to provide them monetary benefits in the guise of cash discounts or coupons. However, providing such benefits to all its customers would prove to be a costly affair for the company. To cut down the costs of this approach it is advisable to provide profits to only those customers who contribute greatly to the benefit of the organization. This would motivate the other customers to contribute more towards the development of the organization in order to acquire benefits.In this paper, we have implemented an approach that selects customers on the basis of their contributions to the enterprise and rewards them on the same basis. Keywords-Expert System, Beneficial Customers, lucrative offers. I. INTRODUCTION he use of Data Mining and Customer Relationship Management (CRM) has greatly risen in the recent years especially by the private organizations [1][6]. Different corporations such as TESCO & Walmart have benefitted greatly by the incorporation of data mining in their usual CRM operations [4].Many methods are followed by organization in order to maintain good relations with their customers. A generally followed technique is to select customers on the basis of his or her frequency i.e. the number of times they visited the organisation in a fiscal year. Even though these techniques are widely employed the effectiveness of these methods is highly doubtful. This is due to the fact that a high frequency customer may actually be generating less revenue for the enterprise. In this project we suggest an approach that actually judges the customers on the basis of their real worth to the organization. II. LITERATURE REVIEW Han And Kamber in 2006 suggested that the most important factor for huge advancements in the field of data mining was the availability of large amount of data and the ability to shape it for useful purposes [3].The Customer Intimacy And Customer Portfolio Analysis (CPA) are two of the most important phases of the CRM model and data mining plays an important role in both of them[2]. The main goal of CPA is to select profitable customers and to suggest the enterprise to continue its relations with them [2]. Data mining makes this possible with the use of techniques such as classification [2].Frank and Buttle have illustrated a use of data mining technique by Natwest which was instrumental in increasing their profit rate. Natwest used the classification technique and used attributes such as credit rating and life time value to divide its customer base into different classes. The most important class of customers enjoyed benefits such as personalised advisers to help them with their investments. 1.1 Existing Systems The output that is generated by most data mining tools may be either customer models or visualized results. These results are studied and the findings are showcased in the form of a graph [4]. Therefore the burden of finding results from the ROC [7][6] (receiver operating characteristic)curve lies on human workers. The actions taken may differ from man to man. This makes it extremely difficult to come up with an analysis report for processing customer relations. The best course of action will be taken by the in house CRM expert. Data mining on being applied to CRM, will display graphical outputs of customer behaviour. III. PROPOSED SYSTEM The ultimate objective of this system is to maintain good relations between customer and the enterprise by providing them appropriate discounts. The calculation of these discounts will be based on an attribute called Life Time Value (LTV) [2]. Here the decision tree algorithm is taken into account as the data mining tool. The best tool to find out optimal customers is the decision tree algorithm [8][7][5]. Based on the result of the decision tree further actions can be taken by the administrator. In other words, we can use this decision tree algorithm to calculate the rewards for customers based on their actions and also get the results in a graphical format[5]. 3.1 Modules The application judges the customers on various fronts. In order for the customer to be deemed profitable, it is necessary to study his activities. The application records every action of T