ABSTRACT

Clustering is a data mining technique used to place data elements into related groups without advance knowledge of the group definitions. In this paper clustering is employed to support efficient decision making by clustering mass storage data available in banks. Risk assessment is important task of Banks, as the failure and success of the Bank depends largely on banks’ ability to evaluate the risk properly. The key problem consists of distinguishing, salubrious (good) and delinquent (bad) customers. Using a novel approach is proposed for risk evaluation of multidimensional risk prediction clustering algorithm (ERPCA). A risk evaluation process is used to determine the good and bad loan applicants. Feature subsets extraction is laid down on the multidimensional data using Information gain technique, to select the valuable attributes. Rules formation is done for each type of loans to avoid redundancy. In order to increase the accuracy of risk computation the risk assessment is performed in two levels, primary and secondary. This method allows for finding percentage of risk to determine whether loan can be sanctioned to a customer or not. This paper mainly concentrates on the clustering of the multidimensional data for efficient risk prediction.