Knowledge Discovery for Banking Risk Management

Economic depression that aggravated in the last decades, led to many banking frauds, attrition and retention disability all over the world. Accordingly, in order to overcome such banking crises, banks policy had been fixated and giving the importance to risk management. To achieve such risk management goals, banks tend to excavate tones of restored transactions in repositories and discover hidden information that affects the decision-making process. Bankers aim to automate the decision-making process to aid and minimize the management risk and enrich bank revenues. This paper presents data preprocessing and feature selection phases entailed in the knowledge discovery process in banking data bases in order to spot the light on the most effective features, which affect the decision making process. In the data preprocessing phase the K-Nearest Neighbor algorithm have been implemented to measure the distance between objects' vectors. The feature selection process has been implemented using Microsoft SQL server analysis services. The results show that the features are reduced from 47 attributes into 17 significant attributes.