Credit Risk Evaluation using Hybrid Feature Selection Method

Abstract

Fund is the greatest variable of the Banking Industry. In Banking Industry achievement and disappointment depends on the credit. Keeping money Industries are focused today with increment in volume, speed and assortment of new and existing information. Managing and analyzing the massive data is more difficult. The credit scoring databases are often large and characterized by redundant and irrelevant features. In credit sanctions can be classified in view of this credit scoring database. With this features, classification methods become more difficult. This difficulty can be solved by using feature selection methods. In machine learning and statistics the greatest issue is to find the optimal feature subset. In data mining feature selection is a most important data preprocessing step. The main objective of the feature selection is to reduce the size of dimensions, costs and increase the classification accuracy. This research paper uses a hybrid model for finding the optimal feature subset to evaluate the credit risk. It uses the filter and wrapper methods. The proposed model is implemented using WEKA tool. The credit dataset which is taken from UC Irvine machine learning repository is used for evaluation. The classifiers used are Decision Tree (C4.5), Logistic Regression (LR), Support Vector Machines (SVM). Comparison study is made to find the credit risk assessment.