Banking Credit Scoring Assessment Using Predictive K-Nearest Neighbour (PKNN) Classifier

Banking Credit Scoring Assessment Using Predictive K-Nearest Neighbour (PKNN) Classifier

Saroj Kanta Jena (BML Munjal University, India), Anil Kumar (BML Munjal University, India) and Maheshwar Dwivedy (BML Munjal University, India)
DOI: 10.4018/978-1-5225-0997-4.ch018

Abstract

Credit scoring models is a scientific methodology adopted by credit providers to assess the credit worthiness of applicants. The primary objective of such models has been to predict the potentiality of the loan applicant. A proper evaluation of the credit can help the service provider to determine whether to grant or to reject credit. Therefore, the objective of the study is to predict banking credit scoring assessment using Predictive K-Nearest Neighbour (PKNN) classifier. For the purpose of analysis two different credit approval datasets: Australian credit and German credit have been used. The results from the study show that the proposed model used for classification works better on credit dataset. Here, the study firstly attempted to find the optimal ‘K' value of the neighbourhood so that the classifier is tuned to forecast the credit worthiness and secondly, validated our proposed model on two credit approval datasets by checking the performance of the proposed models on the basis of classification accuracy.
Chapter Preview
Top

2. Literature Review

Corporate bankruptcies and their prediction remains an important and widely investigated topic even today. Such studies are found to have substantial impact on the entire process of loan approval decisions and profitability. Atiya (2001) after reviewing several Neural Network (NN) models proposed a NN bankruptcy forecast model for bankruptcy prediction. MERTON (1974) is reported to have considered several traditional models for credit risk to propose innovative indicators for the NN system. Similarly, Carling et al. (2001) reported on the different hazard (risk) associated with the loan approval process of the bank. The authors in their study considered the risk of non-payment and the risk of early recovery of loans.

Complete Chapter List

Search this Book:
Reset