Credit Risk Analysis

Authors

  • Sudhir Kumar Pandey, Ashwin Kumar U M

Abstract

score cards. Due to its desirable features (robustness and transparency) logistic regression model is among them the most widely used in the banking industry.Although some modern techniques (support vector machine) were applied to credit scoring and showed superior predictive accuracy, they have problems with interpretability of the results. Therefore, those specialized methods were not commonly used in practice.Logistic regression with random coefficients is suggested to improve predictive accuracy of logistical regression.The proposed model will boost logistic regression prediction accuracy without sacrificing desirable features.The proposed method of developing the credit scorecard is expected to lead to successful credit risk management in practice.

Keywords: logistic regression, Exploratory Data Analysis WOE and information value, DiscreditingPredictors/Binning.

Downloads

Published

2020-05-12

Issue

Section

Articles