Abstract:A credit default is a negative event for individual borrowers, and lenders and at scale, can impact the broader economy. To protect against losses, lenders seek to maintain robust lending practices through optimising credit risk assessment. The credit risk modelling literature is extensive, yet many studies suffer from flaws stifling adoption by industry. Nonlinear algorithms are often incorporated for comparison purposes, nevertheless, a comprehensive explanation of variables and their treatment is not always provided. Furthermore, the publicly available data sets do not necessarily contain important financial variables and their limited sample size raises questions regarding their statistical power. Commonly the focus is on statistical performance as opposed to optimizing the commercial outcome motivating the use of credit scorecards. However, financial institutions are more interested in optimizing profits by reducing the cost arising from false assessments of credit applicants. In this research, we simulated important financial variables recommended by industry to append to German Credit Dataset to compare optimised credit risk assessment models. The performance of Logistic Regression is compared with Random Forest, Gradient Boosting Machine and Artificial Neural Network. A new financial performance metric is developed to estimate the costs arising from both false positives and false negatives, and compared with statistical criteria; Area under the Receiver Operating Curve, Accuracy and Somers' Delta. Four typical asset classes; Credit Cards, Small and Large Loans, and Mortgages are simulated to determine which algorithm outperforms in estimating costs. The results demonstrate that Artificial Neural Network outperforms other methods, and our approach ranks models differently from the purely statistical measures. The proposed cost metric would provide a significant financial benefit to organisations.

Personal Credit Scoring via Logistic Regression with Elastic Net Penalty

Application Analysis of Credit Scoring of Financial Institutions Based on Machine Learning Model

An implementation of ensemble methods, logistic regression, and neural network for default prediction in Peer-to-Peer lending

An Overview of Personal Credit Scoring: Techniques and Future Work

Transfer Learning Using Logistic Regression in Credit Scoring

How Do Machine Learning and Non-Traditional Data Affect Credit Scoring? New Evidence from a Chinese Fintech Firm

A ResNet-LSTM Based Credit Scoring Approach for Imbalanced Data

Research on Credit Big Data Algorithm Based on Logistic Regression

Investigating the beneficial impact of segmentation-based modelling for credit scoring

Personal Credit Risk Evaluation of Commercial Banks Based on Logit Model

Empirical Evaluation of Ensemble Learning for Credit Scoring

Credit Fraud Detection Based on Hybrid Credit Scoring Model

An Interpretable Personal Credit Evaluation Model

Research on credit risk assessment optimization based on machine learning

Research on Personal Credit Evaluation Based on PCA-Logit Model

Prediction and Analysis of Financial Default Loan Behavior Based on Machine Learning Model

An Integrated Machine Learning and Deep Learning Framework for Credit Card Approval Prediction

Optimal Credit Scorecard Model Selection Using Costs Arising from Both False Positives and False Negatives

Statistical and machine learning models in credit scoring: A systematic literature survey

Application of Machine Learning in Credit Risk Scorecard

A Novel Multi-Stage Ensemble Model With a Hybrid Genetic Algorithm for Credit Scoring on Imbalanced Data