Comparative Analysis of Machine Learning, Deep Learning, Statistical Models on Credit Risk Prediction

This study uses a dataset with several borrower characteristics to forecast credit risk using Machine learning (ML) and Deep Learning (DL) techniques in addition to statistical methods. With a 98.1% accuracy, the LGBM classifier outperformed the other models in this investigation and demonstrated the capacity to identify intricate risk patterns. The other machine learning models also shown outstanding performance; however, the statistical and deep learning frameworks were unable to achieve the same accuracy levels as machine learning methods. Sound lending practices and efficient risk management are enhanced by this multidimensional approach, which provides a wealth of information about the borrower's profile.