• Mithun J. Sharma

In the traditional banking business of advancing backed by customer deposits, Public Banks face the risk of debtor evasion in the repayment of either principal or interest. This risk is referred to as "Credit Risk" in lending terminology, and financial records where the payment of principal or interest is not imminent are referred to as "Non-Performing Assets". The existence of Non-Performing Assets (NPAs) is an essential component of finance, and every bank has some NPAs in its loan portfolio. However, any financial institution should be concerned about the high level of NPA. This research was carried out to create machine learning models that could predict the likelihood of bad loans. This study was conducted to develop machine learning models, specifically Artificial Neural Networks (ANN) that include Probabilistic Neural Nets (PNN) with traditional statistical model such as Multivariate Probit Analysis (MPA). The results of the ANN model, which included PNN, showed that all loans were correctly classified. The MPA achieving 98.50%. We discovered that the MPA model would be a perform better in those situations where higher false positive results are ideal. The average score of the two models is 99.25 percent. The main contribution of this study is the empirical nature of the problem and significant correct classification rate of the models for those loan portfolios after sanctioning of the loans. Bank managers can effectively avoid information overload with the models empirically tested in this project, paving the way for proper identification and prevention of bad loans.

Artificial Neural Network (ANN), Multivariate Probit Analysis (MPA), Bad Loans, Banks

"Multivariate Probit Analysis (MPA) and Artificial Neural Nets (ANN) Models for Predicting Bank Loan Portfolio Default Probability", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 2, page no.e45-e51, February-2022, Available :http://www.jetir.org/papers/JETIR2202407.pdf

"Multivariate Probit Analysis (MPA) and Artificial Neural Nets (ANN) Models for Predicting Bank Loan Portfolio Default Probability", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 2, page no. ppe45-e51, February-2022, Available at : http://www.jetir.org/papers/JETIR2202407.pdf