Building predictive models to assess credit risks in banks Case study of public and private banks

Abstract

This research aims to test the ability of credit risk prediction models facing public and private banks, by developing a model whose inputs depend on a set of ratios and financial indicators obtained from a banksampleof (20) banks listed on theIraqstock market over the course of ten years from 2013-2022 with a total of 196 observations. Neural models show the accuracy of high prediction compared to other statistical models, that points to their superiority in comprehending complex relationships between financial variables and improving their predictive power, which contributes to achieve the highest levels of accuracy and effectiveness. The use of multi-layered neural networks enables deep analysis of financial data, which reveals non-linear relationships and complex interactions between variables that enhanceits ability to predict credit risk with greater accuracy. Thus, the role of multi-layered neural networks is evident in improving the ability of financial indicators to predict credit risks for banks. These conclusions are considered an important addition to the available knowledge about the role of neural networks in the field of improving credit risk prediction and developing the banksfinancial performance.