Algorithmic Loan Risk Prediction Method Based on PSO-EBGWO-Catboost

Abstract

Loan risk analysis is a common challenge faced by global financial institutions. Under the background of big data, it is of practical significance to prevent loan risk by the machine learning algorithm. Aiming at the characteristics of unbalanced loan data and high noise, this paper proposes an improved Gray Wolf optimization strategy (PSO-EBGWO). PSO-EBGWO is used to optimize the parameters of the CatBoost model. In this method, the Gray Wolf optimized algorithm (EBGWO) is further optimized by particle swarm optimization (PSO), and when combined with it, the convergence performance of the model is improved, the parameters of the model are reduced, and the model is simplified. To a certain extent, it avoids the inefficiency of the Gray Wolf algorithm, balances the ability of local search and global development, and improves the accuracy of the model. Compared with the traditional credit evaluation model, PSO-EBGWO-CatBoost has better accuracy and practical application value.