Personal credit risk assessment based on stacking ensemble model

Abstract

Nowadays, compared with the traditional artificial risk control audit method, a more sophisticated intelligent risk assessment system is needed to support credit risk assessment. Ensemble learning combines multiple learners that can achieve better generalization performance than a single model. This paper proposes a personal credit risk assessment model based on Stacking ensemble learning. The model uses different training subsets and feature sampling and parameter perturbation methods to train multiple differentiated XGBoost classifiers. According to Xgboost’s high accuracy and susceptibility to disturbances, it is used as a base learner to guarantee every learning “Good and different”. Logistic regression is then used as a secondary learner to learn the results obtained by Xgboost, thereby constructing an evaluation model. Using the German credit data set published by UCI to verify this model and Compared with the single model and Bagging ensemble model, it is proved that the Stacking learning strategy has better generalization ability.