Building a Scoring Model Using the Adaboost Ensemble Model

Abstract

In paper presents and describes modern methods of data analysis, used when using credit scoring. The obtained results of the model allow us to conclude that the classification of borrowers by credit rating can be effectively solved using the machine learning algorithm Adaboost than with the use of Gradient boosting and the standard model of logistic regression even before setting up hyperparameters. Machine learning methods were applied in the work. A correlation analysis of the data was performed to exclude interrelated predictors. The AUC and GINI values of the AdaBoost method were calculated, which show the high efficiency of the model.