BUILDING A MODEL BASED ON MACHINE LEARNING METHODS FOR PREDICTING THE CREDITWORTHINESS OF CUSTOMERS

Abstract

A customer's credit rating is important for financial institutions, as lending can result in real and immediate losses. Scoring models are increasingly used in modern financial technologies and serve professionals to improve their efficiency. They are superior in their capabilities to the subjective assessments of  people, as they are not subject to professional bias and cognitive distortions. In this article, we will focus on building machine learning models to predict  customer creditworthiness. The main goal is to identify the most important factors that will help calculate the creditworthiness of customers, analyze customer characteristics. We will build Support Vector Machine(SVM), Decision Trees(DTs), Xgboost and Random Forest models and explore comparative analysis of their predictive accuracy.