Analysing and Predicting Loan-Default Possibility Using Machine Learning Models

In this paper, the loan-default risk has been calculated using machine learning. It suggests and compares XGBoost, Logistic Regression, Gradient Boosting, and Random Forest models according to their predictive performances on the “Bank Indessa” dataset [1]. The data set is completely preprocessed. In that process, all missing values are replaced with appropriate values and categorical variables are encoded. Among the tested models, the XGBoost model has been proved to give the most accurate and predictive performances. This paper highlights the efficiency of PFR models and the significance of feature engineering in enhancing model performance. It provides relevant insights into using ML models for Financial Risk Assessment, thereby making credit scoring systems faster and more precise[2].