Comparative Analysis of SVM, Logistic Regression, Random Forest, and XGBoost for Credit Risk Assessment

With the evolving world of financial technology, accurate credit risk prediction has become crucial to minimize loan defaults and enhance financial inclusion. Traditional credit assessment relies primarily on historical credit data, rendering credit unaffordable for individuals with thin credit files. This research attempts to bridge this gap by conceptualizing and comparing different machine learning algorithms for loan approval prediction. The study contrasts Support Vector Machines (SVM), Logistic Regression, Random Forest, and Extreme Gradient Boosting (XGBoost) performance in the loan applicant classification task with demographic and financial features. The methodology includes data preprocessing, feature engineering, and unsupervised clustering to identify underlying patterns in the data prior to training. This is followed by model training and testing on the basis of key performance measures such as accuracy, precision, recall, and F1-score. By comparing the performance of models in varying conditions, the paper demonstrates the strengths and weaknesses of each approach. The results are intended to build more reliable, data-driven decision systems that enhance the more equitable and faster loan approval process and reduce the risk of defaults.