Abstract
Credit risk assessment is an essential element of financial stability, impacting banking institutions, governmental financial entities, and individual creditworthiness. This paper presents a machine learning-based credit score model aimed at improving the precision of financial risk evaluations through the use of predictive analytics. The model employs sophisticated algorithms, including Random Forest, XGBoost, and Neural Networks, to categorize individuals into credit risk classifications based on their transaction histories, loan repayment patterns, and income stability. Explainable AI methodologies, such as Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP), enhance transparency in decision-making, hence augmenting trust in credit assessments. The research analyzes the effects of economic volatility and regulatory modifications on credit risk evaluation, specifically in Kazakhstan, where there has been a notable increase in consumer loans. Empirical study and thorough experimentation reveal that machine learning models surpass traditional statistical methods, enhancing predicted accuracy and reducing biases in credit assessment. The proposed approach improves financial inclusion by offering a scalable and transparent method for real-time evaluations of creditworthiness. This research advances the creation of adaptable and robust credit risk models that respond to changing financial environments.