Credit Risk Customers Categorization with Random Forest Classifier using Various Searching Techniques

Effective credit risk assessment holds significant importance for financial institutions and businesses extending credit to customers. It is crucial to evaluate customers’ creditworthiness to minimize default risks and manage financial exposure. This study focuses on enhancing credit risk customer classification through optimized feature selection and classification techniques. The dataset is preprocessed using Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance and enhance performance. Relevant features are then extracted utilizing Cfs Subset evaluation along with various search algorithms. The optimized features are subsequently employed for classification using the Random Forest (RF) classifier. Experimental results reveal that, in credit customer classification, Evolutionary, Tabu, and Scatter search algorithms combined with RF classifier demonstrate a commendable accuracy of 90.31%. The proposed framework also showcases comparable outcomes in terms of other performance metrics. This automated credit customer classification method proves to be highly effective in identifying fraudulent users.