Scholarlink

This study addresses the challenges of loan default prediction and credit card fraud detection, crucial for financial stability and consumer trust. It evaluates the performance of Support Vector Machine (SVM) and Naive Bayes models, focusing on feature selection using the Artificial Bee Colony (ABC) algorithm. The ABC approach is applied to the Kaggle European cardholder dataset for feature reduction, followed by categorization of fraudulent transactions using Naive Bayes and SVM. The SMOTE method is used to handle the highly unbalanced dataset. Performance is measured using precision, recall, f-measure, and accuracy. The results show that the ABC-SVM algorithm achieved 95.4% accuracy, while ABC-NB reached 89.3%, outperforming traditional and population-based feature selection methods.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Evaluation of Machine Learning Models on Loan Default Prediction Using Artificial Bee Colony for Feature Selection

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Asaju La’aro Bolaji,
  • Ebenezer Etim Ekpo,
  • Andrew Ishaku Wreford

摘要

This study addresses the challenges of loan default prediction and credit card fraud detection, crucial for financial stability and consumer trust. It evaluates the performance of Support Vector Machine (SVM) and Naive Bayes models, focusing on feature selection using the Artificial Bee Colony (ABC) algorithm. The ABC approach is applied to the Kaggle European cardholder dataset for feature reduction, followed by categorization of fraudulent transactions using Naive Bayes and SVM. The SMOTE method is used to handle the highly unbalanced dataset. Performance is measured using precision, recall, f-measure, and accuracy. The results show that the ABC-SVM algorithm achieved 95.4% accuracy, while ABC-NB reached 89.3%, outperforming traditional and population-based feature selection methods.