A nature-inspired osprey optimization based feature selection framework with stacking ensemble model for...

Area of Science:

• Data Science
• Machine Learning
• Computational Intelligence

Background:

• Credit card fraud detection is challenging due to imbalanced transaction data and irrelevant features.
• Model performance is significantly impacted by feature selection and class imbalance.

Purpose of the Study:

• To propose an ideal feature selection model for credit card fraud detection.
• To enhance fraud detection performance by addressing class imbalance and feature redundancy.

Main Methods:

• Developed a feature selection model integrating the Osprey Optimization Algorithm with a stacking ensemble classifier.
• Employed a stacking ensemble of Naive Bayes, Decision Tree, Support Vector Machine, and Random Forest as base learners, with Logistic Regression as the meta-learner.
• Utilized a real-world credit card fraud dataset for experimental validation.

Main Results:

• The proposed approach significantly enhances fraud detection performance metrics, including precision, recall, F1-score, and Matthews correlation coefficient.
• Effectively reduces feature dimensionality while maintaining or improving classification accuracy.
• Demonstrates superior capability in handling imbalanced financial datasets compared to existing methods.

Conclusions:

• The Osprey Optimization Algorithm combined with a stacking ensemble classifier offers an efficient solution for credit card fraud detection.
• The method successfully addresses the critical issues of class imbalance and feature redundancy in financial datasets.
• This advanced algorithm proves effective in improving the accuracy and reliability of fraud detection systems.