A Novel Explainable Hybrid Ensemble ML Classification for Digital Risks
Open Access DepositedAs digital adoption increased, the number of digital fraud risks skyrocketed. Every year, digital fraud costs corporations across industries billions of dollars in losses. Governments and organizations have been consistently working to reduce these losses through legislation, regulations, reviews, audits, policies, processes, and the use of cutting-edge tools and techniques. In the age of Artificial Intelligence (AI), when many organizations have already adopted modern machine learning (ML) solutions, these efforts are deemed inadequate, and the need for continued research and development persists. This praxis presents a novel Explainable Hybrid Ensemble Machine Learning (XHEML), that combines the strategies from stacking and bagging ensembles, utilizes heuristic weights with modified Harris Hawk Optimization (DART-HHO), and performs weighted soft voting fusion to classify digital fraud risks efficiently. The model achieved a mean accuracy of 82.06% as validated in the experimental setup. The XHEML achieved a higher mean F1 score compared to the Stacking Logistic Regression (LR) and Bagging-Random Forest (RF). The proposed model also outperformed the Decision Tree (DT) in terms of mean F1 score. The XHEML model researched in this praxis will serve as an asset to business and engineering managers for effectively identifying digital fraud risks and explaining the results, thereby focusing efforts on reducing financial losses.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.