Nowadays, as technology keeps advancing, people are becoming more tech-savvy, doing more transactions online, and the threat of fraudulent actions is high. Understanding which model performs best on imbalanced fraud datasets is essential to ensuring effective fraud detection systems. Furthermore, defining effective strategies for handling such data and using relevant pieces of information about each transaction is crucial for enhancing the accuracy and reliability of the detection mechanisms for online banking services. This research not only strengthens our ability to prevent financial fraud but also safeguards the integrity of digital transactions, bolstering trust and security in the global financial ecosystem.
- Determine the most effective model for detecting fraud within an imbalanced dataset
- Identify the features that make a substantial impact on the model's overall performance
- Outline effective strategies for addressing data imbalance tailored to the nature of fraudulent transaction datasets
pandas, matplotlib, seaborn, plotly, scikit-learn, Synthetic Minority Oversampling Technique (SMOTE)