Fraud detection is a critical application of machine learning in the financial sector. It aims to find out fraud transactions with minimal impact on legitimate transactions. This project uses logistic regression, one of the most popular classification algorithms, to detect fraudulent transactions in a highly imbalanced credit card transaction dataset. This data contains 284,807 transactions, of which only 492 are fraudulent (Class 1), which accounts for less than 0.2% of the total data. This dataset had a highly imbalanced distribution between the classes, with only a tiny fraud class, which usually creates issues in developing an effective predictive model. This project will utilize various techniques, including SMOTE and threshold tuning of the decision, for better model performance and a balanced trade-off between precision and recall for the minority class.

My Presentation Slides: https://docs.google.com/presentation/d/1bi7d2JTGOMlSiqOvATWxqLU1O0JjOdXK/edit?usp=sharing&ouid=116042308112205426978&rtpof=true&sd=true

References: • Dataset: Credit Card Fraud Detection Dataset (Kaggle) • SMOTE Reference: Chawla, N.V., Bowyer, K.W., Hall, L.O., & Kegelmeyer, W.P. (2002). Synthetic Minority Over-sampling Technique. • Logistic Regression: Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning.

Acknowledgements: The Austin Peay State University’s Office of Student Research Innovation Mathematical Association of America’s Southeastern Conference High Point University, North Carolina Dr. Daniel Mayo - My Machine Learning Professor Dr. Ramanjit Sahi - My Graduate Coordinator Dr. Jackie Vogel - The Chair of the Math Department at APSU Ms. Jaime Gaither - The Assistant of the Chair of the Math Department at APSU My Family and colleagues, I could not have done this without your support.