Data Driven-Machine Learning-Based Fraud Detection Models in FinTech Financial Transactions

Fraud examination recognizes and combats improper economic activities by applying data analysis methods and machine learning for the purpose of improving safety and reducing financial threats. Financial fraud plays a big part in the FinTech sector and impacts many people and businesses all around the world. The development of automated fraud detection systems is ongoing, but reliably and rapidly identifying fraudulent transactions is still a significant challenge. The credit card fraud detection dataset from Kaggle is utilized in this study to train the XGBoost model. The dataset receives in-depth preprocessing treatment where researchers perform missing value imputation and handle outliers alongside class imbalance through the implementation of SMOTE. The XGBoost model is evaluated using a variety of significant performance metrics, including accuracy, precision, recall, and AUC. XGBoost delivers superior outcomes than traditional Decision Tree classifiers as well as SVM and Logistic Regression by reaching 99.32% accuracy alongside 99.31% precision and 99.42% recall and 99.7% AUC. The model demonstrates exceptional capability to detect fraudulent transactions because of its superior accuracy rates. The examined research creates stronger financial fraud detection methods for digital systems to protect secure transactions and minimize monetary damages from fraudulent operations.