Using data mining techniques to improve the detection of credit card fraud

The financial industry is constantly under threat in the fight against financial fraud, requiring for strong protection. Data mining emerges as a crucial technique for detecting credit card fraud in online transactions. Detecting credit card fraud proves challenging due to the dynamic nature of legitimate and fraudulent behavior patterns and the inherent skewness in credit card datasets. The accuracy of credit card fraud detection hinges on critical factors like the sampling method of the dataset, the selection of significant variables, and the chosen detection method. This project aims to tackle these challenges by exploring various machine learning models (Random Forest, Logistic Regression, Decision Tree, Naïve Bayes) designed to predict fraudulent activity in credit card transactions. Python will serve as the primary programming language due to its extensive libraries and features for machine learning and data analysis. The objective is to enhance the precision and effectiveness of credit card fraud detection systems, contributing to the ongoing fight against financial fraud. Among the machine learning models assessed, Random Forest stands out with the highest accuracy, achieving 97%, surpassing other models in performance metrics. Our research emphasizes the development and evaluation of predictive models adaptable to the evolving nature of fraud, providing valuable insights for both financial institutions and customers. Keywords: Data Mining, Credit Card, Fraud Detection.