CLICK FRAUD DETECTION USING MACHINE LEARNING

Abstract

As online advertising continues to grow, click fraud has emerged as a critical challenge, undermining the efficiency of advertising campaigns and inflating costs for advertisers. Detecting and mitigating click fraud is, therefore, a vital step in optimizing online advertising efforts. This research focuses on leveraging machine learning models to detect click fraud using features such as IP address, app, device, operating system, and channel. The study explores several machine learning algorithms, including Logistic Regression, Random Forest, SVM, Decision Tree, K Nearest Neighbors, Gradient Boosted Trees, and XGBoost. These models are evaluated based on key metrics such as accuracy, precision, recall, and F1-score. Among the tested models, XGBoost demonstrated the highest performance with an accuracy of 92.12%, highlighting its potential for accurate and efficient fraud detection. This finding underscores the importance of advanced machine learning techniques in combating click fraud, paving the way for more reliable and cost-effective online advertising strategies.