MACHINE LEARNING APPROACH FOR CLICK FRAUD DETECTION

Mobile advertising has gained popularity in recent years as a means for publishers to monetize their free applications due to the increase of Internet usage. Click fraud is one of the main concerns in the in-app advertising industry. Click fraud involves online advertisements that have been clicked on. Pay-per-click fraud involves online advertisements that have been clicked on. Advertisements that pay per click typically target potential customers by charging a fee per click. With machine learning as a solution, we designed the system to detect click fraud using naive bayes, xgboost classifier, random forest, decision tree with gradient boosting, extra tree classifier with gradient boosting, and we observed decision tree with gradient boosting outperformed other algorithms with 96.07% accuracy.