Advertisement Click Fraud Detection Using Machine Learning Techniques

Web browsing has become the integral part of our daily activities. Users come across various advertisement while accessing these web pages. Various companies advertise their products on web pages to increase their sales. The companies hire other Advertisement Agencies for the task of publishing the advertisement. The Advertisement Agency charges the company for clicks made on the Advertisement links using pay-per-click model. Few Advertisements Agency indulge in malpractices to increase their profit by manipulating the number of clicks, this leads to financial loss to the company advertising their products. Click fraud detection helps in identifying the genuine clicks made by the user and the counterfeit clicks made using click bots or other software. In this paper, a comparison is conducted between different machine learning algorithms- Logistic Regression, RNN and various Boosting Algorithm on the basis of accuracy, precision, recall rate, f1-score, subsample, n_estimator, max depth and Learning Rate. The system was experimented on TalkingData AdTracking Fraud Detection Dataset from Kaggle. After experimental analysis it is observed that Logistic Regression shows superiority over the models in comparison.