Click Ad Fraud Detection Using XGBoost Gradient Boosting Algorithm

Abstract

The growth of the online advertising industry has created new business opportunities on the Internet. Companies and advertisers are turning to digital ad platforms like never before to compete for the attention of their audience. In this environment, actions such as clicking an ad result in financial transactions among advertisers, advertising networks and publishers. Since these new opportunities have financial impact, fraudsters have been trying to gain illegal advantages and profit through them. Mitigating the negative effects of illegal traffic is extremely important to the success of any marketing endeavor. Today, false clicks that waste budgets and don’t generate any meaningful value or revenue are costing advertisers billions of dollars. These are the biggest challenge PPC (Pay Per Click) marketers face, although there are efforts made by the advertisers to block fake traffic, they still try to find leading security strategy to identify click fraud. This paper analyzes the click fraud mechanism, focusing on its detection and methods of solution used in recent cases, we try and explain various fundamentals related to online advertising. The objective of this research is to propose solution for click ad fraud present in online advertising using the XGBoost Gradient Boosting algorithm and this model provides the accuracy of 96% with a set of hyperparameters along with features that can be implemented on datasets related to click frauds.