CLICKSENTRY: A HYBRID MACHINE LEARNING AND RULE-BASED SYSTEM FOR CLICK FRAUD DETECTION IN MOBILE APPLICATIONS

Click fraud is a serious problem in the online advertising industry, especially for mobile and web applications that make money through pay-per-click (PPC) models. Fraudulent clicks often come from automated bots, harmful scripts, or fake user actions, causing large financial losses for advertisers and damaging the trustworthiness of online ad platforms. The main detection methods rely on server-side analysis and may not accurately identify widespread and advanced fraudulent activities due to limited insight into real-time user interactions. This paper introduces Click Sentry, a hybrid click fraud detection system that combines machine learning techniques with rule-based URL validation to improve accuracy and reliability in detection. The proposed system uses behavioral analysis by inspecting patterns of user interaction like click frequency along with structural analysis of URLs using keyword-based filtering and regular expression validation The hybrid approach makes sure that malformed suspicious or anomalous URLs are flagged immediately through rulebased logic while machine learning models provide predictive capabilities for identifying complex fraud patterns based on historical data. The system was implemented as a web-based application using the Flask framework which provides an interactive user-friendly interface that supports user registration authentication and real-time fraud detection. An administrator module was also added to allow secure monitoring and management of user activities ensuring controlled access to sensitive data. Role-based access control integration also increases security usability in this system Experimental evaluation shows how well this system can tell apart real clicks from fake ones by using both certain and chance-based detection methods The results show better detection performance than using either method alone with fewer missed detection and greater strength against strange URL patterns and unusual click behavior. A web-based interface was developed using Flask, which allows users to register, login, and do real-time fraud detection. There is also an admin-controlled module for secure monitoring of user activities. The suggested answer gives a flexible and effective setup for dealing with click cheating in today’s ad systems. By mixing machine learning with rule-based checks, Click Sentry presents a hands-on and flexible method that could be used in real-world settings. Future upgrades might include using large-scale data, real-time traffic checks, and better deep learning methods to enhance both detection accuracy and system performance.