Hybrid Ransomeware Detection using Catboost and Random Forest Algorithm

Abstract

Numerous threats to cybersecurity, such as ransomware, malware, spyware, Wannacry, and Cryptolocker assaults continue to cause significant damage to servers, computer systems, and web applications owned by different organizations across the globe. These safety issues are critical and need to be resolved right away. To ensure prompt response and prevention, ransomware detection and classification are essential. The RF algorithms classifiers and CatBoost feature selection are deployed in this work to identify and categorize ransomware assaults. This method entails examining ransomware behavior and identifying important features that can be applied to distinguish between various malware families. The algorithms' efficiency in precisely identifying and categorizing ransomware is demonstrated when they are tested on a ransomware detection dataset used in this study, which has 62,485 samples overall, was gathered from Kaggle, incidents of ransomware attacks and achieved a result of 99.80% accuracy. These findings indicate that the RF Classifiers and CatBoost classifier can accurately distinguish between various ransomware incidents, thereby providing a useful tool to aid cybersecurity.