EVALUATION OF CLASSIFICATION ALGORITHMS ON LOCKY RANSOMWARE USING WEKA TOOL

Abstract

The ongoing danger of ransomware has led to a struggle between creating and identifying novel approaches. Although detection and mitigation systems have been created and are used widely, they are always evolving and being updated due to their reactive nature. This is because harmful code and its behavior can frequently be altered to evade detection methods. In this study, we present a classification method that combines static and dynamic data to improve the precision of locky ransomware detection and classification. We trained supervised machine learning algorithms using cross-validation and used a confusion matrix to observe accuracy, enabling a systematic comparison of each algorithm. In this work, supervised algorithms such as the decision tree algorithm resulted in an accuracy of 97%, naïve baiyes  95%, random tree 63%, and ZeorR 50%.