Feature Selection and Scaling for Random Forest Powered Malware Detection System
Ashutosh Kumar Tripathi, Naman Bhoj, Mayank Khari, Bishwajeet Pandey · Research Square · 2021
Abstract With the rise of internet usage malwares pose a great threat to user security and privacy. Therefore, to mitigate the problem it is essential to develop an efficient malware detection framework. In our research we experimented with various machine learning and feature scaling algorithms. Chi-Square was used as the feature selection technique which selected a set of 48 features from a feature space of 128 features. The empirical results provide us with conclusive evidence that Random Forest is the best algorithm for detection of malware achieving an accuracy of 91.300% on our dataset followed by Gradient Boosting and Support Vector Machines.