Systematic Feature Selection Based on Information Gain in Intrusion Detection Systems

Network traffic has increasing marginally due to the availability of internet used, and cause the overload of dataset, and making data not be understandable. The monitoring of activities on internet using Intrusion Detection Systems (IDS) has been one of essential network infrastructure to ensure the security of internet. This IDS has been implemented based on internet features, therefore some of these features are irrelevant and the correspondents instances are redundant and inconsistent. Feature selection is one of the most important preprocessing stages in data mining and knowledge engineering to overcome the problem of many variables, instances redundancy and inconsistency which make the problem not being approachable. This paper discusses systematic feature selection based on Information Gain to find the relevant subset of features which has effect on targets.