An Efficient Network Intrusion Based on Decision Tree Classifier&K-Mean Clustering using Dimensionality Reduction

As the internet size grows rapidly so that the attacks on network. There is a need of intrusion detection system (IDS) but large and increasing size of network creates huge computational values which can be a problem in estimating data mining results this problem can be overcome using dimensionality reduction as a part of data preprocessing. In this paper we study two decision tree classifiers (J48, Id3) for the purpose of detecting any intrusion and comparing their performances. first we have applied data pre processing steps on each classifier which includes feature selection using attribute selection filter, Intrusion detection dataset is KDDCUP 99 dataset which has 42 features after preprocessing 9 selected attributes remains, then discretization of selected attribute is performed, simple k-Mean algorithm is used for analysis of data and Based on this study, we have concluded that J48 has higher classification accuracy with high true positive rate (TPR) and low false positive rate (FPR) as compared to ID3 decision tree classifiers.