Network traffic is growing exponentially with the increase in use of smart devices and the internet. Commensurately, the chances of network intrusion are also growing. Processing massive data in near real time is the bottleneck in the performance of intrusion detection systems. If the dimension of data in use could be reduced to including only those features which are significantly important for intrusion detection, this can help in increasing the performance of intrusion detection systems. This is where feature selection (FS) techniques play an important role because it provides the classifiers to be fast, cost-effective, and more accurate. In this paper, three feature selection methods are analyzed; FI (Feature Importance), RFE (Recursive Feature Elimination), ANOVA (Analysis of variance). Features from these FS methods are then learned using machine learning models; Random Forest (RF) and Multi-Layer Perceptron (MLP). Later, a comparative analysis of the accuracies and ROC curves for each method is done.