Linear Correlation-Based Feature Selection for Network Intrusion Detection Model

Feature selection is a preprocessing phase to machine learning, which leads to increase the classification accuracy and reduce its complexity. However, the increase of data dimensionality poses a challenge to many existing feature selection methods. This paper formulates and validates a method for selecting optimal feature subset based on the analysis of the Pearson correlation coefficients. We adopt the correlation analysis between two variables as a feature goodness measure. Where, a feature is good if it is highly correlated to the class and is low correlated to the other features. To evaluate the proposed Feature selection method, experiments are applied on NSL-KDD dataset. The experiments shows that, the number of features is reduced from 41 to 17 features, which leads to improve the classification accuracy to 99.1%. Also,The efficiency of the proposed linear correlation feature selection method is demonstrated through extensive comparisons with other well known feature selection methods.