Question 1

Why is feature selection important for anomaly detection in network traffic analytics?

Accepted Answer

High-dimensional data raises computational complexity, increases resource usage, and prolongs execution time. Feature selection reduces dimensionality, helps interpret data, and improves predictive performance for anomaly detection.

Question 2

How does the study use Information Gain to select features?

Accepted Answer

Information Gain ranks and groups features using minimum weight values, selecting relevant and significant subsets for later classifier evaluation.

Question 3

Which classifiers are tested on the CICIDS-2017 dataset, and what performance trends are reported?

Accepted Answer

Random Forest, Bayes Net, Random Tree, Naive Bayes, and J48 are tested. The results show that the number of relevant selected features strongly influences detection accuracy and execution time, with Random Forest reaching the highest accuracy in the reported experiment and J48 achieving a very close accuracy using more features but longer runtime.