In today’s interconnected digital environment, the rise of network-based attacks presents significant risks to individuals and organizations. Conventional security protocols have struggled to keep pace with the increasing complexity of cyber threats. In this study, we delve into the domain of cybersecurity with the aim of leveraging Artificial Intelligence (AI) methodologies such as Machine Learning (ML) and anomaly detection to fortify network defense mechanisms. We make use of different datasets, feature selection techniques, and ML algorithms, in order to compare them and decide which ML algorithm and feature selection technique help reach the best Network Intrusion Detection System (NIDS). Utilizing two distinct datasets, CICIDS2017 and CICIoT2023, the study investigates the efficacy of different ML algorithms, namely Decision Tree (DT), Naïve Bayes (NB), K-Nearest Neighbor (KNN), Logistic Regression (LR), Adaboost and Xgboost and feature selection techniques, namely Random Forest (RF) and Principal Component Analysis (PCA), in enhancing the performance of Intrusion Detection Systems (IDSs). Results show that DT and Xgboost outperform other algorithms in binary classification across both datasets and all feature selection techniques, and that DT with RF is the optimal option for multi-class classification.