Vehicular Ad Hoc Networks (VANETs) are vital in improving safety, security, and overall quality of life in robust development of smart cities. VANETs are highly vulnerable to cyberattacks due to high mobility of vehicles and lack of central security control management. Intrusion Detection Systems (IDS) provides first layer of defense, which is mainly dependent on vehicle collaboration model for possible detection of cyber security threats. Unfortunately, many traditional IDS methods have shown poor performance due to corruption, leading to abnormal behavior and reduced effectiveness. The detection of emerging cyber security threats requires Machine Learning (ML) models. This paper presents the comparative performance analysis of ML algorithms to possibly detect cyber security threats in various computing environments with limited resources including embedded systems, IoT systems and VANETS. The work uses two well-known datasets namely benchmarked network dataset i.e.; NSL-KDD and VANET dataset i.e.; Erlangen for detailed evaluation of ML algorithms: Support Vector Machine (SVM), Random Forest (RF), k-Nearest Neighbor (kNN), and Decision Tree (DT), Logistic Regression (LR). The performance evaluation is carried out on key performance measures including precision, recall, F1-score, and efficiency in computations. This paper presents EdgeAnomSift, a framework that divides the NSL-KDD dataset into smaller parts to evaluate model performance under different training conditions in a scalable way. Experiments with the NSL-KDD dataset show that DT and RF provides high rates of accuracy and efficiency in detecting attacks making it a best algorithm. KNN works well with datasets in smaller size but is ineffective for large computing. SVM is not suitable for real-time detection as it requires a lot of computing power. LR is resource-efficient but needs improvements for the detection of rare attacks. The results show that RF and DT are the best models for Intrusion Detection Systems (IDS) in VANET computing. LR and SVM need more improvement. Tests with the Erlangen dataset confirm that DT is the top choice for VANET security, followed by RF. These findings suggest that efficient ML models can help improve cybersecurity in vehicle networks. The study also explores practical uses, such as linking ML-based IDS with security tools like SNORT. It also suggests future research to enhance intrusion detection in dynamic vehicle systems.