Traffic prediction plays a key role in the management of smart cities and vehicular networks. Accurate forecasting helps improve resource allocation, reduce congestion, and improve overall network performance. Many machine learning(ML) and deep learning(DL) techniques have been developed to predict traffic using data from Vehicle-to-Vehicle (V2V) and Vehicle-to-Roadside Unit (V2R) communications, GPS devices, sensors, and network logs. This paper provides a comprehensive review of various ML methods for traffic prediction. Covers traditional algorithms such as K-Nearest Neighbors(KNN), Gaussian Naive Bayes (Gaussian NB), Support Vector Machines(SVM), Random Forests (RF), and Gene Expression Programming (GEP). Additionally, it explores advanced models like Long Short-Term Memory (LSTM),Gated Recurrent Units (GRU), CNN-LSTM hybrids, XGBoost, LightGBM, and Graph Neural Networks (GNN). We discuss their predictive accuracy, computational efficiency, and how well they handle different types of data. Our review highlights that combining feature selection techniques, such as RF, with time-series models like GRU or LSTM often results in the best prediction accuracy. In addition, incorporating road traffic data from V2V communication along with network traffic data from V2R communication improves prediction performance, especially in complex urban traffic environments.