Urban traffic congestion has emerged as one of the most pressing challenges in today's metropolitan areas, directly influencing economic productivity, environmental sustainability, and the overall quality of life for urban residents. With the continuous rise in vehicle numbers and the increasing complexity of transportation networks, the ability to detect and predict congestion with high accuracy has become critical for effective traffic management and planning. Although machine learning (ML) models have demonstrated considerable promise in traffic prediction tasks, their performance often lacks contextual depth when trained solely on temporal data, such as timestamps or aggregated historical traffic patterns. This study addresses this gap by investigating the integration of real-time spatiotemporal data, incorporating both spatial features (road location, number of lanes, geographic coordinates, and event information) and temporal features (time-of-day, day-of-week, etc.) to enhance the predictive accuracy of ML models for road traffic congestion. Using a comprehensive dataset collected from a metropolitan city, five different ML models were implemented and evaluated, including Support Vector Machine (SVM), Random Forest (RF), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and a hybrid CNN-RNN architecture. Each model was trained and tested under two scenarios: temporal features only and integrated spatiotemporal features. The results indicate that spatiotemporal integration substantially improves prediction accuracy across most models, with ensemble-based methods such as RF achieving near-perfect classification, and hybrid deep learning architectures demonstrating significant gains compared to their temporal-only counterparts. Statistical significance testing further validated these improvements, reinforcing the value of spatiotemporal enrichment for predictive tasks. The findings underscore that spatial and temporal contextualization of traffic data improves model robustness and provides critical insights for developing intelligent transportation systems (ITS) capable of delivering real-time, adaptive congestion management solutions. This research contributes to the growing knowledge in smart mobility by offering empirical evidence that spatiotemporal data integration is a key driver of accuracy and reliability in ML-based traffic prediction.

Spatiotemporal Data Machine Learning Traffic Congestion Real-Time Prediction Random Forest CNN RNN Hybrid Models Intelligent Transportation Systems