An Intelligent Hybrid Supervised Machine Learning Approach for Predicting Vehicular Traffic Congestion
DOI:
https://doi.org/10.51483/IJAIML.2704Keywords:
Vehicular Congestion Forecasting, Hybrid Supervised Classifier, Traffic Prediction, Smart Cities, Intelligent Transportation Systems (ITS), Ensemble Learning, Gradient Boosting, Traffic Flow Analysis, Machine Learning Models, Urban Mobility Management.Abstract
Urban traffic congestion causes delays, fuel consumption, and environmental problems, highlighting the need for intelligent transportation solutions. This study develops an automated supervised learning framework for real-time congestion detection and short-term prediction. Traffic data from sensors, GPS, cameras, weather, and public events are collected and processed using cleaning, feature engineering, normalization, and SMOTE-based balancing. Random Forest, SVM, XGBoost, and LightGBM are evaluated using accuracy, precision, recall, and F1-score. Results show that ensemble and hybrid models outperform individual classifiers and effectively categorize congestion as Low, Normal, High, or Heavy. The proposed framework supports reliable, proactive, and data-driven traffic management.
