Route Analysis And Traffic Jam Prediction Using Deep Learning Techniques: An Enhanced LSTM-GRU Comparative Study

Traffic route analysis plays a critical role in daily urban mobility. This study proposes and evaluates deep learning models for predicting traffic jam status based on contextual weather and event data. Specifically, LSTM and GRU architectures are trained and tested on a dataset of weather conditions, air quality, road characteristics, and events. This study introduces a structured hyperparameter optimization framework and an interpretive analysis explaining observed performance gaps. Data preprocessing includes one-hot encoding, Min-Max normalization, and temporal sequence organization. Both models are evaluated using accuracy, precision, recall, F1-score, and ROC AUC. The LSTM model achieves 87.5% accuracy and 91.67% ROC AUC, significantly outperforming the GRU model (50% accuracy, 67% ROC AUC). The findings offer practical guidance for selecting appropriate recurrent architectures in traffic prediction systems. Traffic prediction, deep learning, LSTM, GRU, weather data, event data, traffic congestion, performance evaluation, hyperparameter optimization.