An Intelligent Approach to Predict the Traffic Congestion Level Entangled With Machine Learning and Explainable Artificial Intelligence

ABSTRACT In intelligent transportation, the prediction of traffic congestion is still a critical challenge because it is very dependent on several dynamic factors. In this research work, the framework represents an innovative model based on machine learning and explainable artificial intelligence (XAI) approaches to congestion prediction, which not only performs accurate predictions but also provides explanations. The proposed SHapley Additive exPlanations and Local Interpretable Model‐agnostic Explanations–powered model based on the random forest performed better and obtained an accuracy of 99.5%, a precision of 99.90%, a recall of 99.99%, and an F1‐score of 99.94% on the validation data. The addition of XAI helped us gain insight into the effects that the presence or absence of features such as road occupancy, accident reports, weather conditions, and traffic signals had on the predictions, which made the projections not only accurate but also interpretable. This research work is novel because it has a dual focus on predictive robustness and interpretability by providing transportation authorities with a fact‐based, reliable, and operational instrument as part of the solution to congestion issues.