Traffic Congestion Prediction Using Machine Learning Techniques

2022-06-22

arXiv.org

AI Summary

To address the challenge of medium- to long-term urban traffic congestion forecasting, this paper proposes a supervised machine learning model that jointly encodes spatiotemporal features and multidimensional meteorological variables (e.g., temperature, humidity) to enable up to seven-day forward-looking congestion predictions. Unlike prior approaches, our work is the first to systematically integrate fine-grained spatiotemporal patterns with heterogeneous meteorological covariates, thereby significantly enhancing model deployability and proactive mitigation utility in real-world urban settings. Evaluated on a large-scale, real-world traffic dataset from New Delhi, the model achieves an average root mean square error (RMSE) of 1.12, demonstrating both high temporal responsiveness and low prediction error. The method delivers reliable, operationally actionable congestion early warnings and supports data-driven, preemptive traffic management interventions.

Abstract

The prediction of traffic congestion can serve a crucial role in making future decisions. Although many studies have been conducted regarding congestion, most of these could not cover all the important factors (e.g., weather conditions). We proposed a prediction model for traffic congestion that can predict congestion based on day, time and several weather data (e.g., temperature, humidity). To evaluate our model, it has been tested against the traffic data of New Delhi. With this model, congestion of a road can be predicted one week ahead with an average RMSE of 1.12. Therefore, this model can be used to take preventive measure beforehand.

Problem

• Predicts traffic congestion using multiple factors
• Incorporates weather data for accurate forecasting
• Evaluates model with real-world New Delhi data

Innovation

• Machine learning predicts traffic congestion
• Model uses day, time, weather data
• Achieves 1.12 RMSE one week ahead