A novel spatiotemporal framework for efficient traffic prediction and visualization

Abstract

This thesis proposes an efficient traffic prediction framework to estimate congestion at intersections depending on neighboring road links. The framework encompasses three major components, data extraction, Bayesian Linear Regression-based traffic prediction model, and an interactive map-based traffic simulator to visualize the results. To collect traffic data, we have developed an open-source web-based data scraper tool to extract and export publicly available traffic data from the Google Maps web interface. We also developed a Bayesian Linear Regression-based traffic prediction model to estimate traffic congestion that leverages Bayesian inference to facilitate model interpretability and quantify model uncertainty. The experiments show that Bayesian linear regression modeling can be trained on small data observations to quantify model uncertainty and predict traffic congestion without sacrificing interpretability and accuracy compared to the frequentist approach. We have also developed a web-based traffic simulator to simulate linear regression-based traffic prediction models and visualize the results on interactive maps.