A Systematic Data-Driven Framework for Evaluating Changes on Traffic Signal Settings

Abstract

Transport agencies worldwide regularly assess the state of traffic corridors and make necessary changes to enhance performance. It is important to systematically evaluate the effect of these changes to quantify their actual effectiveness and inform future decision-making. This study proposes a framework for a formal evaluation and quantification of changes in traffic signal settings while considering multiple key performance measures and likely hidden traffic patterns that could affect system performance and subsequent conclusions. The proposed framework involves: (1) data preparation procedures and recommendations, including a minimum sample size estimation; (2) a before and after analysis considering several performance measures, such as the empirical cumulative distribution function, buffer index, and planning time index; (3) a pattern analysis to account for inherent traffic flow dynamics; and (4) a seasonality analysis to capture long-term variations in traffic conditions. This study illustrates, using three case studies in South East Queensland, Australia, how overlooking certain aspects of an evaluation process can lead to incomplete or inaccurate conclusions. Two of the case studies involved upgrades in signal timings, while the third represented a control scenario with no changes. The proposed framework effectively quantified the effect of the upgrades on the first two case studies, while no statistically significant changes were observed for the third case study. Distinct traffic patterns were identified, and the corresponding effects were assessed accordingly.