A supervised learning approach to calibrating annual average daily traffic against highway roadworks: the impact of demographic and weather conditions

Abstract

ABSTRACT Annual average daily traffic (AADT) is an essential parameter to evaluate the level of mobility in large urban corridors often affected by highway roadworks. However, very little is known about AADT calibration methods that can project the impact of highway roadworks. This study draws upon 13,152 data points collected from the M8 motorway in Scotland, U.K., to propose a machine-learning-driven schematic calibration methodology that can extract the impact of highway roadworks from existing AADT measurements. The robustness of the proposed model is rigorously tested and validated. As the first of its kind, this study provides practical equation models that can extract the impact of roadworks under different demographic and weather conditions from the given AADT. This study should assist governmental transportation agencies in estimating the potential impact of highway roadworks from the very beginning procedures of developing transportation management plans, which is hidden from a single figure of historical AADT.