Travel time prediction for Two-Lane Two-Way undivided carriageway road Section- A case study

Abstract

Significant efforts have been made to enhance the ability to predict travel times along road corridors, considering various influencing factors. However, forecasting travel times remains a complex challenge due to the intricate interactions among numerous variables, which are often difficult to capture fully. This complexity is particularly pronounced on undivided roads, where unrestricted access to the route amplifies the impact of various factors on travel time. This area has received limited research attention. This study aims to develop a travel time prediction model for a 13 km corridor, specifically a two-lane, two-way, undivided rural highway section of the East-West Highway. By analyzing 72-hour datasets on vehicle travel time obtained from traffic volume counts, the study evaluates the performance of machine learning regression techniques, incorporating factors such as through traffic volume, opposing traffic volume, and the percentage of heavy vehicles in through traffic. The regression analysis reveals a moderate correlation between travel time changes and variations in the independent variables. Furthermore, statistical error tests demonstrate that the random forest method outperforms other approaches in predicting travel time. By addressing the effects of mixed traffic conditions and opposing traffic volume on two-lane, two-way undivided roads, this research contributes to improved travel planning, road network design, and advanced transportation modeling.