Comparative approach for predicting travel time reliability (a case study of Virginia interstate)

Travel time reliability (TTR) is known as the temporal variability of travel time and influences many aspects of the traveler's decisions, especially in roadway transport. Investigating literature review shows seven factors can cause congestion and affect travel time reliability. Three of the most common factors investigated in numerous studies include traffic incidents, adverse weather, and traffic volume. Most of the previous studies have focused on modeling the variations of a single variable on the dependent variable (reliability metric), and those considered multivariate analyses have used traditional tools like linear regression. This paper investigates the effects of traffic volume, weather conditions, and traffic incidents on the planning time index (PTI) as a dependent variable in Virginia's interstate highways. Six machine learning (ML) techniques: neural network (NN), support vector regression (SVR), linear, polynomial, and radial basis function (RBF) kernel functions, K-nearest neighbors (KNN), and decision tree (DT) and five ordinary least squares (OLS) linear regression models including linear, inverse, growth, power, and S-shaped functions were used to analyze TTR in this study. Analyzing coefficients, root mean square error (RMSE), and stability of models besides using statistical measures of error terms of maximum, standard deviation, and mean outlined that KNN is the best model for estimating TTR.