Evaluating Moving Average Techniques in Short-Term Travel Time Prediction Using an AVI Data Set

Abstract

This paper reports on the performance of three moving average techniques in predicting average travel speeds up to 10 minutes ahead of time. Evaluation is performed on a dataset obtained by matching recorded Bluetooth MAC IDs at two ends of a freeway segment over a 24 hour period. One simple moving average method with constant weights and two adaptive moving average methods are investigated. Results indicate that all three moving average methods have more or less similar performance in predicting short-term travel times. However, the method using optimized weights produced slightly better predictions at a higher computational cost. Quality of predictions is diminished as the time for which predictions are made is farther in the future. Moving average methods overestimate travel speeds in slow-downs and underestimate them when the congestion is clearing up and speeds are increasing. In general, predictions are not reliable during highly congested periods, while they exhibit improved quality when traffic moves toward less congested and eventually free flow conditions. A comprehensive discussion on the nature and performance of the studied moving average techniques is provided.