Bus Travel Time Prediction Using Time Series Analysis

Abstract

The focus of this paper is on bus travel time prediction in a non-lane based mixed traffic flow conditions. Data obtained from GPS devices fixed on the Metropolitan Transport Corporation (MTC) buses in Chennai, India was used. The travel time patterns at different link types of the considered route were examined. A Monte Carlo method with the time series modelling technique was used to deal with uncertainty in bus travel time prediction. Seasonal Autoregressive Integrated Moving Average (SARIMA) was found to be the best model for the present data. The performance of the proposed models was measured using the Mean Absolute Percentage Error (MAPE) between the actual and the predicted travel time. It was found that the developed SARIMA model outperformed the ARIMA models and was able to predict the travel times reasonably well in different link types.