Travel Model Calibration and Validation

Travel models are used extensively to forecast travel demand and future traffic volumes under a wide variety of socioeconomic, demographic, multimodal, and technological scenarios. As infrastructure investments are often based on forecasts of future travel demand and needs, it is important to ensure that travel models reflect actual conditions in the real-world and are able to provide useful and accurate predictions of traveler behavior into the future. After travel models are estimated using survey or other data, they must be calibrated and validated to match ground truth, usually available in the form of survey data, traffic volumes, highway speeds and travel times, and bus and rail ridership by time of day. This chapter provides an overview of travel model calibration and validation techniques commonly adopted in practice, with an emphasis on the need to perform sensitivity analysis of calibrated models recognizing that such analysis can provide deep insights about the suitability of models for forecasting applications.