Validation of TASHA: A 24-h activity scheduling microsimulation model

The objective of this paper is to verify/validate the results of an application of the Travel Activity Scheduler for Household Agents (TASHA) in the Greater Toronto Area (GTA), Canada. Activity generation and scheduling components of TASHA are validated using 1996 and 2001 travel survey data for the GTA. This validation proceeds in two parts: (a) verification that TASHA replicates the 1996 base case upon which the model was originally built; and (b) comparison of TASHA’s forecasts of 2001 daily travel behaviour with observed travel survey data for 2001. TASHA activity generation and scheduling model components replicate observed activities with good accuracy and precision for the base year. Although TASHA is not able to predict an observed increase in activity participation rate in a five year forecast, the distribution of activities in the day is forecast with greater success. Predictions of average travel distance are oversimulated in the base year by 0.9%. Increased average distances are underpredicted for school, shopping, and other activities, and are overpredicted for work and return home activities. These validation results are promising, although there exist opportunities to improve model performance, and to further validate other elements of the TASHA model.