Validation of an activity-based travel demand modeling system

Abstract

This paper presents the validation analysis of an activity-based travel demand modeling system – shorter-term decisions simulator (SDS). It compares microsimulation results of SDS with observed census and travel-activity survey data for the base year 2006, and forecasting years – 2016 and 2018. Base year validation suggests SDS’s reasonable capability to replicate observed data as indicated by lower absolute percentage error (APE) measures. SDS exhibits satisfactory performance for the forecasting years as well. For instance, Kolmogorov–Smirnov (KS) tests during the activity start time and duration analysis for year 2018 yield smaller D-values than critical values, which indicates a better fit model. Prediction of tour-level mode choice, shared travel choice, and vehicle allocation also exhibit good accuracy (± 4%). Validation results presented in this study demonstrate SDS’s promising forecasting capability so that the model can be utilized efficiently to develop and test alternative land use and travel demand management strategies.