عنوان مقاله [English]
The cost of collecting data for travel demand modeling is very high with an increasing trend. Data collection costs could easily surpass the annual budget of a metropolitan planning organization (MPO) in small or medium-sized area. Spatial transferability of travel forecasting models, or the ability to transfer models from one area to another, can potentially ease important cost and time savings for areas that cannot invest in extensive data-collection and model development procedures. Furthermore, transferability is critical to evaluate the validity of behavioral models. Without transferability in time and space, the use of the model will be compromised due to either over or under-estimating demand, which will lead to an inaccurate assessment of the associated transportation needs and poor allocation for infrastructure investment. Moreover, trip generation has special importance in travel demand modeling, because it is the first step in the classic four-step approach, and any error in this step leads to the transmission of the error to other steps.This study includes two primary research objectives. The first is to test the appropriateness of transferring ordered Logit model for trip production between two cities of Qazvin and Eslamshahr. The second is to determine the best transfer direction (Qazvin to Eslamshahr or Eslamshahr to Qazvin) according to the transferability criteria. The analysis focuses on work trips at the household level. The models are estimated for Qazvin and Eslamshahr cities based on data from the Travel OD Surveys of Qazvin (2008) and Eslamshahr (2013). The surveys collected detailed personal and household characteristics, as well as travel diary information from households in the two cities. Trip production model estimation was conducted in the statistical software package Stata.The measures of spatial stability consist of an analysis of how well the estimated models predict observed shares by stratification and a comparison of model parameters. Various transferability tests are conducted at both aggregate and disaggregate levels. Transferability Test Statistic (TTS) is conducted to test the stability of the model coefficients. Transfer Index (TI) is a measure of predictive accuracy of transferred model relative to a locally estimated model. Transfer Rho-Square is a measure of goodness of fit index and statistical methods such as Root Mean Square Error (RMSE) and Relative Aggregate Transfer Error (RATE) are measures of the aggregate prediction error.The selection of final models is based on different criteria like logical coefficient signs, chi-squared statistics, F-statistic, pseudo R2, and t-statistics. Qazvin and Eslamshahr final models include two explanatory variables: number of employees and car ownership. Models coefficients have correct sign and are all significant at the level of 5 percent.Research results show that Transfer Index rejects the null hypothesis of the equality of the parameters of the two cities. Results also indicate that the transferred models to Eslamshahr and Qazvin have Transfer Rho-Square of 0.06 and 0.15, Transfer Index of 0.50 and 0.71, Root-Mean-Square Error of 0.35 and 0.24, and Relative Aggregate Transfer Errors of 17.5 and 12.0 respectively, indicating multiple aspects of transferability. Other conclusions are also
drawn from different perspectives: the transferred model to Qazvin has a better Transfer Rho-Square, Transfer Index, Root-Mean-Square Error, and Relative Aggregate Transfer Error than the transferred model to Eslamshahr.