|M.Sc Student||Adrian Cotrus|
|Subject||Analysis of Trip Generation Characteristics in Israel for|
the Years 1984, 1996/7 and Spatial and Temporal
Transferability of Trip Generation
|Department||Department of Civil and Environmental Engineering||Supervisors||Professor Prashker Joseph|
|Full Professor Shiftan Yoram|
This research investigates the transferability performance of Disaggregate Trip Generation Models in time and in space. The research also examines the effect of selected socio-economic, demographic and spatial variables as well as household and person characteristics on trip generation. This research investigates two model specifications: Multinomial Linear Regression (MLR) and Tobit at the person level. The models are calibrated for the Metropolitan Areas of Tel Aviv and Haifa, from 1984 & 1996/7 Israel National Travel Habits Survey’s (NTHS). The transferability tests held in this research show that at a 95% confidence level, we reject the null hypothesis, i.e., the models vary in time and in space. Additional findings show that although not transferable (statistically), Tobit and MLR models are quite reliable achieving aggregate demand estimations. Statistical tests held at the 95% confidence level show that most of Tobit model coefficients do not differ in space while in M.L.R. models most of the variable coefficients vary in space. Moreover, we emphasize the need for homogenous specification of N.T.H.S’s at all levels. The heterogeneity of N.T.H.S’s appears from the definition of the surveys (84 N.T.H.S defined as “Activity” survey, 1996/7 N.T.H.S defined as “Rides” survey) and follows with the inconsistency of variables and data file structure. In summary, for the first time in Israel, M.L.R and Tobit disaggregate person level trip generation models were calibrated for the Metropolitan Areas of Haifa and Tel Aviv and tested for transferability in space and in time.