A Method to Identify Market Segmentation and Margin of Error for Trip Production Rates from Household Travel Survey
Corresponding Author: Liang Zhou, North Central Texas Council of Governments
Presented By: Liang Zhou, North Central Texas Council of Governments
Abstract
The purpose of this paper is to introduce a method for the identification of market segments in trip production models with a defined margin of error and confidence interval. It is normal in trip-based models to develop trip rates for trip production, but uncommon to create a margin of error and confidence interval associated with trip production rates. The average trip production rates are subject to the sampling error and the limitations related to travel surveys. A small number of samples and more variation of trip rates will produce less reliable average trip rates for forecasting. Since this information is vital for risk analysis, it is imperative not to limit the model parameters to an average, but include measures of expected errors.
The proposed method assumes the existence of a properly weighted household travel survey. We focused on trip production models for various home-based trip purposes. A set of household characteristic variables such as number of workers and household size from the household survey are selected as candidate market segmentation variables. Using standard statistical testing methods, the rates in neighboring segments were tested to determine if they are significantly different with 95% confidence. If the rates are not statistically different, the two neighboring segments are combined together. The final table for each trip purpose contains a statistically distinct set of rates, associated margin of error, and confidence interval for each candidate variable. The overall normalized margin of error of the production cross-classification table is also calculated as a weighted average of the errors of all rates from each segment based on the number of expanded households. The criteria on which the final segmentation is based include normalized margin of error, number of unique market segments and reasonableness of variables for each trip purpose.
This method was applied to the 2009 household travel survey of Dallas-Fort Worth area. It considered seven home-based trip purposes (work, shopping, social recreation, accompany, education K12, education college and personal business/other) and seven candidate variables, and generated trip production rates based on the best market segmentation from the survey.