INTRODUCTION
Trip generation models are an essential first step to an urban transportation modelling system (UTMS) and transportation planning in general. These models are estimated to predict the total number of trips that are produced (by people living in an area) or attracted (by employment/ destinations located within an area) in each analysis zone within an urban area. These models are also important in order to understand the trip making behaviour of urban population.
Trip generation models are most commonly estimated using a regression approach. However, transportation modeling is an art, as much as it is a science. The variables that are tested for their statistical influence on trip generation (i.e., the number of trips generated) can vary based on the purpose of the modelling exercise. For example, an engineer who intends to forecast trip generation would want to estimate a regression model with the best “fit” or predictability. In contrast, a planner who wants to understand the trip-making behaviour of individuals would probably take a more behavioural approach, and explore variables that provide a better theoretical explanation of travel behaviour.
Travel mode choice, in practice, is better explored using a disaggregated modelling approach, because mode choice is more influenced by an individual’s behaviour, preferences and circumstances more so than any other travel-related decisions. However, in the absence of detailed data, influences on travel mode choice can be examined using an aggregated linear regression approach.
LEARNING OBJECTIVES
In this assignment, students will estimate regression models of work trip generation, and will use the same technique to produce aggregated estimates of transportation mode choice, specifically the transit usage rate for commuting. Students will interpret the results and critically discuss the modelling approaches.
Keep in Mind:
1. Responses must be type-written in complete sentences; proper paragraph structure must be used. The assignment must have a cover page with Your full name, course umber and assignment number.
2. Save the SPSS, R or Excel output file (i.e., your original model results) and include with your submission as appendix.
3. Information taken from other sources (if needed) must be properly referenced.
THE PROBLEM
The data required to complete this assignment is saved in a file named data.xlsx, which can be downloaded from the course website. Travel data has been extracted from the 2001 Transportation Tomorrow Survey (TTS) database (http://www.dmg.utoronto.ca/transportationtomorrowsurvey/index.html). A description of variables is also provided in the file (separate worksheet named: Description). You can create more variables using the given data (e.g., population density = population / area; employment density = employment / area).
The data table contains one record (i.e., row) for each traffic analysis zone (TAZ) in the City of Toronto; the values for travel and socio-demographic variables represent the total or average for all households living in that TAZ. The built environment variables were computed from other sources using GIS.
Using the given data, estimate TWO REGRESSION MODELS of –
(1) household work trip rates (i.e., number of work trips per day per household). In this model, you are expected to adopt a predictive approach to modelling. Your goal here is to produce a best-fit equation to be used in predicting trip generation for a future year in the GTHA, where every independent variable included in the model must be statistically associated with the dependent variable. and
(2) transit mode share (i.e., proportion of daily work trips by transit). You are expected to take a behavioural approach to modelling, and explore and discuss associations as well as non-associations between independent and dependent variables.
The independent variables must be chosen carefully (do not use a kitchen sink approach). Justify their inclusion either by citing relevant literature or by providing logical explanation. Also when choosing the variables, please consider their appropriateness. For example, The population of a TAZ is not a good variable to include in a model, because it can vary by the size/area of a TAZ. A better variable is population density, where population data is normalized by area [15 + 15 = 30 marks]
For each case, present and discuss the results of your analysis in terms of the statistical significance of the parameter estimates and the goodness-of-fit of the model. [6 + 6 = 12 marks]
Compare the two models in terms of the variables that were significant. Also discussion the limitations of this modeling exercise. [8 marks]