I have cleaned disaggregate travel data down to a household level (of about 75% of households) and socio economic data aggregated into 100-200 household blocks. What I'm wanting to do is display a map of the household vehicle travel to visualize the geographical influence of urban form on transport energy consumption. Using an IDW interpolation of the vector data, I have done this already to produce the following map:


However this doesn't take into account the spatial distribution of the demography/income characteristics of residents. I need to account for other explainable factors including household size and income to isolate the unexplained influence of the built environment. The end result I am wanting is a map similar to above of the amount of travel a 'mean resident' would carry out if they moved to that area. I realize such an inference is subject to the ecological fallacy but I'm fine with a crude analysis.

My experience with statistics is rather beginner level (a few 100/200 level papers) as I'm coming at this problem with a background as a mechanical engineer, but I'm happy to dedicate a bit of time into getting this done.

Where I am at

As a first step I have aggregated all data to a larger block size (See images 2,3 and 4). I'll eventually start working down to a lower level of aggregation when I sort out I understand that to control for income and HH size, I can use an OLS, obtain the coefficients and use average values for the controlled variables. However if I am not explaining the built environment variables in the equation, how do I account for that in the equation. I have tried this already using the above data to produce a map of residuals i.e how much more or less the a resident of the area travels than explained by socioeconomic variables(see image 5), but this isn't quite what I want.


A bit blunt but: what is the correct way of doing this?


  1. Median VKT


  1. Median Income


  1. Mean HH size

Household size

  1. Residuals

Std Resid

  • $\begingroup$ Can you say why the residuals are not what you want? From the rest of your description, it sounds like they are what you want. You can covert the residuals into "what the average HH would travel" by just adding average VKT to them (not to standardized residuals, but to the actual residuals). $\endgroup$ – Bill Oct 15 '14 at 18:50
  • $\begingroup$ Hi Bill, thank you. I was looking at the problem as if it were far more complicated than it needed to be. By using the percentage change between expected and actual result and then multiplying by an average resident value, I got exactly what I was wanting. $\endgroup$ – Tmontgom Oct 15 '14 at 22:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.