I have done this analysis before but want to open it up to the community because I have not considered alternative methods.

I essentially am trying to "score" ZIP codes for how much a ZIP code demands private school and related goods and services. I have data for each ZIP code on...

  1. Household income (e.g., % of families earning above certain thresholds
  2. Level of education attainment (e.g., % of families earning bachelors, doctorates, etc.
  3. Level of private school enrollment (e.g., # and X% of the ZIP code is enrolled in PS
  4. Number of high quality public school alternatives

I would like to be able to create a function so that I can enter the values of each ZIP code and then have a result that returns and says demand is X%ile along a curve, but the problem is I don't have a Y variable that is just "private school demand". I can and in the past have used private school enrollment as a proxy, but I am wondering what you all think I should do here.

Thanks very much.


If you don't have the $y$, then you can't learn a function which gives you $\mathbb{E}(y\vert x)$, plain and simple. Using a proxy measure, as you've done, is probably the next best thing but you open yourself up to problems without knowing how and when the proxy is a bad proxy for the outcome.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.