My dataset consists of hourly values by weekday across several sites, where the sites vary by spatial location and by other common characteristics, such as type, or 'cafe,' 'restaurant,' and 'bar.'

For some sites I am missing days of hourly data. I would like to develop an estimate for mean and variance for the missing hours from the hourly data I do have.

What statistical models / approaches might I use to estimate the mean and variance for a missing hourly at a given site for a given day of the week, where the value is missing in my dataset?

I'm closing this question.

I had been seeking a methodology for developing estimators for hourly values at sites where location and site type were likely to explain a given site's hourly values. I guess that is imputation, but I think, rather, I am looking for some form of regression where the modeling would traverse hourly values across week days and across sites spatially.

In the end, I find my quest for seeking better definition of this modeling challenge a poor fit for this forum.

  • $\begingroup$ If you only want to know Python / R modules / code to deal w/ this situation, that would be off topic here. If you are wondering how to deal w/ this generally (ie in a software-neutral sense), please edit to clarify. $\endgroup$ – gung Jan 30 '16 at 0:08
  • 1
    $\begingroup$ Thank you for your recommendation; I have edited my question consistent with your comment. $\endgroup$ – gallygator Jan 30 '16 at 0:42
  • $\begingroup$ Thank you for clarifying your question. I have retracted my close vote. $\endgroup$ – gung Jan 30 '16 at 0:45
  • 1
    $\begingroup$ It sounds like you have time series-cross-sectional data -- Amelia is one multiple imputation tool that was developed specifically for this type of data. $\endgroup$ – Sycorax Jan 30 '16 at 1:53

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.