mattwigway / peak-spreading

Are transportation peak hours getting longer?
MIT License
0 stars 0 forks source link

Daytime entropy hours code and test #33

Closed samzhang111 closed 2 years ago

samzhang111 commented 2 years ago

For #32.

I've added a column in the preprocessing code for computing the daytime entropy, and written some corresponding tests to check that it works at the unit level. I don't actually have the raw data for regenerating the data and doing the downstream analyses, but hopefully this is enough of a push that it can make the updates downstream fairly straightforward.

I avoided overwriting the "occ_entropy" column since I figured we wanted the old results to stay reproducible without renaming all of the variables, since we are discussing some of the issues with the 24-hour entropy in the paper. Thus to plumb this through the analysis, we'll need to plug in "occ_entropy_daytime" in a few places in the notebooks and rerun that, but I figure that's straightforward enough that I don't feel (too) bad to leave this for Matt :). (Or Matt: once you regenerate the data and check that back in, I can do that too if you prefer)

For the actual hours that are "daytime", I picked 5AM to 8PM since traffic can start fairly early and end rather late, but those are arbitrary decisions and I'm happy for them to be revised if need be.