maxachis
commented
2 years ago It turns out that with fresh_produce, we weren't identifying it anywhere at all. We did it previously in a prep script, but code changes meant that was no longer the case.
Fortunately, that's an opportunity for more modular scripting, as now we can create a script explicitly dedicated for setting types. I've made a branch for doing just that.
I did a comparison of my branch's merged_dataset with the most recent merged_dataset of the main branch -- the chances mostly seem acceptable. But I note a few things I want clarification on before moving forward with a pull request
- In the new branch, some columns are being interpreted as floats rather than ints (0.0 instead of 0, for example). My code didn't do anything explicitly with the other columns, but it's possible loading the code led to some sort of automatic float conversion. Might not be a problem. Or it might! Should I add code to make sure the flag columns are interpreted as ints?
hellonewman
A script called prep_snap.R that is pulling in sites from SNAP. We thought it was categorizing them by type and saying if the type is a farmer's market or supermarket, then type is 1. But we create a type field and just put NA in it, because were trying to determine a type by name, like if it contains Giant Eagle it's a supermarket, etc., but we moved that to a different script that might be downstream. So we're saying that supermarkets have fresh produce but the script doesn't know what site is a supermarket yet so we haven't told it yet.
So we need to find where we're identifying things as supermarkets or fresh produce.