Compare Fire Propensity Model and Fire Severity Model about rcp2 HOT 3 OPEN

I've been (accidentally) working from the tract model, not block groups. I'll be updating the code accordingly in the next few days.

@kelsonSS , I'm planning to use the full modeling data (X) to compare the predictions since the test sets won't necessarily match up. Any concerns there?

Here are the initial feature comparisons (from the tract model):

from rcp2.

the only difference in code in tract vs block model should be how its aggregated at the beginning so it should be an easy fix.

Let me know what the current severe fire block model is ( and push it to main if you need to). I'll be trying to make the first pass on the ensemble model on Tuesday.

from rcp2.

Here's the comparisons at the block group level:

from rcp2.