Compare Fire Propensity Model and Fire Severity Model

[kelsonSS]

Determine:

1. Are the risk factors for severe fires the same for fire propensity.
2. What is the overlap in the model predictions?

• how many blocks are high in severe fires but not total fires?
• how many blocks have a high number of fires but not severe fires?

[jessezlotoff]

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):

[kelsonSS]

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.

[jessezlotoff]

Here's the comparisons at the block group level:

true_all_fires	true_sev_fires	total	pred_all_false	pred_all_true	pred_sev_false	pred_sev_true
FALSE	FALSE	4329	3437	892	3006	1323
FALSE	TRUE	292	187	105	81	211
TRUE	FALSE	304	88	216	149	155
TRUE	TRUE	249	90	159	60	189