EMS Heatmap: Update to use ALL EMS incidents vs those that used the NFIRS EMS module

[meilinger]

No description provided.

[garnertb]

Before starting this, please confirm it actually mitigates the issue where there is no EMS heatmap for departments.

[meilinger]

@zunware could you run the EMS incident counts by state (akin to the ems.ems counts by state) and post in here?

[josellausas]

I have generated counts and totals for the ems.basicincident query for each state and analyzed how many entries have geom vs not.

Total entries for all states:

• Total entry count = 117383071
• with geom = 75929729
• without geom = 41453342

Total entries without geom is 35.3%

Most missing geoms are from 2012 and 2013, and no geoms for 2016.

Total missing geom by year

• Most complete year is 2015.

All States

Charts for all states are here:
figures.zip

Some states with entire years missing geo

AK

• No geo for 2014

AL

• No geo for 2012, 2013, 2014

CO

CT

GA

MI

MO

MT

OR

SD

[josellausas]

Python notebook can be found here: https://github.com/FireCARES/firecares/blob/ems-update-analysis/scripts/ems-heatmap-analysys.ipynb

[meilinger]

@zunware could you compare these results against the ems.ems[year]_geocode tables? those ems tables contain the geocoded results (in wkb_geometry) If states are missing geocoded addresses for specific years then it makes me think that there was a bad join when trying to add geometries to ems.incidentaddress records from these tables.