Fill in some data gaps for the 4/12 content draft

[futurechris]

These questions are from https://docs.google.com/document/d/19ha8W6b8GXKGgO_vsyM71khuFHZcFKFCX2oAh7-MVKk/edit#

Might be a little tricky to match the work below to actual editing, if changes are made, but just comment if you run into trouble.

Section "What does Portland Fire & Rescue do?"

• [x] "13% of the calls Portland Fire & Rescue responds to are fires*? The vast majority (78%)".
  • I filled these in based off of Mark's comments in another ticket (#31 on April 11th). I don't know if he meant his answers to be presented as numerical facts, though, so we should run the numbers ourselves to make sure we don't misrepresent PF&R.
• [x] Other examples of what comes under the 'Fire' category: "explosions, ___, and __."

Section "Who does PF&R serve?"

• [x] "To achieve this, we mapped the FMAs separately against their specific demographic data from the _____census." @sky-t We might need to rewrite the copy a little bit since it's not a specific census year, right?

Section "How busy is PF&R? When are they most busy?"

• [x] "On an average day each FMA responds to approximately __ calls."

Section "The First Five Minutes"

• [x] "The average response time for the city of Portland is _____." I think the median time is 4:31, but there's some trickiness to getting this number, since we have some outliers (events that have null or no entry for dispatch, etc.). Do we want to split this on fire/medical or is that just complicating the copy?

Feel free to edit the document directly, or else drop answers and comments and I can do that.

[futurechris]

I've computed the median times (for the 5th check-box), but I'm a bit perplexed. My previous computation had the median as 4m31s, but now I'm getting 4m35s. Not a huge difference, and it's likely explained by the new calculation (with Kyle's help) removing all the entries that didn't have dispatch & onscene entered. Previously we had a handful of "-731 day" response times. These outliers were mostly taken care of (on both ends) by taking the median, but there was some impact.

I'll add my notebook to the repo sometime soon so even if it's wrong it'll there for someone else to figure out why. :)