search

insight-lane / crash-model

Build a crash prediction modeling application that leverages multiple data sources to generate a set of dynamic predictions we can use to identify potential trouble spots and direct timely safety interventions.
https://insightlane.org
MIT License
113 stars 40 forks source link

Trend analysis for various elements of the data #171

Closed bpben closed 11 months ago

bpben commented 5 years ago

We heard from BTD that they'd be interested in how some of the features we use and outcomes we target predicting have changed over time. Particularly this was pitched as "has volume or speed on a certain segment been increasing?"

I think there's a lot of exploration that could be done here, along a couple of lines:

Segment characteristic changes: I'm not sure what data are available, but it would be interesting to see if things like speed limit or number of lanes has changed over time across a city. These would be make interesting additions to the model (e.g. if a segment has recently had a reduction in speed limit, maybe that's changed its risk by some amount)

Temporal characteristic changes: The easiest one of these would just be to look at have accidents been on the rise. But likely a lot of cities already know that. Something they may not know is if traffic volumes have been on the rise and where they've been on the rise. If we get some measures of traffic volume, this would be a valuable analysis.

bpben commented 11 months ago

This is tricky, and stale, closing.