Evaluate overall project results - which projects have positive/zero crash change or negative/zero travel change

bicyclingplus / atp-bc-tool-analysis

Analyzing inputs/outputs from the CTC Active Transportation Benefits/Costs tool, to identify and investigate potential issues (analysis notebooks only - input/output data is not included)

0 stars 0 forks source link

Evaluate overall project results - which projects have positive/zero crash change or negative/zero travel change #1

Open mRaffill opened 1 year ago

mRaffill commented 1 year ago

[x] Identify projects with at least one "unreasonable" travel/safety result
- [x] proportion of projects with "unreasonable" vs "reasonable" results
[x] Graph of travel and safety benefits across all projects
- line graph with x axis as project ID and y axis as change in travel/change in safety
- Separated by mode, location type, outcome
[ ] Identify projects with all "unreasonable" travel/safety results (eg. across all modes and location types and outcomes

mRaffill commented 11 months ago

Miles traveled across all projects:	existing (weighted) travel	change in travel

There are no obvious errors (eg. negative miles traveled) with the miles calculations*, although I haven't looked into the calculations in detail yet (that will be in another issue).

*There are 20 projects with a negative lower estimate for change in travel, this is just because conventional-bike-lane has a lower estimate of -21% change in travel so this is not an error.

mRaffill commented 11 months ago

Change in travel sorted low -> high (just to quickly see the distribution of the change in travel, again, more details in the future)	change in travel - bicycling	change in travel - walking

Most projects have around 0 change in miles traveled, a few have several thousands (which does seem a bit unreasonable, but maybe makes more sense looking at the project length.

Also note that this isn't split by location type, and is only for the mean estimate. So it might be different between segments and intersections - have to look into that in a later issue.

mRaffill commented 11 months ago

Crashes/change in crashes:	all	split by mode

So almost every single project has an increase in crashes, and this is consistent across all modes.

Should look into location types later (the results table I used in this file was not split by location type, need to use a different table)
There are also a few projects with bicycling > combined > walking, which seems strange - shouldn't combined be greater than both walking and bicycling?

bicycling low -> high walking low -> high combined low -> high

There are a few crash change estimates from projects which are negative (the tiny "tail" at the end) but a large amount have positive crashes in the hundreds and thousands.

mRaffill commented 11 months ago

These graphs were added in b3948cc665f19c9551791e8acf1290af2331ad85 and 84d1ec3ed4e9e05f517e75cd1a4cd72779d1e07f