usds / justice40-tool

A tool to identify disadvantaged communities due to environmental, socioeconomic and health burdens

https://screeningtool.geoplatform.gov/

Creative Commons Zero v1.0 Universal

125 stars 42 forks source link

2119 there are a few tracts places where the values over the 90th percentile are not showing as blue #2160

Closed sampowers-usds closed 1 year ago

sampowers-usds commented 1 year ago

The main change here is in the score/constants.py file, adjusting IS_ET to pull on NON_NATURAL_PCTILE_THRESHOLD as opposed to NON_NATURAL_LOW_INCOME_FIELD_NAME. Also adding the ipython notebook I used to investigate to the ipython folder

vim-usds commented 1 year ago

@sampowers-usds - seems as tho maybe some of the smoke tests are failing. I think it may have to do with the donut calculations?

github-actions[bot] commented 1 year ago

Score Deployed! Find it here:

github-actions[bot] commented 1 year ago

Map Deployed! Map with Staging Backend: https://screeningtool.geoplatform.gov/en?flags=stage_hash=2160/2d31365ea6bbd7719accbb660157faba980f97a3 Find tiles here: https://justice40-data.s3.amazonaws.com/data-pipeline-staging/2160/2d31365ea6bbd7719accbb660157faba980f97a3/data/score/tiles

vim-usds commented 1 year ago

@sampowers-usds - This looks great!!

It seems that each of the original tracts outlined by @KameronKerger for the lack of green space not being highlighted when over 90th percentile seems functional!

[ ] originally failing tract on prod is now passing on staging
[ ] originally failing tract on prod is now passing on staging
[ ] originally failing tract on prod is now passing on staging

Out of curiosity, were the total number of tracks designated as disadvantaged changed by any amount due to the fix? Are there other questions that should be asked along the same line of thinking @KameronKerger?

vim-usds commented 1 year ago

@sampowers-usds - the original ticket also calls out some issues around the lead paint indicator. I'm not sure I understand the problem and if others are feeling the same, maybe we can create a new ticket to identify this, as this PR seems to fix the lack of green space issue.

github-actions[bot] commented 1 year ago

Score Deployed! Find it here:

github-actions[bot] commented 1 year ago

Score Comparison Summary

Hi! I'm the Score Comparator. I compared the score in production (version 1.0) to the freshly calculated score. Here are the results.

Columns

I compared the columns. Here's what I found.

There are no differences in the column names.

Scores

I compared the scores, too. Here's what I found.

The production score has 74,134 census tracts, and the freshly calculated score has 74,134. They match!
I compared all of the census tracts. There are 51 tracts with at least one score difference. Please examine the logs or run the score comparison locally to view them all.

travis-newby commented 1 year ago

I compared all of the census tracts. There are 51 tracts with at least one score difference. Please examine the logs or run the score comparison locally to view them all.

These are the same deltas we've seen in main

github-actions[bot] commented 1 year ago

Map Deployed! Map with Staging Backend: https://screeningtool.geoplatform.gov/en?flags=stage_hash=2160/f8d5ec7e00ca380519398b8f6663806383106e0f Find tiles here: https://justice40-data.s3.amazonaws.com/data-pipeline-staging/2160/f8d5ec7e00ca380519398b8f6663806383106e0f/data/score/tiles

vim-usds commented 1 year ago

All three problem tracts still have the lack of green space field fixed!