bayesimpact / encompass Goto Github PK

Uploads can be in 1 of 3 formats: Zip Codes only, Counties only, or Zip Code and County pairs.

Today, we assume that both zips and counties are defined. But we should assume that one or the other is defined.

Let's disable these selections for demo purposes. We can still display them so that we allude to the eventual capabilities, but I don't want to mix dummy data with real data-- it's just confusing to track what's actually computed vs. what's placeholder.

corresponding issue in backend: https://github.com/bayesimpact/time-distance-standards/issues/89

Steps to reproduce:

1. Select San Francisco County, all zips
2. Change distance from 0.5 to 4 miles

Expected behavior:
Bigger points show up

Actual behavior:
No points appear

That way progress indicators are trivial to implement.

Open questions:

• How well would backend be able to handle this volume of queries?
• Would sending 100k queries at once saturate the browser's TCP thread?

1. Go to http://localhost:8081/#/service-areas
2. Uncheck 94014 and 94102
3. Go to analytics pane

Expected: Analytics dropdown should be set to All service areas
Actual: Analytics dropdown still shows San Francisco / 94014

For both Service Area and Providers, add info on the expected format of the CSV.

Expected behavior:
Push zoom button and map scale to the right when drawers are in use.

It's nice to have some feedback after a user has uploaded a list of providers.
I understand the geocoding of providers is not optimized currently, but it's helpful to understand that the application is working.

A few options:

1. make the application inactive. use a module overlay that shows a loading circle or bar with some text about what process is currently happening (geocoding, running time and distance analysis, etc)
2. put loading messages in respective drawers. ie: when geocoding providers, a loading circle or bar will show up in the Providers drawer (a user will have just uploaded a CSV here so this should make sense). then the map auto-zooms to encompass the new providers as well. (should be considered in #12). when running the analysis, the charts/figures in the analysis drawer should be grayed out until complete.

Expected behavior:
Analytics drawer should have two rows:

• percentage by number of representative point
• percentages by population

Actual behavior:
Analytics drawer shows percentage by rps only

• Color representative points appropriately
• Color bounding polygons appropriately

“Clear Inputs” button to clear inputs and start a new analysis should appear in Service Area andf Provider drawers.

Favicon throws an error and does not load properly.

Use an environment variable to define the API endpoint (in services/api.ts)
This will make deploys easier and more straightforward

I think this should at least match what we have built for RPG.
We can anticipate what the most common headers will be--ie: County, county, County Name, County name. This will also lessen the need for issue #67

The experience of uploading an incorrectly formatted CSV is worsened by the fact that there is no "Clear Inputs" button-- I have to reload the page each time.
nudge for issue #14

Please connect this repo at https://reviewable.io/repositories so that all PRs get linked to reviews. Thanks!

We now return all populations as a hash, and the API is no longer parameterized by distance.

ie: if "Time/Distance" is changed to "Time", the Measures values should change from combined time / distance values to only time values

Steps to reproduce:

1. Select San Francisco, all Zips
2. Toggle from 0.5 mile distribution to 5 mile distribution

Expected behavior:
Points get larger as they represent more people at the 5 mi setting.

Actual behavior:
The points are clustered appropriately for 0.5 but then shrink to be the same size at 5 mi.

Currently the only option for a user is to reload the page or upload a new CSV.
It would be nice to have a Clear Inputs button that removes the uploaded CSV, just to be able to start with a clean slate. Mostly useful for debugging.

• Map should auto-adjust to specified service area and appropriate zoom level
  • If two non-adjacent service areas selected, map should adjust to a zoom-level that encompasses both areas

Adding a button will degrade UX - is there another way to improve performance instead?

Moved from #61

The name of the uploaded CSV should be shared so that we can easily know which files were uploaded when we take a screenshot and or compare results. @viviandien on where to do this thoughts?

eg. zipcode 95678-6194 should be simplified to 95678

Steps to reproduce:

1. Upload service area file in Service Area drawer.
2. Upload PCPs file in Providers drawer.
3. Go to Analytics drawer.
4. Change dropdown from "All service areas" to "Amador/94501"

Expected behavior:
Map zooms to selected service area and displays only points in that service area.

Actual behavior:
Map does not react and all points are displayed.

1. Aggregate adequacy metrics for the selected service area
   a. Create a Percentage Circle showing the percentage of population adequately covered.
   b. Reveal aggregate metrics and “worst case scenario”.
   c. Allow user to download a report.
   d. When user highlights a different service area in the map, update analytics

Mockup:

Representative points in adjacent service areas should have different colors.

Steps to reproduce:

1. Upload PCPs file in Providers drawer.
   https://drive.google.com/open?id=1x31zlVpEi6QPC1xQl5XT1SZBzIaUiBJ4

Expected behavior:
Two snack bars appear, 1 to give an error about the data that was uploaded, and 1 to say that providers have successfully geocoded.

Actual behavior:
Both messages should be visible. Suggested UI - stack the snack bars vertically on top of each other.

Add a button to load the app with demo data, maybe Fresno and Fresno providers?

The idea is to make it easier for people to debug the primary functionalities of the app. It would also make it easier to share the link externally.

Two options:

1. Have a button to download two csvs that can then be used with the app
2. Have a button to autoload a state with the corresponding data

@viviandien @bcherny thoughts on the above?

Steps to reproduce:

1. Upload a CSV of known addresses for providers
2. Wait for the backend to process them
3. If there is no error, it will show a Failed - 0 providers failed to geocode

Expected behavior:
Success - All N providers geocoded

Actual behavior:
Failed - 0 providers failed to geocode (N)

Steps to reproduce:

1. Navigate to http://localhost:8081/#/service-areas
2. Select Alameda county
3. Check all zips (observe that representative points load)
4. Also select Fresno county
5. Check all zips

Expected behavior:

Representative points for Fresno should appear

Actual behavior:

POST /api/representative_points returns a 500 error

TODO

Just like what we already do for Service Areas: