chicago / clear-water-app

Generate water quality predictions for Chicago beaches, and upload to https://data.cityofchicago.org/Parks-Recreation/Beach-E-coli-Predictions/xvsz-3xcj

Clear Water App

This is a companion repository to the Clear Water model repo for predicting beach water quality. It can be used as a production-ready solution to deliver model predictions to a Socrata data portal and map.

Notes

The file model.Rds is obtained by running Master.R in the Clear Water model repo. To generate the file, set the productionMode variable in the model repo to TRUE. Place the file in the data directory of this repo.

Make sure you also set the datasetUrl variable in app.R.

Socrata Credentials

To upload predictions to a Socrata portal, this application uses Chicago's RSocrata R package. You will have to authenticate with your data portal creditionals in order to add data to a predictions dataset. To authenticate with Socrata and upload predictions using this automated app, create files to save login and app token details in the credentials folder, named as follows:

• email.txt
• password.txt
• token.txt

App.R will read these credentials and use them to send the data to the Socrata dataset.

Model Settings

This app is currently configured to predict water quality at beaches within a regional system. Specifically, it predicts the quality of some beaches along Chicago's Lake Michigan coastline based on the same-day bacteria observed at 5 predictor beaches.

Contributing

If you are interested in contributing to this project, see our Contribution Guide.

clear-water-app's Issues

Not enough DNA test results to issue a prediction at "today"

Dear Tom Schenk Jr,
I'm sorry if I'm asking something obvious but I'm a freshman in programming and I'm encountering 2 different problems in Running the app.R code (Run with sys.time=2017-11-18 12:43:00"):

1. The datasetUrl value "https://data.cityofchicago.org/Parks-Recreation/Beach-Lab-Data/2ivx-z93u/data" gives the error 403, so I changed this value to "https://data.cityofchicago.org/api/views/2ivx-z93u/rows.csv?accessType=DOWNLOAD" and then I'm able to assign the datatable to labPortal. Have I done the right thing?

2. Then I tried to generate predictions but the variable readyToModel is FALSE and so I'm not able to perform the predictions. So I tried to change the today value to an old date (such as may 2017) but readyToModel is still FALSE.

What I can do to solve these issues?
Thank you

Giada

Enhance technical instructions in README.md

Add more detailed instructions that compliment the instructions in the model repo. See also Chicago/clear-water#132.

Randomize website

We should see if it is possible to randomize which website is seen by end-users. This sort of randomization can then be used to estimate which visualizations and webpages are most effective.

Missing `credentials` folder

The instructions ask users to place credentials such as email.txt, password.txt, and token.txt in the credentials/ folder. However that folder isn't placed in the repo because, presumably, it is empty because the aforementioned text files are in .gitignore. This may a bit confusing for users.

Git does not display empty folders, but there is an emerging practice to place a .gitkeep file as a placeholder file so the folder will display.

Add automated email notifications

Right now, the results are written to a log. Add code to send email notifications.