iArsenic – Instant Arsenic screening of hand pump tubewells in Bangladesh

About

Web-based application to estimate arsenic levels in untested wells in Bangladesh.

the main application is a set of static files:
- the HTML and JS for the page that allows a user estimate arsenic levels
- JS files with pre-processed data (see below)
- request log backed by a Google Cloud Function with a database of requests
the estimation is based on input data with measured arsenic levels in wells in Bangladesh; this input data lives in data/
- this data is processed statistically using one of the available models (see models/)
- we don't process the data on every user request, but rather we pre-process it so we can satisfy user requests with a simple data lookup
- a model generates aggregate data files that capture the estimate for every region (i.e. mouza) and every depth stratum and every combination of staining/flooding inputs, these become part of the static files above
every user request is logged in a little Google Cloud database (see server/) for usage analytics and to show impact

data/ — source data
- it specifies measured arsenic concentrations in tubewells around Bangladesh
docs/ — the website hosted in GitHub Pages
preprocessing/ — processing scripts
- these turn data from the source CSV form into something consumed by the website and other tools
- then there are tools for CLI running of estimates and testing
preprocessing/geodata/ – scripts that use Bangladesh geo-boundary data
- includes experiments with visualization of the various data we have
server/ — server for request log database
- in a Google Cloud Function

Original scripts in R can now be found here.