MIT License
We run the analysis using make
version 4.4.1
make data-clean
will generate all of the intermediate files we need for the analysis. On my computer it just took 4 minutes to run all targets. This does not run two categories of targets: downloading data and the water ML pipeline. Those both take a while to run, and thus I recommend you just get their data outputs from dropbox for now, along with the data we downloaded manually---they are grouped together in the data/download-manual
, data/download-script
, and data/watershed
directories.
I've set it up so that make data-raw
will create all of the files in the data/raw
directory, make data-clean
creates all of the files in the data/clean
directory, make water
creates all of the files in the data/watershed
directory, and make data-download
creates all of the files in the data/download-script
directory.
For analysis:
make desc-figs
will create the descriptive time series plots and mapsmake cnty-results
runs the county level analysis (event studies, DiD, TSLS for many outcomes and different treatment vars) and creates the event study figures. It also does the Ag district level analysisThe following API keys are required, save them to .Renviron with with usethis::edit_r_environ()
NASS_KEY
CENSUS_KEY
BEA_KEY
Ag district to county crosswalk from here.