These are the analytics portion of our proposed study. This is a self-contained set of scripts that does analysis and prepares reports on that analysis. All of this is a work in progress. The data files are not included and need to be obtained from our PI but simulated data-files are bundled with the project and have everything you need in order to test out these scripts.
As far as I know this project will work on Mac, Windows, and Linux.
The first time you clone or download this project from GitHub onto a new
computer you will need to install some required R packages in the script
onetime_setup.R. You can do this automatically from an R console started in
the project folder with source('onetime_setup.R'). Or you can do it from the
command-line in the project folder with R -f onetime_setup.R.
At the moment the only script you are likely to need is variableselection.R.
In an interactive session you can open it and run it line by line, or you can
source('variableselection.R'). You can also compile it directly into a
self-contained HTML document with the following command:
rmarkdown::render('variableselection.R') and that is how the results are
intended to be viewed.
If you want to compile a Word version of the output, you could use:
rmarkdown::render('variableselection.R',output_format='word_document')
By default, all the data used is the simulated version.
If you downloaded the SDOH_RSR_2013_prelim.csv and SDOH_RSR_2013_prelim.csv
spreadsheets from our shared folder just put them in the top level of this
project folder and start working with variableselection.R (or a new script of
your own). Please remember to not add the data files to your repository.
If you have the raw data and wish to use build SDOH_RSR_2013_prelim.csv and
SDOH_RSR_2013_prelim.csv from scratch, create a script named local.config.R
in the top level of this project folder and put the following code into it,
replacing all the file paths that begin with SIM_ with paths to the real
versions of those files wherever you keep them on your computer...
inputdata <- c(dat0='data/SIM_SDOH_ZCTA.xlsx' # census data by ZCTA
,cx0='data/SIM_ALLCMS.csv' # RSA-ZCTA crosswalk
,rsa0='data/SIM_RSAv4 SCD RSRs.csv' # outcomes (RSR)
,dct0='data/data_dictionary.tsv' # data dictionary for the
# dat1 dataset that _this_
# scriport produces
,dat1='SDOH_RSR_2013_prelim.csv' # the dat1 dataset
,dat2='SDOH_RSR_2013_scaled_prelim.csv' # the scaled version of
# dat1
);
This is all you need in order to generate the same results that I've been posting to our shared space. There is no need to change paths for any of the other files-- all of them get generated as needed if they are missing.
If you are collaborating with us you are welcome to create forks. We welcome your pull requests and suggestions. You should never check the following items into github:
-
local.config.R. Doing so will ruin the portability of your version of the project. As a result, anybody who lacks the needed files or keeps them in a different location than you do will get error messages when they try to run it instead of example results from the simulated test-data. -
Any real data, unless explicitly directed to do so by TR. Unless TR says otherwise, we are not posting the real data publicly. Furthermore, data is potentially large and will slow down the repo. Git is for code. So even if and when we do release the data proper place for datasets is https://zenodo.org or something similar