data management

This repo provides code and metadata for releasing MEDSL datasets.

We use the medsl module to:

Validate precinct-level returns from each state
Combine and transform these returns into release-ready datasets
Generate accompanying documentation
Update the elections R package

use

To generate docs and datasets:

$ python release.py

We assume the following directory structure:

├── 2016-precinct-data
├── data-management
├── elections
└── precinct-returns

Where:

2016-precinct-data contains input data, the returns for each state (not yet available online);
data-management is this repository;
elections contains our R package for election data, and is an output target;
precinct-returns is the repo for released datasets, and is an output target.

installation

Requires Python 3. Not yet tested on anything but Fedora Linux and Python 3.6, but should be fine on MacOS.

virtualenv installation:

$ git clone git@github.com:MEDSL/data-management.git data-management
$ cd data-management
$ python3 -m virtualenv env -p python3
$ source env/bin/activate
$ pip install -r medsl/requirements.txt

The feather-format library depends on pyarrow, whose availability on pip varies by platform (instructions).