Dataset: Election
=================

This dataset contains voting results per county in the 2012 presidential
election, along with various population statistics about the county.

Not all counties are represented in this dataset. Two random, non-overlapping
subsets were drawn from the full dataset to produce the training and the
testing sets.

:Dependent variable: 
    The column ``VOTES_DIFF`` is the percentage point difference between votes
    for Obama compared to votes for Romney in a specific county. If Obama
    received 25% of votes and Romney 30% of votes, this value will be -5.

:Predictor variables: 
    The predictors are population statistics about the county:

    :``POPULATION``: Population in log10 (i.e. 4 means 10,000 people)
    :``INCOME``: Median annual income per household in $1000
    :``AGE``: Median age
    :``UNEMPLOYMENT``: Unemployment rate
    :``BACHELORDEG``: Percentage of population with a bachelor's degree
    :``HOMEOWNERS``: Percentage of population that is homeowner
    :``FEMALE``: Percentage of population that is female
    :``65PLUS``: Percentage of population that is 65 or older
    :``LATINO``: Percentage of population that is Latino
    :``WHITE``: Percentage of population that is white
    :``BLACK``: Percentage of population that is black

:File paths:
    ::

        data/election/training.csv
        data/election/testing.csv