Decision Tree Classifier
Author: Dane Larsen

DEPENDENCIES:

Requires c99
Only tested on linux, but _should_ work on everything

BUILDING:

If on *nix, run `make`, otherwise take a look at the Makefile

RUNNING:

./dt_main [entropy|gini] [prune|noprune] <train csv> <validate csv> <test csv> <prediction output>

Parameters:
    [entropy|gini]    - choose the splitting metric, either information gain
                        (entropy) or population diversity (gini). In general,
                        population diversity performs better

    [prune|noprune]   - either prune the trained decision tree, or don't prune it.
                        pruning can sometimes improve accuracy, and can often
                        greately improve classification speed and memory usage.

    <train csv>       - the csv with training data, assumes that the last column is
                        the Y values

    <validation csv>  - used to score the trained tree, and used to prune if
                        requested, also assumes that the last column is y values

    <test csv>        - the test set without y values, will be used to generate
                        the prediction set

    <prediction file> - the file to write the final predictions to