Instructions for running our code:

All code is written in python, and is to be run from the command line.

First, a metadata file must be created. This is a simple, one-line CSV file, with one comma-separated value for each column in the dataset. The values describe each column as one of {numeric, categorical, class}. The outcome / predicted column should be "class", and there should only be one of these.

Creating a Tree
================
For this, use the decision_tree.py script, from the command line. There are two required command-line options, and three optional. The training data should be passed with the -d flag, and the metadata after the -m flag. To test the tree against a validation set, pass the validation data after the -v flag. The accuracy will be printed to the terminal. If the validation set should be used for pruning, add the -p flag.

By default, the tree will be saved as a Python pickle file named finished_tree.pkl in the current directory. To change this, pass the -o flag.

Example: python decision_tree.py -d ../btrain.csv -v ../bvalidate.csv -m ../metadata.csv -o pruned_tree.pkl -p

This example would use the btrain.csv data to train on, would prune that tree based on the bvalidate.csv data, and would create a file named pruned_tree.pkl.

Disjunctive Normal Form
=======================
To convert a tree into disjunctive normal form, use the tree_to_dnf.py script. Pass the pickle version of the decision tree after the -d flag, and the location of the output file after the -o flag.

Example: python tree_to_dnf.py -d pruned_tree.pkl -o pruned_tree_DNF.txt

Predicting Test Data
====================
To predict the test data, run the label_test_data.py script, passing the test data after the -t flag, the pickled version of the tree after the -d flag, and the output file after the -o flag.

Example: python label_test_data.py -t ../btest.csv -d pruned_tree.pkl -o btest_predicted.csv

Generating Learning Curves
=========================

Run the decision_tree.py method using the --plot flag and passing in training data to create a tree with and validation data to test on. Use the -p flag to use pruning. The curve is saved to a file with the name given with the --plot_file flag.

Example: 
python decision_tree.py -d btrain.csv -m metadata.csv -v bvalidate.csv --plot --plot_file prune.png -p