Forest Cover Prediction

Disclaimer

Not much planning or forethought went into these scripts. I created them to learn and explore. Until near the end of the project I never thought I would use them again

I make no guarantees nor am I responsible for any harm that may be caused by running these scripts

Pre-requisites

• python 2.7
• pyyaml
• Updated orange:
  I found an issue with the Orange source code. The fix can be found in my fork:
  https://github.com/maroy/orange

To run the classification:

1. Get train.csv and test.csv from Kaggle.
2. To generate the train.tab file:
   python 2tab.py True False train.csv
3. To generate the test.tab file:
   python 2tab.py False False test.csv
4. To run the actual classification:
   python mp_runner.py

• Not much feedback is given during training
  2 warnings were reported during the Neural Network training, I never looked into the cause
• On subsequent runs the pickled classifier will be used and training will not run
  Delete the classifier.pickle file to re-run training

The results will be in a file named out.csv

To check the correct ratio of the output:

python score_classifier.py