kdehaan / ntwist-machine-learning-entry Goto Github PK

Kevin de Haan, 2017



Usage instructions:
    ** NOTE **
    main.py will work (as far as I am aware) on python 2, while model.py will not,
    as input() is not cross-compatible. If you only have python 2 and do not want
    to risk messing up your environment, it is possible to remove the input statements
    from model.py to hard code the files you want to use. 
    That being said, this code was written for python 3 and any python 2 compatibility
    may be considered a happy accident. It is not designed to be cross-compatible.

If required dependencies are installed (on any system), simply run 'python3 main.py' or 'python3 model.py':
    -NumPy
    -SciPy
    -SciKit-learn

On linux, with python >= 3.3 installed:
    extract files to appropriate file location
    navigate to file location
    type 'venv/bin/python3.5 main.py' to create a model
    type 'venv/bin/python3.5 model.py' to test a model

    
On Windows, with python >= 3.3 installed:
    extract files to appropriate file location
    navigate to file location
    type '.\win_env\Scripts\activate'
    this will run the appropriate file for cmd or powershell, whichever you are using
    with the virtual environment active, type 'python3 .\main.py' to create a model
    type 'python3 .\model.py' to test a model

    to exit the virtual environment, type 'deactivate'

    if you encounter a script permission error, it may be bypassed by changing your system settings in powershell:
	'Set-ExecutionPolicy Unrestricted'
    which you can reset afterwards.
    
    
Description of files:
    main.py: model creation and training
    model.py: model testing
    data.csv: provided data
    requirements.txt: exact versions of modules
    results.csv: results of data testing created immediately after model creation
    sample_results.csv: an example of a typical results.csv output
    model_results.csv: results of model testing when target results are provided
    sample_model_results.csv: model_results when run with sample_model.pkl
    predictions.csv: results of model testing when target results are unknown
    sample_predictions.csv: predictions when run with sample_model.pkl
    model.pkl: created when main.py is run
    sample_model.pkl: one of the better models created from main.py
    sample_tests.csv: an example csv with the first column (actual outputs) removed