To execute this program, download the feature-finder files and run the following from your command line:
virtualenv -p python3 venv; . ./venv/bin/activate
python3 setup.py install
feature-finder <model type> <path/to/data.csv> <index or name of y-column> <optional: --header>
For example:
feature-finder logistic feature_finder/tests/sample_data/01_feature_logistic.csv 1
The following help menu is available via feature-finder --help.
usage: feature-finder [-h] [--header] [--plugins PLUGINS [PLUGINS ...]]
model data y
positional arguments:
model Model type: linear or logistic.
data Data in CSV format.
y y-value of column (index position or header name).
optional arguments:
-h, --help show this help message and exit
--header Data CSV contains header row.
--plugins PLUGINS [PLUGINS ...]
Use plugin(s) defined in plugins directory to add or
clean up feature columns.
Your CSV should include at least one x-column with numerical data for input as a dependent variable into the model, and one y-column with either quantitative or categorical data. e.g.
1, 22.2, 30
0, 20.5, 28
0, 30, 33.5
Plugins are customizations that can be added to the plugins/plugins.py file. A sample plugin has been included for reference. Plugins must be python functions that take in a pandas DataFrame object of the original CSV, which it is allowed to modify.
The plugin must return a pandas DataFrame that includes the features whose error/accuracy should be calculated.
Modify your data CSV with plugins using the following input format:
feature-finder <model type> <path/to/data.csv> <index or name of y-column> <optional: --header> --plugins <plugin1 plugin2 etc>
Tests can be run from the feature-finder directory with nosetests ..