philipkalinda / geneticfs

GeneticFS is a library for feature selection in Machine Learning using a Genetic Algorithm as an optimisation method.

GeneticFS

GeneticFS is a library for feature selection in Machine Learning using a Genetic Algorithm as an optimisation method. More about the development and tests related to the package are explained HERE

New Features!

• Replaced accuracy score for classification models with F1 score
• Added hyperparameter descriptions to improve ease of use

Tech

GeneticFS uses a number of open source projects to work properly:

• Sci-Kit Learn
• Numpy
• Matplotlib

And of course GeneticFS itself is open source with a public repository on GitHub.

Installation

Install the dependencies

$ pip install numpy
$ pip install matplotlib
$ pip install sklearn

Install the package

$ pip install geneticfs

How to Use

Regression Model:

from geneticfs import GeneticFS
from sklearn.linear_model import LinearRegression

lin_model = LinearRegression()
gfs = GeneticFS()

# fit the optimizer
gfs.fit(model=lin_model, _type='regression', X=X, y=y) # regression model

# get results output
binary_output_of_optimal_variables, indicies_of_optimal_variables = gfs.results()

# plot results of progress
gfs.plot_progress()

Classification Model:

from geneticfs import GeneticFS
from sklearn.linear_model import LogisticRegression

log_model = LogisticRegression()
gfs = GeneticFS()

# fit the optimizer
gfs.fit(model=log_model, _type='classification', X=X, y=y) # classification model

# get results output
binary_output_of_optimal_variables, indicies_of_optimal_variables = gfs.results()

# plot results of progress
gfs.plot_progress()

Todos

• Write more Tests
• Add more flexibility in use

License

MIT

geneticfs's Issues

do you have example with complicated task when strong ML packages needed like lightGBM?

do you have example with complicated task when strong ML packages needed like lightGBM?