Implementation of Geometric Distance Based Feature Selection

This package is intended for use with 'geometric distance based feature selection'. For more information on this feature selection, please refer to 'https://github.com/seo-young-kim/GeometricDistanceBasedFeatureSelection'

pip install gdbfs 

How to Use

Currently, only functions implemented in the sequential forward selection method exist. Therefore, import sfs of the package and use it with fit() function.

from gdbfs import sfs

fit()

arguments

1. the entire dataframe for the data.
2. the target columns name.

return
Print the process of selecting with added feature name and evaluated value.
return a subset of the selected features using the gdbfs.

for a instance, we can select best features int iris problem as follows:

1. make dataframe that contains all information (data,target)

from sklearn.datasets import load_iris
import pandas as pd

data = load_iris()
df = pd.DataFrame(data.data)
df['result']=data.target

2. call fit with two arguments

sfs.fit(df,'result')

3. result

print

   feature to add     gdbfs
0               2  2.215350
1               0  2.969618
2               3  2.521875
3               1  2.533693

return

array([[2],
       [0]], dtype=int64)

seoykim996@gmail.com