License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Simple feature selection - Pandas and Scipy

Imports



In [2]:

    
import pandas as pd              # pandas for handling mixed data sets 
import numpy as np               # numpy for basic math and matrix operations

# scipy for stats and more advanced calculations
from scipy.stats import chi2_contingency

Perform simple feature selection

Create a sample data set



In [3]:

    
scratch_df = pd.DataFrame({'x1': pd.Series(np.arange(0, 10)),
                           'x2': pd.Series(np.random.randn(10)), 
                           'x3': ['A', 'A', 'A', 'A', 'A', 'B', 'B', 'B', 'B', 'B'],
                           'x4': ['C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L'],
                           'y' : [0, 0, 0, 0, 0, 1, 1, 1, 1, 1]})

scratch_df

Calculate Pearson correlation for numeric variables

pandas.DataFrame.corr() function shows that x1 is much more correlated with y than x2.



In [4]:

    
scratch_df.corr()

Calculate Chi-Square statistic for categorical variables

pandas.crosstab() creates frequency tables
scipy.stats.chi2_contingency() function on the contingency tables shows that the frequency of values in x3 is related to y more so than the frequency of values in x4.



In [5]:

    
pd.crosstab(scratch_df.x3, scratch_df.y)



In [6]:

    
chi2, p, dof, ex = chi2_contingency(pd.crosstab(scratch_df.x3, scratch_df.y))
print('chi2 =', chi2)
print('p-value =', p)









    



chi2 = 6.4
p-value = 0.011412036386



In [7]:

    
pd.crosstab(scratch_df.x4, scratch_df.y)



In [8]:

    
chi2, p, dof, ex = chi2_contingency(pd.crosstab(scratch_df.x4, scratch_df.y))
print('chi2 =', chi2)
print('p-value =', p)









    



chi2 = 10.0
p-value = 0.350485212323

	x1	x2	x3	x4	y
0	0	-0.965284	A	C	0
1	1	0.278069	A	D	0
2	2	0.223738	A	E	0
3	3	0.158793	A	F	0
4	4	-0.433190	A	G	0
5	5	-0.368828	B	H	1
6	6	-0.947958	B	I	1
7	7	-0.873526	B	J	1
8	8	0.820806	B	K	1
9	9	-0.755244	B	L	1

	x1	x2	y
x1	1.000000	-0.089896	0.870388
x2	-0.089896	1.000000	-0.234613
y	0.870388	-0.234613	1.000000