Mean Encoding – Machine Learning

During Feature Engineering the task of converting categorical features into numerical is called Encoding.
There are various ways to handle categorical features like OneHotEncoding and LabelEncoding, FrequencyEncoding or replacing by categorical features by their count. In similar way we can uses MeanEncoding.

Created a DataFrame having two features named subjects and Target and we can see that here one of the features (SubjectName) is Categorical, so we have converted it into the numerical feature by applying Mean Encoding.
Code:

filter_none

edit
close

play_arrow

link
brightness_4
code

# importing libraries
import pandas as pd
  
# creating dataset
data={'SubjectName':['s1','s2','s3','s1','s4','s3','s2','s1','s2','s4','s1'],
      'Target':[1,0,1,1,1,0,0,1,1,1,0]}
  
df = pd.DataFrame(data)
  
print(df)

chevron_right


Output:

     SubjectName  Target
0    s1    1
1    s2    0
2    s3    1
3    s1    1
4    s4    1
5    s3    0
6    s2    0
7    s1    1
8    s2    1
9    s4    1
10    s1    0

Code : Counting every datapoints in SubjectName

filter_none

edit
close

play_arrow

link
brightness_4
code

df.groupby(['SubjectName'])['Target'].count()

chevron_right


Output:



subjectName
 s1         4
 s2         3
 s3         2
 s4         2
Name: Target, dtype: int64

Code: groupby data with SubjectName with their mean according to their positive target value

filter_none

edit
close

play_arrow

link
brightness_4
code

df.groupby(['SubjectName'])['Target'].mean()

chevron_right


Output:

subjectName
s1         0.750000
s2         0.333333
s3         0.500000
s4         1.000000
Name: Target, dtype: float64

The output shows the mean mapped with data point in SubjectName with their positive target value (1-positive and 0-Negative).

Code : Finally assigning the mean value and map with df[‘SubjectName’]

filter_none

edit
close

play_arrow

link
brightness_4
code

Mean_encoded_subject = df.groupby(['SubjectName'])['Target'].mean().to_dict()
  
df['SubjectName'] =  df['SubjectName'].map(Mean_encoded_subject)
  
print(df)

chevron_right


Output : Mean Encoded Data

    SubjectName    Target
0    0.750000    1
1    0.333333    0
2    0.500000    1
3    0.750000    1
4    1.000000    1
5    0.500000    0
6    0.333333    0
7    0.750000    1
8    0.333333    1
9    1.000000    1
10    0.750000    0

Pros of MeanEncoding:

  • Capture information within the label, therefore rendering more predictive features
  • Creates a monotonic relationship between the variable and the target

Cons of MeanEncodig:

  • It may cause over-fitting in the model.



My Personal Notes arrow_drop_up


If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to contribute@geeksforgeeks.org. See your article appearing on the GeeksforGeeks main page and help other Geeks.

Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.


Article Tags :
Practice Tags :


6


Please write to us at contribute@geeksforgeeks.org to report any issue with the above content.