Python discrete feature encoding

There are two kinds of discrete feature coding, which have the meaning of size and character.

1, the characteristic does not have the size significance direct single-heat code

2, the characteristics of the size of the significance of the use of mapping code

[Python]View PlainCopy

2. Import Pandas as PD
4. DF = PD. DataFrame ([
6. [' green ', ' M ', 10.1, ' Label1 '],
8. [' Red ', ' L ', 13.5, ' Label2 '],
10. [' Blue ', ' XL ', 15.3, ' Label2 ']]
12. # color, label does not have size meaning, size has significance
14. Df.columns = [' color ', ' size ', ' length ', ' label ']
16. Df

[Python]View PlainCopy

2. Size_mapping = {
4. ' XL ': 3,
6. ' L ': 2,
8. ' M ': 1}
10. df[' size ' = df[' size '].map (size_mapping)
14. label_mapping = {Lab:idx for Idx,lab in enumerate (set (df[' label '))}
16. df[' label '] = df[' label '].map (label_mapping)
18. Df

Direct use of functions for single-hot coding

does not differentiate whether it has a size meaning

[Python]View PlainCopy

2. Import Pandas as PD
4. DF = PD. DataFrame ([
6. [' green ', ' M ', 10.1, ' Label1 '],
8. [' Red ', ' L ', 13.5, ' Label2 '],
10. [' Blue ', ' XL ', 15.3, ' Label2 ']]
12. # color, label does not have size meaning, size has significance
14. Df.columns = [' color ', ' size ', ' length ', ' label ']
16. Pd.get_dummies (DF)

Get_dummies usage:

[Python]View PlainCopy

1. Import Pandas as PD
2. s = PD. Series (List (' ABCA '))
3. Pd.get_dummies (s)

[Python]View PlainCopy

1. DF = PD. DataFrame ({' a ': [' a ', ' B ', ' a '], ' B ': [' B ', ' a ', ' C '],
2. ' C ': [1, 2, 3]}

[Python]View PlainCopy

1. Pd.get_dummies (DF, prefix=[' col1 ', ' col2 '])

Python discrete feature encoding