LabelEncoder specify classes in DataFrame

问题:

I’m applying a LabelEncoder to a pandas DataFrame, df

Feat1  Feat2  Feat3  Feat4  Feat5
  A      A      A      A      E
  B      B      C      C      E
  C      D      C      C      E
  D      A      C      D      E

I'm applying a label encoder to a dataframe like this -

from sklearn import preprocessing
le = preprocessing.LabelEncoder()
intIndexed = df.apply(le.fit_transform)

This is how the labels are mapped

A = 0
B = 1
C = 2
D = 3
E = 0

I'm guessing that E isn't given the value of 4 as it doesn't appear in any other column other than Feat 5 .

I want E to be given the value of 4 - but don't know how to do this in a DataFrame.

回答1:

You could fit the label encoder and later transform the labels to their normalized encoding as follows:

In [4]: from sklearn import preprocessing
   ...: import numpy as np

In [5]: le = preprocessing.LabelEncoder()

In [6]: le.fit(np.unique(df.values))
Out[6]: LabelEncoder()

In [7]: list(le.classes_)
Out[7]: ['A', 'B', 'C', 'D', 'E']

In [8]: df.apply(le.transform)
Out[8]: 
   Feat1  Feat2  Feat3  Feat4  Feat5
0      0      0      0      0      4
1      1      1      2      2      4
2      2      3      2      2      4
3      3      0      2      3      4

---

One way to specify labels by default would be:

In [9]: labels = ['A', 'B', 'C', 'D', 'E']

In [10]: enc = le.fit(labels)

In [11]: enc.classes_                       # sorts the labels in alphabetical order
Out[11]: 
array(['A', 'B', 'C', 'D', 'E'], 
      dtype='<U1')

In [12]: enc.transform('E')
Out[12]: 4

回答2:

You can fit and transform in single statement, Please find the code for encoding single column and assigning back to data frame.

df[columnName] = LabelEncoder().fit_transform(df[columnName])