LABEL ENCODING

Transforming in a numerical quantity the features that represent categories.

there are 2 types of encoders

ENCODERS APPLIED TO PREDICTORS

• ONE HOT ENCODER

  It transforms features in a set of columns that integrate the predictor X:

  • it counts all the categories within the features;
  • it replaces the analyzed features with as many columns as there are categories;
  • all the new columns values will be 0 or 1. For every sample (X’s rows), it will be put the 1 value only if the column corresponds to the represented category.

  here a usage example:

  from sklearn.preprocessing import OneHotEncoder
  ohe = OneHotEncoder() # creating object
  ohe.fit(X) # fit the data
  ohe.categories_ # show categories founded
  ohe.transform(X) # apply the transformation

• ORDINAL ENCODER

  • it locates all the categories available;
  • it assignes an incremental value to every category
  • a single column with the corresponding incremental values is returned

  here a usage example:

from sklearn.preprocessing import OrdinalEncoder
oe = OrdinalEncoder() # creating object
oe.fit(X) # fit the data
oe.categories_ # show categories founded
oe.transform(X) # apply the transformation

ENCODERS APPLIED TO TARGET

The most famous is LabelEncoder. It’s similar to OrdinalEncoding, but it’s applied to the targed instead of the predictor.

here a usage example:

from sklearn.preprocessing import LabelEncoding
le = LabelEncoding() # creating object
le.fit(y) # fit the data
le.classes # show classes founded
le.transform(y) # apply the transformation