A categorical variable, as the name suggests, is used to represent categories or labels. For instance, a categorical variable could represent major cities in the world, the four seasons in a year, or the industry (oil, travel, technology) of a company. […] The categories of a categorical variable are usually not numeric. [..] Thus, an encoding method is needed to turn these non-numeric categories into numbers.¹

Both Pandas and scikit-learn propose encoders to deal with categorical variables. But the distinction between each technique and implementation is not obvious. This is what I will try to clarify it in this article.

Basically we can distinct two kinds of encoder:

• Encode labels (categorical variables) into numeric variables: Pandas factorize and scikit-learn LabelEncoder. The result will have 1 dimension.
• Encode categorical variable into dummy/indicator (binary) variables: Pandas get_dummies and scikit-learn OneHotEncoder. The result will have n dimensions (or n-1 dimensions), one by distinct value of the encoded categorical variable.

The main difference between pandas and scikit-learn encoders is that they are made to be used in scikit-learn pipelines with fit and transform methods.

Encode labels into numerical variables

Pandas factorize and scikit-learn LabelEncoder belong to the first category. They can be used to encode labels (nonnumerical variables) to numerical variables.

from sklearn import preprocessing
# Test data
df = DataFrame(['A', 'B', 'B', 'C'], columns=['Col'])
df['Fact'] = pd.factorize(df['Col'])[0]
le = preprocessing.LabelEncoder()
df['Lab'] = le.fit_transform(df['Col'])

#   Col  Fact  Lab
# 0   A     0    0
# 1   B     1    1
# 2   B     1    1
# 3   C     2    2

But pay attention since converting nonnumerical variables to numbers is not the end of the road.

The values may be represented numerically. However, unlike other numeric variables, the values of a categorical variable cannot be ordered with respect to one another. (Oil is neither greater than nor less than travel as an industry type.) They are called non-ordinal.$^1$

Encode categorical variable into dummy/indicator (binary) variables

Pandas get_dummies and scikit-learn OneHotEncoder can be used to create binary variables. OneHotEncoder can only be used with categorical integers while get_dummies can be used with other type of variables. Another difference is that they refer to two feature engineering techniques:

• One-hot encoding: It uses k bit to encode k values. It is implemented by both OneHotEncoder and get_dummies.
• Dummy coding: It uses k-1 bit to encode k values. It is implemented only by get_dummies.

df = DataFrame(['A', 'B', 'B', 'C'], columns=['Col'])
df = pd.get_dummies(df)

#    Col_A  Col_B  Col_C
# 0    1.0    0.0    0.0
# 1    0.0    1.0    0.0
# 2    0.0    1.0    0.0
# 3    0.0    0.0    1.0

We can see here that 3 bits are required to encode 3 distinct values where the variable itself needs only 2 bits (k-1 bits). In this case the extra feature is dropped (A) thanks to the parameter drop_first and so it is represented implicitly by all 0.

This is known as the reference category.

df = DataFrame(['A', 'B', 'B', 'C'], columns=['Col'])
df = pd.get_dummies(df, drop_first=True)

# Col_B  Col_C
# 0      0      0
# 1      1      0
# 2      1      0
# 3      0      1

It’s trickier to do the same thing with scikit-learn since data has to be converted first to numeric before using the OneHotEncoder.

from sklearn.preprocessing import OneHotEncoder, LabelEncoder

df = DataFrame(['A', 'B', 'B', 'C'], columns=['Col'])
# We need to transform first character into integer in order to use the OneHotEncoder
le = preprocessing.LabelEncoder()
df['Col'] = le.fit_transform(df['Col'])
enc = OneHotEncoder()
df = DataFrame(enc.fit_transform(df).toarray())

#      0    1    2
# 0  1.0  0.0  0.0
# 1  0.0  1.0  0.0
# 2  0.0  1.0  0.0
# 3  0.0  0.0  1.0

Note: This post is an augmented version of my Stack Overflow answer²