How to Use Label Encoding with sklearn in Python

Use LabelEncoder from sklearn.preprocessing to convert categorical labels into numeric values. Fit the encoder with fit() or fit_transform() on your label data, then transform labels to numbers with transform().

Syntax

The main steps to use LabelEncoder are:

• from sklearn.preprocessing import LabelEncoder: Import the encoder.
• encoder = LabelEncoder(): Create an encoder object.
• encoder.fit(labels): Learn the mapping from labels to numbers.
• encoded_labels = encoder.transform(labels): Convert labels to numbers.
• Or use encoded_labels = encoder.fit_transform(labels) to fit and transform in one step.

python

from sklearn.preprocessing import LabelEncoder

encoder = LabelEncoder()
encoder.fit(labels)  # Learn label mapping
encoded_labels = encoder.transform(labels)  # Convert labels to numbers

# Or combine:
encoded_labels = encoder.fit_transform(labels)

Example

This example shows how to encode a list of fruit names into numbers using LabelEncoder. It fits the encoder and transforms the labels, then prints the original and encoded labels.

python

from sklearn.preprocessing import LabelEncoder

labels = ['apple', 'banana', 'apple', 'orange', 'banana', 'orange']

encoder = LabelEncoder()
encoded_labels = encoder.fit_transform(labels)

print('Original labels:', labels)
print('Encoded labels:', encoded_labels.tolist())

Output

Original labels: ['apple', 'banana', 'apple', 'orange', 'banana', 'orange']
Encoded labels: [0, 1, 0, 2, 1, 2]

Common Pitfalls

Common mistakes when using LabelEncoder include:

• Trying to encode features with multiple columns instead of just labels.
• Using fit_transform on training data but only transform on test data without fitting first.
• Assuming encoded numbers have order or magnitude meaning (they do not).

Always fit the encoder on training labels and then transform test labels to avoid errors.

python

from sklearn.preprocessing import LabelEncoder

# Wrong: fitting separately on train and test labels
train_labels = ['cat', 'dog', 'cat']
test_labels = ['dog', 'cat']

encoder = LabelEncoder()
train_encoded = encoder.fit_transform(train_labels)
# This will cause issues if test labels have unseen classes
# test_encoded = encoder.fit_transform(test_labels)  # Wrong

# Right: fit on train, transform on test
encoder = LabelEncoder()
encoder.fit(train_labels)
test_encoded = encoder.transform(test_labels)

Quick Reference

Method	Description
fit(labels)	Learn label to number mapping from labels
transform(labels)	Convert labels to numbers using learned mapping
fit_transform(labels)	Fit and transform labels in one step
inverse_transform(numbers)	Convert numbers back to original labels

Key Takeaways

Use sklearn's LabelEncoder to convert categorical labels to numeric values easily.

Always fit the encoder on training labels before transforming test labels.

Label encoding assigns arbitrary numbers without implying order or magnitude.

Use fit_transform() for quick encoding when training and transform() for new data.

Avoid encoding multi-column features with LabelEncoder; it is for single label arrays.