Practical Guide And Tutorial To Label Encoding In Python

by | Sep 4, 2023 | Data Science, Machine Learning

What is label encoding machine learning?

Label encoding is a technique used in machine learning and data preprocessing to convert categorical data (data that consists of categories or labels) into numerical format so that machine learning algorithms can work with it effectively. Each unique category or label is assigned a unique integer value in label encoding. This can be useful when working with algorithms that require numerical input, which is the case for many machine learning models.

How does label encoding work?

  1. Identify Categorical Features: First, you need to identify which features in your dataset are categorical. These are usually non-numeric columns, such as “gender,” “colour,” “city,” etc.
  2. Assign Integer Labels: You assign a unique integer label for each category within a categorical feature. These labels are typically assigned in ascending order starting from 0 or 1.
    • For example, if you have a “colour” feature with categories [“red”, “green”, “blue”], you can label encode it as:
      • “red” → 0
      • “green” → 1
      • “blue” → 2
  3. Replace Categorical Values: Replace the original categorical values in your dataset with their corresponding integer labels. This transforms the categorical feature into a numerical one.
  4. Machine Learning Model: You can now use this transformed dataset with label-encoded features to train your machine learning model.
Label encoding is used for encrypting the colout of a pixel

Label encoding can be used to encode the colour of a pixel (“red”, “green”, “blue”)

Label encoding is suitable for nominal categorical data with no inherent order or ranking among the categories. However, it may not be ideal for ordinal categorical data, where there is a specific order or hierarchy among the categories, as label encoding does not capture this information.

Label encoding with Python

Example with Scikit-Learn (sklearn)

Here’s a Python example using the LabelEncoder class from the scikit-learn library to perform label encoding:

from sklearn.preprocessing import LabelEncoder

# Sample categorical data
colors = ["red", "green", "blue", "red", "green"]

# Initialize the LabelEncoder
label_encoder = LabelEncoder()

# Fit the encoder to the data and transform the categories into integers
encoded_colors = label_encoder.fit_transform(colors)



[2 1 0 2 1]

In this example, “red” is encoded as 2, “green” as 1, and “blue” as 0, as shown in the output. These encoded values can be used as features in a machine learning model.

Example with Pandas

Label encoding can also be performed using the Pandas library in Python, which is especially useful when working with DataFrame objects. Here’s how to implement it using Pandas:

import pandas as pd

# Create a DataFrame with a categorical column
data = {'Category': ["A", "B", "A", "C", "B", "C"]}
df = pd.DataFrame(data)

# Use the pandas factorize method to perform label encoding
df['Category_encoded'] = pd.factorize(df['Category'])[0]

# View the DataFrame with the encoded column


  Category  Category_encoded
0        A                 0
1        B                 1
2        A                 0
3        C                 2
4        B                 1
5        C                 2

In this example, we create a DataFrame with a categorical column ‘Category’ and then use the pd. factorize method to perform label encoding. The encoded values are stored in a new column, ‘Category_encoded.’

You can access the encoded column as a Pandas Series and use it as numerical features in your machine learning models.

Advantages and disadvantages of label encoding

Label encoding is a simple and commonly used technique for handling categorical data in machine learning, but it has its advantages and disadvantages, which are essential to consider:


  1. Simplicity: It is straightforward to implement and understand. It involves converting categories into numerical labels, making it easy to work with for beginners.
  2. Compatibility with Algorithms: Many machine learning algorithms, especially traditional ones like linear regression, decision trees, and support vector machines, require numerical input features. Label encoding provides a way to represent categorical data in a format compatible with these algorithms.
  3. Efficient Memory Usage: It typically uses less memory than other encoding methods like one-hot encoding. This can be important when working with large datasets.


  1. Ordinal Misinterpretation: It assumes an ordinal relationship between the categories, which may not always be true. For nominal categorical data (where there’s no inherent order among categories), label encoding can mislead the algorithm into treating the encoded values as ordinal, potentially leading to incorrect model results.
  2. Arbitrary Numerical Values: The integers assigned during label encoding are random and do not convey meaningful category information. Machine learning algorithms might misinterpret these values as having numerical significance or relationships when there aren’t any.
  3. Impact on Model Performance: It can introduce unintended relationships between categories, especially in algorithms that rely on distance metrics (e.g., k-means clustering). The algorithm may interpret smaller numerical differences as more significant than they should be.
  4. Not Suitable for High Cardinality: When dealing with categorical features with a high number of unique categories, label encoding can lead to a significant increase in dimensionality. This can negatively impact model performance and increase the risk of overfitting.
  5. Loss of Information: It can result in a loss of information about the original categorical data, especially if the encoded values do not represent the true nature of the categories.

When to Use Label Encoding

Label encoding can be a suitable choice when:

  • You are working with nominal categorical data with a clear ordinal relationship among categories (e.g., “low,” “medium,” “high”).
  • The number of unique categories in a feature is reasonably small.
  • You are using machine learning algorithms that can handle ordinal encoding, or you have verified that the encoding does not adversely affect the model’s performance.

However, alternative encoding methods like one-hot encoding or target encoding may be more appropriate for nominal categorical data without a natural ordering or when working with algorithms that might misinterpret the encoded values. The choice of encoding method should be based on the specific characteristics of your data and the requirements of your machine learning model.

What are the alternatives to label encoding?

Label encoding is one way to handle categorical data. Still, it’s not always the best choice, especially when dealing with nominal categorical variables (with no inherent order among categories) or using algorithms that can misinterpret the encoded values. Here are some alternatives:

1. One-Hot Encoding:

  • Each category is transformed into a binary vector in one-hot encoding, where each category becomes a new binary feature (0 or 1).
  • This method is suitable for nominal categorical data as it avoids the ordinal assumptions made by label encoding.
  • One-hot encoding can increase dimensionality, especially when dealing with categorical features with many unique categories, which can be a disadvantage regarding memory and computation.

2. Ordinal Encoding:

  • Ordinal encoding is suitable when there’s a clear ordinal relationship among the categories, meaning they have a natural order or ranking.
  • You assign integer values to the categories based on their order.
  • Be cautious with ordinal encoding, as it assumes an order that may not always be valid.

3. Binary Encoding:

  • Binary encoding combines the advantages of label encoding and one-hot encoding. It represents each category as a binary code.
  • This method helps reduce dimensionality compared to one-hot encoding.

4. Frequency (Count) Encoding:

  • In this technique, you encode categories based on their frequency or count in the dataset.
  • It can help capture the importance of each category in the dataset.

5. Target Encoding (Mean Encoding):

  • Target encoding replaces each category with the mean of the target variable for that category.
  • It is often used in classification tasks and can help capture relationships between the categorical variable and the target.

6. Embedding Layers (for Neural Networks):

  • You can use embedding layers to automatically learn embeddings for categorical features when working with deep learning models.
  • This is common in natural language processing (NLP) and recommendation systems.

7. Hash Encoding:

  • Hash encoding maps categories to a fixed number of bins using a hash function.
  • It can be helpful when dealing with high cardinality categorical data.

The choice of encoding method depends on the nature of your data, the machine learning algorithm you plan to use, and the specific problem you are solving. It’s essential to consider the advantages and disadvantages of each method and carefully preprocess your data to achieve the best results.

Best Practices and Tips for Label Encoding

Label encoding is a helpful technique for converting categorical data into a numerical format, but like any data preprocessing method, it requires careful consideration and attention to detail. Here are some best practices and tips to help you use label encoding effectively:

1. Verify the Ordinal Relationship:

  • Before applying label encoding, make sure that there is a genuine ordinal relationship among the categories. In other words, ensure that the order of the encoded values reflects meaningful differences between the categories. Consider alternative encoding methods like one-hot or target encoding if there’s no inherent order.

2. Handle Missing Values:

3. Consider Scaling and Standardization:

  • After label encoding, consider scaling or standardizing your data if you use algorithms that rely on distances between data points (e.g., k-nearest neighbours). Scaling ensures that numerical values are on a similar scale, preventing features with large values from dominating those with smaller values.

4. Use Feature Engineering Techniques:

  • Sometimes, you may need to create additional features based on label-encoded data. For example, you can calculate statistical measures (e.g., mean, median, variance) for numerical features within each category. These new features can capture valuable information from the original categorical data.

5. Avoid Data Leakage:

  • Be cautious when encoding categorical data before splitting your dataset into training and testing sets. Encoding the entire dataset before splitting can lead to data leakage, where information from the test set inadvertently influences the training process. To avoid this, apply it separately to the training and test sets after the split.

6. Consider Model Interpretability:

  • Keep in mind that label encoding may affect the interpretability of your model. If interpretability is essential, explore alternative encoding methods like target or binary encoding, which can preserve the original meaning of categorical variables to a greater extent.

7. Choose the Right Encoding Method:

  • While label encoding is suitable for some scenarios, remember there is no one-size-fits-all solution. Evaluate your dataset, the nature of your categorical features, and the requirements of your machine learning model to determine the most appropriate encoding method. Consider both the advantages and disadvantages of label encoding in your decision-making process.

8. Documentation and Communication:

  • Document the encoding process in your data preprocessing pipeline. Ensure that your team or collaborators understand how categorical data has been encoded. This documentation helps maintain consistency and transparency in your machine learning project.

By following these best practices and tips, you can effectively use label encoding while minimizing potential pitfalls and ensuring that your encoded data suits the machine-learning algorithms you intend to use. Remember thoughtful data preprocessing is critical in building accurate and reliable machine learning models.


Dealing with categorical data is a common and critical challenge in machine learning. Label encoding is a valuable tool in your data preprocessing toolkit, offering a straightforward way to convert categorical data into a numerical format. However, as explored in this blog post, it’s essential to use label encoding judiciously and be aware of its advantages and limitations.

Label encoding shines when there is a clear ordinal relationship among the categories and is compatible with many machine learning algorithms. Its simplicity and efficiency make it an attractive choice for specific datasets and scenarios.

However, it’s equally important to acknowledge the drawbacks. Most notably, label encoding assumes an ordinal relationship even when there isn’t one, which can lead to misinterpretations by machine learning algorithms. Additionally, it may not be suitable for nominal categorical data or when model interpretability is a priority.

As a data practitioner, your role is to apply techniques like label encoding and make informed decisions about the most appropriate encoding method for your specific dataset and machine learning task. Sometimes, alternative encoding methods like one-hot encoding, target encoding, or binary encoding may be better suited to preserve the integrity of your data and enhance model performance.

In conclusion, mastering data preprocessing techniques like label encoding is essential to building robust and accurate machine learning models. Remember to exercise caution, consider the nature of your data, and choose the encoding method that aligns with your modelling goals. With the right approach, you can harness the power of categorical data and unlock the potential of your machine learning projects.

About the Author

Neri Van Otten

Neri Van Otten

Neri Van Otten is the founder of Spot Intelligence, a machine learning engineer with over 12 years of experience specialising in Natural Language Processing (NLP) and deep learning innovation. Dedicated to making your projects succeed.

Recent Articles

online machine learning process

Online Machine Learning Explained & How To Build A Powerful Adaptive Model

What is Online Machine Learning? Online machine learning, also known as incremental or streaming learning, is a type of machine learning in which models are updated...

data drift in machine learning over time

Data Drift In Machine Learning Explained: How To Detect & Mitigate It

What is Data Drift Machine Learning? In machine learning, the accuracy and effectiveness of models heavily rely on the quality and consistency of the data on which they...

precision and recall explained

Classification Metrics In Machine Learning Explained & How To Tutorial In Python

What are Classification Metrics in Machine Learning? In machine learning, classification tasks are omnipresent. From spam detection in emails to medical diagnosis and...

example of a co-occurance matrix for NLP

Co-occurrence Matrices Explained: How To Use Them In NLP, Computer Vision & Recommendation Systems [6 Tools]

What are Co-occurrence Matrices? Co-occurrence matrices serve as a fundamental tool across various disciplines, unveiling intricate statistical relationships hidden...

use cases of query understanding

Query Understanding In NLP Simplified & How It Works [5 Techniques]

What is Query Understanding? Understanding user queries lies at the heart of efficient communication between humans and machines in the vast digital information and...

distributional semantics example

Distributional Semantics Simplified & 7 Techniques [How To Understand Language]

What is Distributional Semantics? Understanding the meaning of words has always been a fundamental challenge in natural language processing (NLP). How do we decipher...

4 common regression metrics

10 Regression Metrics For Machine Learning & Practical How To Guide

What are Evaluation Metrics for Regression Models? Regression analysis is a fundamental tool in statistics and machine learning used to model the relationship between a...

find the right document

Natural Language Search Explained [10 Powerful Tools & How To Tutorial In Python]

What is Natural Language Search? Natural language search refers to the capability of search engines and other information retrieval systems to understand and interpret...

the difference between bagging, boosting and stacking

Bagging, Boosting & Stacking Made Simple [3 How To Tutorials In Python]

What is Bagging, Boosting and Stacking? Bagging, boosting and stacking represent three distinct ensemble learning techniques used to enhance the performance of machine...


Submit a Comment

Your email address will not be published. Required fields are marked *

nlp trends

2024 NLP Expert Trend Predictions

Get a FREE PDF with expert predictions for 2024. How will natural language processing (NLP) impact businesses? What can we expect from the state-of-the-art models?

Find out this and more by subscribing* to our NLP newsletter.

You have Successfully Subscribed!