One-hot encoding

As explained in Data Without Labels

One-hot encoding represents each word or category with a binary vector containing one active position and zeros elsewhere.

One-hot encoding turns categories or words into binary vectors: each distinct value gets its own position, set to 1 for that value and 0 elsewhere. For categorical variables, this creates a column for each distinct class. section 2.7 section 7.13

Figure 2.36 Using one-hot encoding to convert categorical variables into numeric ones

The city example shows categorical values represented in separate binary columns. section 2.7

Encoding a short sentence

For “It is raining heavily,” lowercase and alphabetize the words as heavily, is, it, raining. Encode “it” as [0, 0, 1, 0], “is” as [0, 1, 0, 0], “raining” as [0, 0, 0, 1], and “heavily” as [1, 0, 0, 0]. section 7.7

Encoding can let algorithms such as k-means use categorical inputs numerically. But many distinct categories create many mostly-zero columns; in text, large vocabularies also make vectors high-dimensional and sparse. The representation does not preserve relationships between words, so word embeddings offer an alternative for capturing context. section 2.7 section 7.13

For categorical data, pandas provides get_dummies() to convert variables into numeric columns. Check whether a variable offers useful variation: values that are mostly identical or all distinct may contribute little to analysis. section 11.6

How Data Without Labels covers it

How deep it goes

Defines category columns and word vectors, works through a four-word example, and identifies sparsity and vocabulary-growth limits.

Examples

Uses London and New Delhi categories and a four-word sentence to demonstrate binary columns and vectors.

What it assumes

Introduces binary-vector encoding while assuming categorical variables or a cleaned collection of text words is available.

What you can do after

Apply it via pandas get_dummies() to convert categorical variables into numeric columns.

How it connects

Connects categorical encoding to clustering and word encoding to text feature extraction and word embeddings.

liveBookmanning.com

Read Data Without Labels on liveBook Get the book at manning.com

Also explained in

• 

  liveBookmanning.com
  The Well-Grounded Data AnalystOne-hot encoding transforms categorical values into separate binary indicator variables, with each variable representing one category.
• 

  liveBookmanning.com
  Machine Learning for Tabular Data: XGBoost, Deep Learning, and AIOne-hot encoding represents each category of a categorical variable as a separate binary feature.
• 

  liveBookmanning.com
  Outlier Detection in PythonOne-hot encoding represents each category of a categorical feature as a separate binary indicator feature.
• 

  liveBookmanning.com
  Ensemble Methods for Machine LearningOne-hot vectorization represents each category of a categorical variable as a separate binary indicator component.
• 

  liveBookmanning.com
  Graph Neural Networks in ActionA representation technique that encodes a categorical value as a vector with one active entry and all other entries inactive.

Compare how 6 books explain one-hot encoding →

Related concepts

• labelincludes · A label is a value or annotation associated with an item that identifies its target category, outcome, or reference answer for learning or evaluation.
• categorical datauses · Categorical data consists of values representing membership in a discrete set of named or ordered categories rather than measured numerical magnitudes.
• binary featureproduces · A binary feature is an input variable whose value is restricted to exactly two alternatives.
• categorical encodingkind of · Categorical encoding transforms discrete category values into numerical representations, such as separate binary indicators or vectors with one active position per category.
• indicator variableproduces · An indicator variable is a binary-valued variable representing whether a specified condition or category applies.
• label encodingalternative to · Label encoding is a categorical-data transformation that maps each category to a numerical code, which may imply an ordering among categories.

This explainer was generated by Manning's knowledge graph from the text of Data Without Labels. The section links point to the parts of the book it draws on.