Categorical Features
← Back to Feature Engineering
Encoding non-numeric categorical data into numeric representations that models can process.
Techniques
- One-Hot Encoding — binary column per category, sparse for high cardinality
- Label Encoding — integer assignment, implies ordering
- Target Encoding — replace category with mean of target variable
- Embeddings — learned dense representations (for neural networks)
Related
- Numerical Features (numeric transformations)
- Text Features (text as categorical)