How do you handle categorical variables in machine learning models?

Categorical variables require encoding techniques to convert text/categories into numerical formats suitable for algorithms.\n\n• One-hot encoding: Binary columns for each category, good for nominal data\n• Label encoding: Numerical mapping, suitable for ordinal data\n• Target encoding: Mean target value per category, handles high cardinality\n• Embedding: Dense representations, effective for deep learning\n\nExample: Product category with 50 unique values uses target encoding to avoid curse of dimensionality from one-hot encoding. Validate with cross-validation to prevent overfitting and compare performance against other encoding methods.