Signature
pd.get_dummies(data, columns=None, drop_first=False, dtype=None)What it does
get_dummies is a fast pandas shortcut for one-hot encoding. For production pipelines prefer scikit-learn OneHotEncoder so the training categories, unknown handling, and column order are stored with the model.
Parameters
dataDataFrame or Series- Data containing categorical columns.
columnslist, optionaloptional- Columns to encode; default encodes every categorical column.
drop_firstbool, optionaloptional- Drop one level to avoid perfect collinearity.
Returns
DataFrame
Example
pythonpd.get_dummies example
import pandas as pd
df = pd.DataFrame({"team": ["red", "blue", "red"]})
print(pd.get_dummies(df, columns=["team"]))Output
team_blue team_red 0 False True 1 True False 2 False True
Watch out for
- Encoding train and test separately can produce different columns.
- drop_first changes interpretation; keep it consistent between fit and inference.