Signature
Pipeline(steps)What it does
A Pipeline applies each step in order, ending with a final estimator. Fitting the pipeline fits every transformer and the model on the same training rows, which keeps preprocessing inside cross-validation and prevents leakage.
Parameters
stepslist of (name, estimator) tuples- Ordered transforms followed by a final estimator.
Returns
Pipeline
Example
pythonPipeline example
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LinearRegression
model = Pipeline([
("scale", StandardScaler()),
("regress", LinearRegression()),
])
model.fit(X_train, y_train)Watch out for
- Intermediate steps must implement fit and transform; only the final step needs predict.
- Step names must be unique and cannot contain a double underscore.