Scikit Learn Pipeline allows to sequentially apply a list of transformers to preprocess the data and, if desired, conclude the sequence with a final predictor for predictive modeling.
Main benefits are:
- It allows us to keep all the definitions and components of our model in one place, which makes it easier to reuse the model or change it in the future.
- We can use grid search and cross-validate all the steps of the model together.
Name | Description |
---|---|
pipeline.Pipeline(steps, *[, memory, verbose]) | A sequence of data transformers with an optional final predictor |
Name | Description |
---|---|
compose.ColumnTransformer(transformers, *[, ...]) | Applies transformers to columns of an array or pandas DataFrame |