Several pipelines that merge into the Sklearn pipeline?

Question

Sometimes I design machine learning pipelines that look something like this:

Usually I have to hack these "split" pipelines together using my own "Combined Functions" functions. However, it would be great if I could put this in a Piple Pipeline object. How should I do it? (Pseudo code is fine.)

Answer 1

While “Complete Dataset” means the same functions, this is exactly what FeatureUnion:

make_pipeline(make_union(PolynomialFeatures(), PCA()), RFE(RandomForestClassifier()))

, , , FeatureUnion , . [ , FunctionTransformer(), ]