sklearn set_output

[mglisse]

Since version 1.2, scikit-learn has a set_output API that can make transformers output pandas dataframes.

• Investigate how that affects us.

[mglisse]

@martinroyer suggests adding to RipsPersistence

def get_feature_names_out(self):
    return [f"H{i}" for i in self.homology_dimensions]

so that RipsPersistence(homology_dimensions=[0, 2]).set_output(transform="pandas").fit_transform([X]) can output a pandas dataframe with column labels "H0" and "H2".