Using feature selector as part of sklearn pipeline

[jtlz2]

Awesome package, thanks!

I'm trying to use the feature-selector transformer within a sklearn pipeline but keep getting errors like

AssertionError: The index of X and y need to be the same

Now, this raises a few questions for me:

1. Although https://github.com/blue-yonder/tsfresh/blob/main/notebooks/examples/02%20sklearn%20Pipeline.ipynb mentions the feature augmenter, it does not give the exact syntax for using the feature selector in a pipeline step. What is the syntax precisely?
2. Presumably feature selection should only be carried out on the CV dataset - how do I ensure this in the context of a sklearn pipeline?
3. This raises another spectre - I want to perform (tsfresh's) feature selection on tsfresh features, while simultaneously fitting non-tsfresh-derived features. Is this even possible and if so how can we make it work?

Thanks again!