Anyone have opinions on best practices around using scikit-learn pipelines as Kedro nodes in Kedro pipelines? One of my concerns is that scikit-learn pipelines may obscure information from the kedro-viz, making it more difficult to use kedro-viz to quickly understand what is in the project. https://scikit-learn.org/stable/modules/generated/sklearn.pipeline.Pipeline.html

Indeed, a big scikit-learn pipeline will be seen from Kedro as just one node. It depends on how much granularity do you want. Usually a pipeline = “a model”, so it should be fine to treat it as a black box - but this might not work in all cases of course