馃殌 Feature Request
Currently Cover classes expose a fit and a search method. However, it would be useful to also have a transform method that works similarly as scikit-learn transformers: you call fit on a training set and call transform on the test set.
馃攳 Motivation
The main benefit is that we can construct the Mapper graph on a training set and then check if a test set matches the graph created on the train set.
馃 Algorithm Details (Optional)
The implementation would require that landmarks are somehow stored in the fit method, and then used in the transform method. The additional landmarks needed to cover the test set should be consistent with the ones already created.
馃殌 Feature Request
Currently Cover classes expose a
fitand asearchmethod. However, it would be useful to also have atransformmethod that works similarly asscikit-learntransformers: you callfiton a training set and calltransformon the test set.馃攳 Motivation
The main benefit is that we can construct the Mapper graph on a training set and then check if a test set matches the graph created on the train set.
馃 Algorithm Details (Optional)
The implementation would require that landmarks are somehow stored in the
fitmethod, and then used in thetransformmethod. The additional landmarks needed to cover the test set should be consistent with the ones already created.