This PR allows for projection into LSI spaces, as planned with an sklearn-esque interface. We port over both variable feature selection, and LSI. fit(), project() and short_description() methods have been created for both these steps.
Currently the implementations utilize a lot of the previously created LSI implmementation. I am refraining from moving that over until review is concluded. However, this PR already contains some initial tests, and a demonstration that we can concat two PipelineSteps together, to fit and project.
Tests
Tests were created to make sure initialization, as well as fit() and project() methods work for variable feature selection and LSI.
Tests were also created to make sure pipelines can be created between these two individual steps, with fit() and project() working from the Pipeline implementations of these two methods
Details
This PR allows for projection into LSI spaces, as planned with an sklearn-esque interface. We port over both variable feature selection, and LSI.
fit(),project()andshort_description()methods have been created for both these steps.Currently the implementations utilize a lot of the previously created LSI implmementation. I am refraining from moving that over until review is concluded. However, this PR already contains some initial tests, and a demonstration that we can concat two
PipelineStepstogether, to fit and project.Tests
fit()andproject()methods work for variable feature selection and LSI.fit()andproject()working from thePipelineimplementations of these two methods