nv-morpheus / Morpheus

Morpheus SDK
Apache License 2.0
309 stars 119 forks source link

[FEA]: Automated Feature Selection in DFP and building DataFrame schemas from JSON definition files #1685

Open dnandakumar-nv opened 2 months ago

dnandakumar-nv commented 2 months ago

Is this a new feature, an improvement, or a change to existing functionality?

New Feature

How would you describe the priority of this feature request

Low (would be nice)

Please provide a clear description of problem this feature solves

I want Morpheus to be able to provide initial heuristics and recommendations of what features would be a good starting point for the DFP workflow.

I also want to be able to use different input datasets into DFP without having to change the source code of the workflow to accommodate difference schemas. It would be helpful if instead, I could provide a schema definition file as config, from which the schemas are built.

Describe your ideal solution

Two new classes, the first being an AutoEncoderFeatureSelector class that takes in input data and some configurable parameters for feature selection and performs some basic checks to see which features might be helpful in the DFP workflow.

Second, a JSONSchemaBuilder class that takes an input JSON (dictionary) schema definition for Morpheus DataFrameInputSchema objects and builds those objects dynamically at runtime.

Additional context

No response

Code of Conduct