Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
Description
The predict functions for MLP and SVM have a parameter named id in it, and the name of the independent variable is inferred from the model summary table. This poses a serious limitation; test data table is forced to follow a schema, where the independent variable column has the same name as what was in the training data table. This can cause issues when the consumer of an MLP/SVM model is not the producer of the model.
Acceptance:
1. Add a new parameter for independent variable.
2. Evaluate if id column from the current predict interface is required (think about grouping too.)
3. Find other modules that may have this issue, and raise new JIRAs if need be.