See the three new png test results, and the last cell of the modified doc/...FICO... notebook.
This draft adds the multi-plot to BinningProcess, and modifies the BinningTable.plot method to suit. We need to add the same sort of thing to the other single-var binners, ContinuousBinningTable.plot and MulticlassBinningTable, and reconcile any differences in how those three operate.
I'm not so familiar with the multidimensional, piecewise, or streaming binners; maybe something needs/should be done there too.
Other TODOs:
write something to select the subplots grid shape; there's probably something we can borrow from pandas for this
I'd really like to be able to share the y-axis of the response aggregates (WoE etc.); this would surface a quick feature importance.
(I wonder whether we can refactor a bit to make the various flavors share more code, and make this easier?...)
Closes #315
See the three new png test results, and the last cell of the modified
doc/...FICO...
notebook.This draft adds the multi-plot to
BinningProcess
, and modifies theBinningTable.plot
method to suit. We need to add the same sort of thing to the other single-var binners,ContinuousBinningTable.plot
andMulticlassBinningTable
, and reconcile any differences in how those three operate.I'm not so familiar with the multidimensional, piecewise, or streaming binners; maybe something needs/should be done there too.
Other TODOs:
(I wonder whether we can refactor a bit to make the various flavors share more code, and make this easier?...)