feat: add support to sklearn TargetEncoder #1137
Conversation
|
Thanks for the contribution. One line should be removed. Everything else looks good. |
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| [("input", StringTensorType([None, X.shape[1]]))], | ||
| target_opset=TARGET_OPSET, | ||
| ) | ||
| self.assertTrue(model_onnx is not None) |
Check notice
Code scanning / CodeQL
Imprecise assert Note test
| target_opset=TARGET_OPSET, | ||
| ) | ||
| self.assertTrue(model_onnx is not None) | ||
| self.assertTrue(model_onnx.graph.node is not None) |
Check notice
Code scanning / CodeQL
Imprecise assert Note test
| [("input", Int64TensorType([None, X.shape[1]]))], | ||
| target_opset=TARGET_OPSET, | ||
| ) | ||
| self.assertTrue(model_onnx is not None) |
Check notice
Code scanning / CodeQL
Imprecise assert Note test
| target_opset=TARGET_OPSET, | ||
| ) | ||
| self.assertTrue(model_onnx is not None) | ||
| self.assertTrue(model_onnx.graph.node is not None) |
Check notice
Code scanning / CodeQL
Imprecise assert Note test
| model_onnx = convert_sklearn( | ||
| model, "ordinal encoder two string cats", inputs, target_opset=TARGET_OPSET | ||
| ) | ||
| self.assertTrue(model_onnx is not None) |
Check notice
Code scanning / CodeQL
Imprecise assert Note test
Signed-off-by: boccaff <[email protected]>
Signed-off-by: boccaff <[email protected]>
Signed-off-by: boccaff <[email protected]>
boccaff
force-pushed
the
feat-target_encoder_support
branch
from
e945a85
to
e243865
Compare
|
Thanks for the comments @xadupre. I've removed the line, and solved a couple of the CodeQL suggestions (removed an unused import and an unused variable). The rest of the CodeQL suggested changes would diverge from the other implementations. For the |
Signed-off-by: boccaff <[email protected]>
Signed-off-by: boccaff <[email protected]>
This PR implements a converter and a shape calculator for the TargetEncoder class introduced in Scikit-learn 1.5. The code follows much of the implementation of the converter for Ordinal Encoder.
A partial suit of tests is already implemented, but there is at least a couple of additional tests that I would like to add (missing values and using the smooth parameter from sklearn, even though I think it shouldn't matter).