Multiclass pipeline #21
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jeremyliweishih
commented
Aug 23, 2019
jeremyliweishih
commented
Aug 23, 2019
jeremyliweishih
commented
Aug 23, 2019
kmax12
suggested changes
Aug 24, 2019
kmax12
suggested changes
Aug 27, 2019
kmax12
suggested changes
Sep 9, 2019
kmax12
suggested changes
Sep 9, 2019
kmax12
suggested changes
Sep 10, 2019
kmax12
suggested changes
Sep 10, 2019
…evalml into multiclass_pipeline
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Added multiclass functionality for classification models, updated metrics and included tests.
Does not include changes for feature-importance
However, I don't think we would need to change the feature_importance functions for logistic regression, RF or XGClassifier as they're are not one. vs all classifiers. For LR, the loss function is changed to multinomial logistic loss for multiclass classification, RF inherently works with multiple classes, and XGClassifier uses softmax loss.Upon further review, there will need to be changes for feature importance.Feature importance has been added for LR. Changes will not be needed for RF or XGClassifier.
Metrics
Currently for classification metrics, I am using themicroparameter which uses the global count to define metrics. In the case of precision, the total count of FP etc is used. Themacroparameter would calculate FP etc. per label and then use the average. I think either parameter could work and its more of a design choice.Separate metrics have been created for each type of averaging.