The discussion in #35 showed that the one-vs-rest approach to multi-class classification for logistic regression was not importing as expected. This method of multi-class is encoded as a modelChain Segmentation, rather than a single RegressionModel.
This PR adds explicit one-vs-rest support for logistic regression.
The discussion in #35 showed that the one-vs-rest approach to multi-class classification for logistic regression was not importing as expected. This method of multi-class is encoded as a modelChain
Segmentation, rather than a singleRegressionModel.This PR adds explicit one-vs-rest support for logistic regression.