Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/12402#discussion_r60271513
--- Diff:
mllib/src/main/scala/org/apache/spark/ml/clustering/GaussianMixture.scala ---
@@ -105,6 +108,15 @@ class GaussianMixtureModel private[ml] (
def gaussians: Array[MultivariateGaussian] = parentModel.gaussians
@Since("2.0.0")
--- End diff --
I'd prefer to have a wrapper since, without it, it would be difficult for
users to do anything with the gaussians' means and covariance matrices.
However, I should have realized on the initial PR that we should define a new
MultivariateGaussian class within spark.ml so that we do not reference
spark.mllib publicly. I can send a quick PR to add this on the Scala side.
@yanboliang does this sound reasonable to you?
For serialization to the Python side, we could use DataFrames if it
simplifies the code. To do that, we could keep a private reference to the
SQLContext from training.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
