Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/16222
@mengxr
I used the following R code and glmnet to check whether `regParam = 0.5`
fits a good model.
> iris2 <- iris[iris$Species %in% c("versicolor", "virginica"), ]
> iris.x = as.matrix(iris2[, 1:4])
> iris.y = as.factor(as.character(iris2[, 5]))
> cvfit = cv.glmnet(iris.x, iris.y, family = "binomial", type.measure =
"class")
> cvfit$lambda.min
[1] 0.000423808
> cvfit = cv.glmnet(iris.x, iris.y, family = "multinomial", type.measure =
"class")
> cvfit$lambda.min
[1] 0.01324703
If I understand correctly, `regParam = 0.5` doesn't fit a good model for
both binomial and multinomial cases, as the minimal lambda is < 0.1.
---
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]
