Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
@felixcheung I opened #15365 based on our discussion in this PR. I close
this PR now and please review #15365 . Thanks!
---
If your project is set up for it, you can reply to this email and
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
Thanks for your reply! Now, I am working in a fresh branch. Will submit
another PR and close this old one later.
---
If your project is set up for it, you can reply to this email and have
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
make sense, I think then we should have a separate function
---
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
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
For example:
`LogisticRegression` has `Threshold`, `Thresholds` and `AggregationDepth`
etc as parameters.
`GeneralizedLinearRegression` doesn't have such parameters. So the
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
Could you elaborate what's incompatible?
---
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
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
@felixcheung If I added it to `spark.glm`, it will break the current
`spark.glm` interface.
`setMethod("spark.glm", signature(data = "SparkDataFrame", formula =
"formula"),
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
Since `spark.glm` is different from the SparkR `glm` function, adding it to
`spark.glm` should be fine. I will follow this path. Thanks!
---
If your project is set up for it, you can reply
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
If we are closely matching the capability of glmnet then we could name it
`glmnet` or `spark.glmnet`, if this fits in the existing `spark.glm`
implementation (which is different from the SparkR
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/65833/
Test FAILed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Merged build finished. Test FAILed.
---
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
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #65833 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/65833/consoleFull)**
for PR 14818 at commit
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #65833 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/65833/consoleFull)**
for PR 14818 at commit
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
@felixcheung For Spark, it supports both `binomial` and `multinomial`
logistic regression. The link function `logit` under `glm` only supports
`binomial`. I found that `glmnet` supports both.
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
`glm` has a `link=logit` parameter? not sure if it maps to this
http://www.statmethods.net/advstats/glm.html
---
If your project is set up for it, you can reply to this email and have your
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
@felixcheung https://web.stanford.edu/~hastie/glmnet/glmnet_alpha.html
`glmnet` has more functionalities than Logistic Regression (BLOR and MLOR).
So `logit` could be a better name. Based
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
Yes. I am reviewing the code in #14834 to check if we just have to modify
existing wrapper to be consistent with Scala side API.
---
If your project is set up for it, you can reply to this
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
thanks - let's review if the name `glmnet` should be used or should this be
part of our support for logit
---
If your project is set up for it, you can reply to this email and have your
reply
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
Thanks! I will update the PR accordingly.
---
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
Github user sethah commented on the issue:
https://github.com/apache/spark/pull/14818
https://github.com/apache/spark/pull/14834 is merged now :)
---
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
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64489/
Test FAILed.
---
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Merged build finished. Test FAILed.
---
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
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #64489 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64489/consoleFull)**
for PR 14818 at commit
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #64489 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64489/consoleFull)**
for PR 14818 at commit
Github user wangmiao1981 commented on the issue:
https://github.com/apache/spark/pull/14818
OK. I will wait for the interface refactoring to continue this JIRA. Now, I
just leave it as a placeholder. Later on, I will change the name of the JIRA.
Thanks!
---
If your project is set
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
Right - I suspect we do need to change to R interface though which is why I
mention it. Thanks!
---
If your project is set up for it, you can reply to this email and have your
reply appear on
Github user sethah commented on the issue:
https://github.com/apache/spark/pull/14818
In fact, we are actually just eliminating the
`MultinomialLogisticRegression` interface and merging into the existing
`LogisticRegression` estimator. So, maybe we won't need a change after all? I'm
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/14818
cool, thanks for the heads up @sethah - please loop us in for the R side
changes.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as
Github user sethah commented on the issue:
https://github.com/apache/spark/pull/14818
This is going to have to wait. We are changing the interface completely.
See https://issues.apache.org/jira/browse/SPARK-17163.
---
If your project is set up for it, you can reply to this email
Github user junyangq commented on the issue:
https://github.com/apache/spark/pull/14818
We may want to use a different name? glmnet related name could be confusing
if it is actually only multiclass logistic.
---
If your project is set up for it, you can reply to this email and have
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #64441 has
finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64441/consoleFull)**
for PR 14818 at commit
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/64441/
Test FAILed.
---
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/14818
**[Test build #64441 has
started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/64441/consoleFull)**
for PR 14818 at commit
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/14818
Merged build finished. Test FAILed.
---
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
33 matches
Mail list logo
