diff options
| author | Xiangrui Meng <meng@databricks.com> | 2015-03-13 10:27:28 -0700 |
|---|---|---|
| committer | Xiangrui Meng <meng@databricks.com> | 2015-03-13 10:27:28 -0700 |
| commit | 7f13434a5c52b815c584ec773ab0e5df1a35ea86 (patch) | |
| tree | b715eaee2309bb66bb8eda4ce00f30ff52fa82f9 /docs/mllib-guide.md | |
| parent | dc4abd4dc40deacab39bfa9572b06bf0ea6daa6d (diff) | |
| download | spark-7f13434a5c52b815c584ec773ab0e5df1a35ea86.tar.gz spark-7f13434a5c52b815c584ec773ab0e5df1a35ea86.tar.bz2 spark-7f13434a5c52b815c584ec773ab0e5df1a35ea86.zip | |
[SPARK-6278][MLLIB] Mention the change of objective in linear regression
As discussed in the RC3 vote thread, we should mention the change of objective in linear regression in the migration guide. srowen
Author: Xiangrui Meng <meng@databricks.com>
Closes #4978 from mengxr/SPARK-6278 and squashes the following commits:
fb3bbe6 [Xiangrui Meng] mention regularization parameter
bfd6cff [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-6278
375fd09 [Xiangrui Meng] address Sean's comments
f87ae71 [Xiangrui Meng] mention step size change
Diffstat (limited to 'docs/mllib-guide.md')
| -rw-r--r-- | docs/mllib-guide.md | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/docs/mllib-guide.md b/docs/mllib-guide.md index 598374f66d..f8e879496c 100644 --- a/docs/mllib-guide.md +++ b/docs/mllib-guide.md @@ -102,6 +102,8 @@ In the `spark.mllib` package, there were several breaking changes. The first ch * In `DecisionTree`, the deprecated class method `train` has been removed. (The object/static `train` methods remain.) * In `Strategy`, the `checkpointDir` parameter has been removed. Checkpointing is still supported, but the checkpoint directory must be set before calling tree and tree ensemble training. * `PythonMLlibAPI` (the interface between Scala/Java and Python for MLlib) was a public API but is now private, declared `private[python]`. This was never meant for external use. +* In linear regression (including Lasso and ridge regression), the squared loss is now divided by 2. + So in order to produce the same result as in 1.2, the regularization parameter needs to be divided by 2 and the step size needs to be multiplied by 2. ## Previous Spark Versions |