Github user tengpeng commented on a diff in the pull request:
https://github.com/apache/spark/pull/19638#discussion_r148641672
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/evaluation/RegressionEvaluatorSuite.scala
---
@@ -73,6 +73,11 @@ class RegressionEvaluatorSuite
evaluator.setMetricName("r2")
assert(evaluator.evaluate(predictions) ~== 0.9998387 absTol 0.01)
+ // Adjusted r2
+ evaluator.setMetricName("r2adj")
+ assert(evaluator.evaluate(predictions) ~== 0.9998 absTol 0.01)
+ print(evaluator.evaluate(predictions))
--- End diff --
> Alternatively, you can just compute n from the size of one of the stats
vectors.
I tried this. In old `RegressionMetrics`, only `y` related information is
passed, observed `y_i` & fitted `\hat{y}`. The length of `x` is sent from
`MultivariateStatisticalSummary` but the width does not.
Maybe I am missing something here. Do you have any hints to get the width
of `x` using `states vectors`?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
