SPARK-4111 [MLlib] add regression metrics #2978
Conversation
|
Can one of the admins verify this patch? |
| * Computes R^2^, the coefficient of determination. | ||
| * @return | ||
| */ | ||
| def r2_socre(): Double = { |
|
Update the title with SPARK-XXXX [MLLIB] |
| private lazy val summarizer: MultivariateOnlineSummarizer = { | ||
| val summarizer: MultivariateOnlineSummarizer = valuesAndPreds.map{ | ||
| case (value,pred) => Vectors.dense( | ||
| Array(value, pred, value - pred, math.abs(value - pred), math.pow(value - pred, 2.0)) |
There was a problem hiding this comment.
I don't see that stats for pred are used?
There was a problem hiding this comment.
Yes, it's not used and I have remove it in a new commit.
yanboliang
changed the title
|
Rename re_score() and remove unused column. |
| * @return | ||
| */ | ||
| def r2_score(): Double = { | ||
| 1 - summarizer.mean(3) * summarizer.count / (summarizer.variance(0) * (summarizer.count - 1)) |
There was a problem hiding this comment.
I think this might be worth a comment to explain what sums of squares you are trying to compute in the numerator and denominator. A link to the definition might be good, here and for explained variance, since they are related.
|
This is picky now, but you might write out "meanAverageError" instead of saying "mae". Is "r2_score" style-wise correct vs "r2Score"? (Sorry should have thought of that.) Finally consider using |
| private lazy val summarizer: MultivariateOnlineSummarizer = { | ||
| val summarizer: MultivariateOnlineSummarizer = valuesAndPreds.map{ | ||
| case (value,pred) => Vectors.dense( | ||
| Array(value, value - pred, math.abs(value - pred), math.pow(value - pred, 2.0)) |
There was a problem hiding this comment.
Also picky but you can avoid math.pow and avoid computing value - pred 3 times here with a local var. Might be cleaner. This LGTM for what it's worth.
There was a problem hiding this comment.
The third and the forth columns are not necessary. You can use normL1 and normL2 on the second column:
|
Rename parameter and function names to be consistent with spark naming rules. |
|
ok to test |
|
test this please |
|
Test build #22457 has started for PR 2978 at commit
|
|
Test build #22457 timed out for PR 2978 at commit |
|
Test FAILed. |
| * @param predictionAndObservations an RDD of (prediction,observation) pairs. | ||
| */ | ||
| @Experimental | ||
| class RegressionMetrics(predictionAndObservations: RDD[(Double, Double)]) extends Logging { |
| */ | ||
| private lazy val summarizer: MultivariateOnlineSummarizer = { | ||
| val summarizer: MultivariateOnlineSummarizer = predictionAndObservations.map{ | ||
| case (prediction,observation) => Vectors.dense( |
|
Test build #22527 has started for PR 2978 at commit
|
|
Test build #22528 has started for PR 2978 at commit
|
|
Test build #22527 has finished for PR 2978 at commit
|
|
Test PASSed. |
|
Test build #22528 has finished for PR 2978 at commit
|
|
Test PASSed. |
|
LGTM. Merged into master. Thanks! |
Add RegressionMetrics.scala as regression metrics used for evaluation and corresponding test case RegressionMetricsSuite.scala.