Skip to content

[SPARK-14900][ML]:spark.ml classification metrics should include accuracy #12882

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wangmiao1981 wants to merge 7 commits into from
Closed

[SPARK-14900][ML]:spark.ml classification metrics should include accuracy #12882

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
fb52b97
add accuarcy to multiclassmetrics
wangmiao1981 May 3, 2016
41ab674
deprecate precision and use one big case for checking metrics
wangmiao1981 May 4, 2016
68c0a67
simply return true now.
wangmiao1981 May 4, 2016
9685c73
add unit test
wangmiao1981 May 5, 2016
ecccef7
address review comments
wangmiao1981 May 5, 2016
4a4481b
revert a comment
wangmiao1981 May 5, 2016
11132db
change deprecated precision to accuracy in many places
wangmiao1981 May 12, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 5 additions & 10 deletions mllib/src/main/scala/org/apache/spark/ml/evaluation/MulticlassClassificationEvaluator.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,15 +40,15 @@ class MulticlassClassificationEvaluator @Since("1.5.0") (@Since("1.5.0") overrid

/**
* param for metric name in evaluation (supports `"f1"` (default), `"precision"`, `"recall"`,
* `"weightedPrecision"`, `"weightedRecall"`)
* `"weightedPrecision"`, `"weightedRecall"`, `"accuracy"`)
* @group param
*/
@Since("1.5.0")
val metricName: Param[String] = {
val allowedParams = ParamValidators.inArray(Array("f1", "precision",
"recall", "weightedPrecision", "weightedRecall"))
"recall", "weightedPrecision", "weightedRecall", "accuracy"))
new Param(this, "metricName", "metric name in evaluation " +
"(f1|precision|recall|weightedPrecision|weightedRecall)", allowedParams)
"(f1|precision|recall|weightedPrecision|weightedRecall|accuracy)", allowedParams)
}

/** @group getParam */
Expand Down Expand Up @@ -86,18 +86,13 @@ class MulticlassClassificationEvaluator @Since("1.5.0") (@Since("1.5.0") overrid
case "recall" => metrics.recall
case "weightedPrecision" => metrics.weightedPrecision
case "weightedRecall" => metrics.weightedRecall
case "accuracy" => metrics.accuracy
}
metric
}

@Since("1.5.0")
override def isLargerBetter: Boolean = $(metricName) match {
case "f1" => true
case "precision" => true
case "recall" => true
case "weightedPrecision" => true
case "weightedRecall" => true
}
override def isLargerBetter: Boolean = true

@Since("1.5.0")
override def copy(extra: ParamMap): MulticlassClassificationEvaluator = defaultCopy(extra)
Expand Down
17 changes: 14 additions & 3 deletions mllib/src/main/scala/org/apache/spark/mllib/evaluation/MulticlassMetrics.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -139,7 +139,8 @@ class MulticlassMetrics @Since("1.1.0") (predictionAndLabels: RDD[(Double, Doubl
* Returns precision
*/
@Since("1.1.0")
lazy val precision: Double = tpByClass.values.sum.toDouble / labelCount
@deprecated("Use accuracy.", "2.0.0")
lazy val precision: Double = accuracy

/**
* Returns recall
Expand All @@ -148,14 +149,24 @@ class MulticlassMetrics @Since("1.1.0") (predictionAndLabels: RDD[(Double, Doubl
* of all false negatives)
*/
@Since("1.1.0")
lazy val recall: Double = precision
@deprecated("Use accuracy.", "2.0.0")
lazy val recall: Double = accuracy

/**
* Returns f-measure
* (equals to precision and recall because precision equals recall)
*/
@Since("1.1.0")
lazy val fMeasure: Double = precision
@deprecated("Use accuracy.", "2.0.0")
lazy val fMeasure: Double = accuracy

/**
* Returns accuracy
* (equals to the total number of correctly classified instances
* out of the total number of instances.)
*/
@Since("2.0.0")
lazy val accuracy: Double = tpByClass.values.sum.toDouble / labelCount

/**
* Returns weighted true positive rate
Expand Down
9 changes: 5 additions & 4 deletions mllib/src/test/scala/org/apache/spark/mllib/evaluation/MulticlassMetricsSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,11 +69,12 @@ class MulticlassMetricsSuite extends SparkFunSuite with MLlibTestSparkContext {
assert(math.abs(metrics.fMeasure(1.0, 2.0) - f2measure1) < delta)
assert(math.abs(metrics.fMeasure(2.0, 2.0) - f2measure2) < delta)

assert(math.abs(metrics.recall -
assert(math.abs(metrics.accuracy -
(2.0 + 3.0 + 1.0) / ((2 + 3 + 1) + (1 + 1 + 1))) < delta)
assert(math.abs(metrics.recall - metrics.precision) < delta)
assert(math.abs(metrics.recall - metrics.fMeasure) < delta)
assert(math.abs(metrics.recall - metrics.weightedRecall) < delta)
assert(math.abs(metrics.accuracy - metrics.precision) < delta)
assert(math.abs(metrics.accuracy - metrics.recall) < delta)
assert(math.abs(metrics.accuracy - metrics.fMeasure) < delta)
assert(math.abs(metrics.accuracy - metrics.weightedRecall) < delta)
assert(math.abs(metrics.weightedFalsePositiveRate -
((4.0 / 9) * fpRate0 + (4.0 / 9) * fpRate1 + (1.0 / 9) * fpRate2)) < delta)
assert(math.abs(metrics.weightedPrecision -
Expand Down