Skip to content

[SPARK-5990] [MLLIB] Model import/export for IsotonicRegression #5270

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
yanboliang wants to merge 5 commits into from
Closed

[SPARK-5990] [MLLIB] Model import/export for IsotonicRegression #5270

Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
2b2f5a1
Model import/export for IsotonicRegression
yanboliang Mar 30, 2015
429ff7d
store each interval as a record
yanboliang Apr 4, 2015
49600cc
address comments
yanboliang Apr 9, 2015
f80ec1b
address comments
yanboliang Apr 20, 2015
872028d
fix code style
yanboliang Apr 21, 2015
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
Prev Previous commit
Next Next commit
address comments
  • Loading branch information
@yanboliang
yanboliang committed Apr 9, 2015
commit 49600cc3be22383d787d40a73cae09ed7106c083
28 changes: 14 additions & 14 deletions mllib/src/main/scala/org/apache/spark/mllib/regression/IsotonicRegression.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -133,9 +133,7 @@ class IsotonicRegressionModel (
}

override def save(sc: SparkContext, path: String): Unit = {
val intervals = boundaries.toList.zip(predictions.toList).toArray
val data = IsotonicRegressionModel.SaveLoadV1_0.Data(intervals)
IsotonicRegressionModel.SaveLoadV1_0.save(sc, path, data, isotonic)
IsotonicRegressionModel.SaveLoadV1_0.save(sc, path, boundaries, predictions, isotonic)
}

override protected def formatVersion: String = "1.0"
Expand All @@ -153,9 +151,14 @@ object IsotonicRegressionModel extends Loader[IsotonicRegressionModel] {
def thisClassName: String = "org.apache.spark.mllib.regression.IsotonicRegressionModel"

/** Model data for model import/export */
case class Data(intervals: Array[(Double, Double)])

def save(sc: SparkContext, path: String, data: Data, isotonic: Boolean): Unit = {
case class Data(boundary: Double, prediction: Double)

def save(
sc: SparkContext,
path: String,
boundaries: Array[Double],
predictions: Array[Double],
isotonic: Boolean): Unit = {
val sqlContext = new SQLContext(sc)
import sqlContext.implicits._
Copy link
Contributor

@mengxr mengxr Apr 20, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Remove this line because no implicits are used.


Expand All @@ -164,21 +167,18 @@ object IsotonicRegressionModel extends Loader[IsotonicRegressionModel] {
("isotonic" -> isotonic)))
sc.parallelize(Seq(metadata), 1).saveAsTextFile(metadataPath(path))

val dataRDD: DataFrame = sc.parallelize(Seq(data), 1).toDF()
dataRDD.saveAsParquetFile(dataPath(path))
sqlContext.createDataFrame(boundaries.toList.zip(predictions.toList)
.map { case (b, p) => Data(b, p) }).saveAsParquetFile(dataPath(path))
Copy link
Contributor

@mengxr mengxr Apr 9, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It may read better with the following style:

sqlContext.createDataFrame(
  boundaries.toSeq.zip(predictions).map { case (b, p) => Data(b, p) }
).saveAsParquetFile(dataPath(path))

Copy link
Contributor

@mengxr mengxr Apr 20, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please see my previous comment about the code style here. The following has a better separation of the logic:

sqlContext.createDataFrame(
  boundaries.toSeq.zip(predictions).map { case (b, p) => Data(b, p) }
).saveAsParquetFile(dataPath(path))

}

def load(sc: SparkContext, path: String): (Array[Double], Array[Double]) = {
val sqlContext = new SQLContext(sc)
val dataRDD = sqlContext.parquetFile(dataPath(path))

checkSchema[Data](dataRDD.schema)
val dataArray = dataRDD.select("intervals").take(1)
assert(dataArray.size == 1,
s"Unable to load IsotonicRegressionModel data from: ${dataPath(path)}")
val data = dataArray(0)
val intervals = data.getAs[Seq[(Double, Double)]](0)
val (boundaries, predictions) = intervals.unzip
val dataArray = dataRDD.select("boundary", "prediction").collect()
val (boundaries, predictions) = dataArray.map {
x => (x.getAs[Double](0), x.getAs[Double](1)) }.toList.unzip
Copy link
Contributor

@mengxr mengxr Apr 9, 2015

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

getAs[Double](0) -> getDouble

The data might not be loaded into a single partition. So we need to sort (boundary, prediction) list first based on isotonic.

(boundaries.toArray, predictions.toArray)
}
}
Expand Down