Github user jkbradley commented on a diff in the pull request:
https://github.com/apache/spark/pull/11933#discussion_r57956411
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/clustering/BisectingKMeansModel.scala
---
@@ -92,4 +100,83 @@ class BisectingKMeansModel private[clustering] (
*/
@Since("1.6.0")
def computeCost(data: JavaRDD[Vector]): Double =
this.computeCost(data.rdd)
+
+ @Since("2.0.0")
+ override def save(sc: SparkContext, path: String): Unit = {
+ BisectingKMeansModel.SaveLoadV1_0.save(sc, this, path)
+ }
+
+ override protected def formatVersion: String = "1.0"
+}
+
+object BisectingKMeansModel extends Loader[BisectingKMeansModel] {
+
+ @Since("2.0.0")
+ override def load(sc: SparkContext, path: String): BisectingKMeansModel
= {
+ BisectingKMeansModel.SaveLoadV1_0.load(sc, path)
+ }
+
+ private case class Data(index: Int, size: Long, center: Vector, norm:
Double, cost: Double,
+ height: Double, children: Seq[Int])
+
+ private object Data {
+ def apply(r: Row): Data = Data(r.getInt(0), r.getLong(1),
r.getAs[Vector](2), r.getDouble(3),
+ r.getDouble(4), r.getDouble(5), r.getSeq[Int](6))
+ }
+
+ private[clustering] object SaveLoadV1_0 {
+ private val thisFormatVersion = "1.0"
+
+ private[clustering]
+ val thisClassName =
"org.apache.spark.mllib.clustering.BisectingKMeansModel"
+
+ def save(sc: SparkContext, model: BisectingKMeansModel, path: String):
Unit = {
+ val sqlContext = SQLContext.getOrCreate(sc)
+ import sqlContext.implicits._
+ val metadata = compact(render(
+ ("class" -> thisClassName) ~ ("version" -> thisFormatVersion)
+ ~ ("rootId" -> model.root.index)))
+ sc.parallelize(Seq(metadata),
1).saveAsTextFile(Loader.metadataPath(path))
+
+ val data = getNodes(model.root).map(node => Data(node.index,
node.size,
+ node.centerWithNorm.vector, node.centerWithNorm.norm, node.cost,
node.height,
+ node.children.map(_.index)))
+ val dataRDD = sc.parallelize(data).toDF()
+ dataRDD.write.parquet(Loader.dataPath(path))
+ }
+
+ private def getNodes(node: ClusteringTreeNode):
Array[ClusteringTreeNode] = {
+ if (node.children.isEmpty) {
+ Array(node)
+ } else {
+ node.children.flatMap(getNodes(_)) ++ Array(node)
+ }
+ }
+
+ def load(sc: SparkContext, path: String): BisectingKMeansModel = {
+ implicit val formats = DefaultFormats
+ val sqlContext = SQLContext.getOrCreate(sc)
+ val (className, formatVersion, metadata) = Loader.loadMetadata(sc,
path)
+ require(className == thisClassName)
+ require(formatVersion == thisFormatVersion)
+ val rootId = (metadata \ "rootId").extract[Int]
+ val rows = sqlContext.read.parquet(Loader.dataPath(path))
+ Loader.checkSchema[Data](rows.schema)
+ val data = rows.rdd.map(Data.apply).collect().map(d => (d.index,
d)).toMap
+ val rootNode = buildTree(rootId, data)
+ new BisectingKMeansModel(rootNode)
+ }
+
+ private def buildTree(rootId: Int, data: Map[Int, Data]):
ClusteringTreeNode = {
+ val root = data.get(rootId).get
--- End diff --
```data(rootId)```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
