Init cluster centers with norm, which save computings

Li Jingqin · Li Jingqin · commit 12ce7310a46d · 2014-11-14T10:54:55.000+08:00
diff --git a/mllib/src/main/scala/org/apache/spark/mllib/clustering/KMeansModel.scala b/mllib/src/main/scala/org/apache/spark/mllib/clustering/KMeansModel.scala
@@ -29,6 +29,7 @@ class KMeansModel (val clusterCenters: Array[Vector]) extends Serializable {
 
   /** Total number of clusters. */
   def k: Int = clusterCenters.length
+  val ccWithNorm: Array[BreezeVectorWithNorm] = clusterCenters.map(new BreezeVectorWithNorm(_))
 
   /** Returns the cluster index that a given point belongs to. */
   def predict(point: Vector): Int = {
@@ -43,7 +44,7 @@ class KMeansModel (val clusterCenters: Array[Vector]) extends Serializable {
   }
 
   def predict2(point: Vector): (Int, Double) = {
-    KMeans.findClosest(clusterCentersWithNorm, new BreezeVectorWithNorm(point))
+    KMeans.findClosest(ccWithNorm, new BreezeVectorWithNorm(point))
   }
 
   /** Maps given points to their cluster indices. */

Original file line number	Diff line number	Diff line change
`@@ -29,6 +29,7 @@ class KMeansModel (val clusterCenters: Array[Vector]) extends Serializable {`
`29`	`29`
`30`	`30`	`/** Total number of clusters. */`
`31`	`31`	`def k: Int = clusterCenters.length`
	`32`	`+ val ccWithNorm: Array[BreezeVectorWithNorm] = clusterCenters.map(new BreezeVectorWithNorm(_))`
`32`	`33`
`33`	`34`	`/** Returns the cluster index that a given point belongs to. */`
`34`	`35`	`def predict(point: Vector): Int = {`
`@@ -43,7 +44,7 @@ class KMeansModel (val clusterCenters: Array[Vector]) extends Serializable {`
`43`	`44`	`}`
`44`	`45`
`45`	`46`	`def predict2(point: Vector): (Int, Double) = {`
`46`		`- KMeans.findClosest(clusterCentersWithNorm, new BreezeVectorWithNorm(point))`
	`47`	`+ KMeans.findClosest(ccWithNorm, new BreezeVectorWithNorm(point))`
`47`	`48`	`}`
`48`	`49`
`49`	`50`	`/** Maps given points to their cluster indices. */`