Github user dbtsai commented on a diff in the pull request:
https://github.com/apache/spark/pull/4622#discussion_r29738695
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/clustering/AffinityPropagation.scala
---
@@ -0,0 +1,475 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.mllib.clustering
+
+import scala.collection.mutable
+
+import org.apache.spark.{Logging, SparkException}
+import org.apache.spark.annotation.Experimental
+import org.apache.spark.api.java.JavaRDD
+import org.apache.spark.graphx._
+import org.apache.spark.graphx.impl.GraphImpl
+import org.apache.spark.rdd.RDD
+
+/**
+ * :: Experimental ::
+ *
+ * Model produced by [[AffinityPropagation]].
+ *
+ * @param id cluster id.
+ * @param exemplar cluster exemplar.
+ * @param members cluster members.
+ */
+@Experimental
+case class AffinityPropagationCluster(val id: Long, val exemplar: Long,
val members: Array[Long])
+
+/**
+ * :: Experimental ::
+ *
+ * Model produced by [[AffinityPropagation]].
+ *
+ * @param clusters the clusters of AffinityPropagation clustering results.
+ */
+@Experimental
+class AffinityPropagationModel(
+ val clusters: RDD[AffinityPropagationCluster]) extends Serializable {
+
+ /**
+ * Set the number of clusters
+ */
+ lazy val getK: Long = clusters.count()
+
+ /**
+ * Find the cluster the given vertex belongs
+ * @param vertexID vertex id.
+ * @return a [[Array]] that contains vertex ids in the same cluster of
given vertexID. If
+ * the given vertex doesn't belong to any cluster, return null.
+ */
+ def findCluster(vertexID: Long): Array[Long] = {
+ val cluster = clusters.filter(_.members.contains(vertexID)).collect()
+ if (cluster.nonEmpty) {
+ cluster(0).members
+ } else {
+ null
+ }
+ }
+
+ /**
+ * Find the cluster id the given vertex belongs to
+ * @param vertexID vertex id.
+ * @return the cluster id that the given vertex belongs to. If the given
vertex doesn't belong to
+ * any cluster, return -1.
+ */
+ def findClusterID(vertexID: Long): Long = {
+ val clusterIds = clusters.flatMap { cluster =>
+ if (cluster.members.contains(vertexID)) {
+ Seq(cluster.id)
+ } else {
+ Seq()
+ }
+ }.collect()
+ if (clusterIds.nonEmpty) {
+ clusterIds(0)
+ } else {
+ -1
+ }
+ }
+}
+
+/**
+ * The message exchanged on the node graph
+ */
+private case class EdgeMessage(
+ similarity: Double,
+ availability: Double,
+ responsibility: Double) extends Equals {
+ override def canEqual(that: Any): Boolean = {
+ that match {
+ case e: EdgeMessage =>
+ similarity == e.similarity && availability == e.availability &&
+ responsibility == e.responsibility
+ case _ =>
+ false
+ }
+ }
+}
+
+/**
+ * The data stored in each vertex on the graph
+ */
+private case class VertexData(availability: Double, responsibility: Double)
+
+/**
+ * :: Experimental ::
+ *
+ * Affinity propagation (AP), a graph clustering algorithm based on the
concept of "message passing"
+ * between data points. Unlike clustering algorithms such as k-means or
k-medoids, AP does not
+ * require the number of clusters to be determined or estimated before
running it. AP is developed
+ * by [[http://doi.org/10.1126/science.1136800 Frey and Dueck]].
+ *
+ * @param maxIterations Maximum number of iterations of the AP algorithm.
+ * @param lambda lambda parameter used in the messaging iteration loop
+ * @param normalization Indication of performing normalization
+ * @param symmetric Indication of using symmetric similarity input
+ *
+ * @see [[http://en.wikipedia.org/wiki/Affinity_propagation Affinity
propagation (Wikipedia)]]
+ */
+@Experimental
+class AffinityPropagation private[clustering] (
+ private var maxIterations: Int,
+ private var lambda: Double,
+ private var normalization: Boolean,
+ private var symmetric: Boolean) extends Serializable {
+
+ import org.apache.spark.mllib.clustering.AffinityPropagation._
+
+ /** Constructs a AP instance with default parameters: {maxIterations:
100, lambda: `0.5`,
+ * normalization: false, symmetric: true}.
+ */
+ def this() = this(maxIterations = 100, lambda = 0.5, normalization =
false, symmetric = true)
+
+ /**
+ * Set maximum number of iterations of the messaging iteration loop
+ */
+ def setMaxIterations(maxIterations: Int): this.type = {
+ this.maxIterations = maxIterations
+ this
+ }
+
+ /**
+ * Get maximum number of iterations of the messaging iteration loop
+ */
+ def getMaxIterations(): Int = {
+ this.maxIterations
+ }
+
+ /**
+ * Set lambda of the messaging iteration loop
+ */
+ def setLambda(lambda: Double): this.type = {
+ this.lambda = lambda
+ this
+ }
+
+ /**
+ * Get lambda of the messaging iteration loop
+ */
+ def getLambda(): Double = {
+ this.lambda
+ }
+
+ /**
+ * Set whether to do normalization or not
+ */
+ def setNormalization(normalization: Boolean): this.type = {
+ this.normalization = normalization
+ this
+ }
+
+ /**
+ * Get whether to do normalization or not
+ */
+ def getNormalization(): Boolean = {
+ this.normalization
+ }
+
+ /**
+ * Set whether the input similarities are symmetric or not.
+ * When symmetric is set to true, we assume that input similarities only
contain triangular
+ * matrix. That means, only s,,ij,, is included in the similarities. If
both s,,ij,, and
+ * s,,ji,, are given in the similarities, it very possibly causes error.
+ */
+ def setSymmetric(symmetric: Boolean): this.type = {
+ this.symmetric = symmetric
+ this
+ }
+
+ /**
+ * Get whether the input similarities are symmetric or not
+ */
+ def getSymmetric(): Boolean = {
+ this.symmetric
+ }
+
+ /**
+ * Calculate the median value of similarities
+ */
+ private def getMedian(similarities: RDD[(Long, Long, Double)]): Double =
{
+ import org.apache.spark.SparkContext._
+
+ val sorted = similarities.sortBy(_._3).zipWithIndex().map {
+ case (v, idx) => (idx, v)
+ }
--- End diff --
Do you want to persistence the sorted RDD, and unpersistence later after
you find the median? I will add type of `sorted` to explicitly indicate it's
RDD. PS, `val count = sorted.count()` can be used to persistence the sorted
RDD.
---
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]
