[
https://issues.apache.org/jira/browse/FLINK-2030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14705053#comment-14705053
]
ASF GitHub Bot commented on FLINK-2030:
---------------------------------------
Github user tillrohrmann commented on a diff in the pull request:
https://github.com/apache/flink/pull/861#discussion_r37538357
--- Diff:
flink-staging/flink-ml/src/main/scala/org/apache/flink/ml/statistics/DiscreteHistogram.scala
---
@@ -0,0 +1,126 @@
+/*
+ * 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.flink.ml.statistics
+
+import scala.collection.mutable
+
+/** Implementation of a discrete valued online histogram
+ *
+ * =Parameters=
+ * -[[numCategories]]:
+ * Number of categories in the histogram
+ */
+class DiscreteHistogram(numCategories: Int) extends OnlineHistogram {
+
+ require(numCategories > 0, "Capacity must be greater than zero")
+ val data = new mutable.HashMap[Double, Int]()
+
+ /** Number of categories in the histogram
+ *
+ * @return number of categories
+ */
+ override def bins: Int = {
+ numCategories
+ }
+
+ /** Increment count of category c
+ *
+ * @param c category whose count needs to be incremented
+ */
+ override def add(c: Double): Unit = {
+ data.get(c) match {
+ case None =>
+ require(data.size < numCategories, "Insufficient capacity. Failed
to add.")
+ data.put(c, 1)
+ case Some(value) =>
+ data.update(c, value + 1)
+ }
+ }
+
+ /** Merges the histogram with h and returns a new histogram
+ *
+ * @param h histogram to be merged
+ * @param B number of categories in the resultant histogram.
+ * (Default: ```0```, number of categories will be the size of
union of categories in
+ * both histograms)
+ * @return Merged histogram with capacity B
+ */
+ override def merge(h: OnlineHistogram, B: Int = 0): DiscreteHistogram = {
+ h match {
+ case h1: DiscreteHistogram => {
+ val finalMap = new mutable.HashMap[Double, Int]()
+ data.iterator.foreach(x => finalMap.put(x._1, x._2))
+ h1.data.iterator.foreach(x => {
+ finalMap.get(x._1) match {
+ case None => finalMap.put(x._1, x._2)
+ case Some(value) => finalMap.update(x._1, x._2 + value)
+ }
+ })
+ require(B == 0 || finalMap.size <= B, "Insufficient capacity.
Failed to merge")
--- End diff --
Why can't we adaptively fuse bins together and store the values which are
in each bin, if we run out of capacity. Then we can later add values to their
corresponding bins. Seems to me the better solution than to throw an
`IllegalArgumentException`.
> Implement an online histogram with Merging and equalization features
> --------------------------------------------------------------------
>
> Key: FLINK-2030
> URL: https://issues.apache.org/jira/browse/FLINK-2030
> Project: Flink
> Issue Type: Sub-task
> Components: Machine Learning Library
> Reporter: Sachin Goel
> Assignee: Sachin Goel
> Priority: Minor
> Labels: ML
>
> For the implementation of the decision tree in
> https://issues.apache.org/jira/browse/FLINK-1727, we need to implement an
> histogram with online updates, merging and equalization features. A reference
> implementation is provided in [1]
> [1].http://www.jmlr.org/papers/volume11/ben-haim10a/ben-haim10a.pdf
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
