##### Site index · List index
Message view
Top
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-1933) Add distance measure interface and basic implementation to machine learning library
Date Thu, 07 May 2015 09:00:10 GMT
```
]

ASF GitHub Bot commented on FLINK-1933:
---------------------------------------

Github user tillrohrmann commented on a diff in the pull request:

---
@@ -0,0 +1,45 @@
+/*
+ * 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
+ * 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
+ *
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * 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.
+ */
+
+
+
+/** This class implements a cosine distance metric. The class calculates the distance
between
+  * the given vectors by dividing the dot product of two vectors by the product of their
lengths.
+  * We convert the result of division to a usable distance. So, 1 - cos(angle) is actually
returned.
+  *
+  * @see http://en.wikipedia.org/wiki/Cosine_similarity
+  */
+class CosineDistanceMeasure extends DistanceMeasure {
+  override def distance(a: Vector, b: Vector): Double = {
+    checkValidArguments(a, b)
+
+    val dotProd: Double = a.dot(b)
+    val denominator: Double = a.magnitude * b.magnitude
+    if (dotProd == 0 && denominator == 0) {
--- End diff --

what if `a` and `b` are both zero? Are they then similar with respect to the cosine similarity?

> Add distance measure interface and basic implementation to machine learning library
> -----------------------------------------------------------------------------------
>
>          Issue Type: New Feature
>          Components: Machine Learning Library
>            Reporter: Chiwan Park
>            Assignee: Chiwan Park
>              Labels: ML
>
> Add distance measure interface to calculate distance between two vectors and some implementations
of the interface. In FLINK-1745, [~till.rohrmann] suggests a interface following:
> {code}
> trait DistanceMeasure {
>   def distance(a: Vector, b: Vector): Double
> }
> {code}
> I think that following list of implementation is sufficient to provide first to ML library
users.
> * Manhattan distance [1]
> * Cosine distance [2]
> * Euclidean distance (and Squared) [3]
> * Tanimoto distance [4]
> * Minkowski distance [5]
> * Chebyshev distance [6]
> [1]: http://en.wikipedia.org/wiki/Taxicab_geometry
> [2]: http://en.wikipedia.org/wiki/Cosine_similarity
> [3]: http://en.wikipedia.org/wiki/Euclidean_distance
> [4]: http://en.wikipedia.org/wiki/Jaccard_index#Tanimoto_coefficient_.28extended_Jaccard_coefficient.29
> [5]: http://en.wikipedia.org/wiki/Minkowski_distance
> [6]: http://en.wikipedia.org/wiki/Chebyshev_distance

--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

```
Mime
View raw message