Github user dorx commented on a diff in the pull request:
https://github.com/apache/spark/pull/1367#discussion_r14836617
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/stat/correlation/Correlation.scala
---
@@ -0,0 +1,121 @@
+/*
+ * 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.stat.correlation
+
+import org.apache.spark.mllib.linalg.{DenseVector, Matrix, Vector}
+import org.apache.spark.rdd.RDD
+
+/**
+ * New correlation algorithms should implement this trait
+ */
+trait Correlation {
+
+ /**
+ * Compute correlation for two datasets.
+ */
+ def computeCorrelation(x: RDD[Double], y: RDD[Double]): Double
+
+ /**
+ * Compute the correlation matrix S, for the input matrix, where S(i, j)
is the correlation
+ * between column i and j.
+ */
+ def computeCorrelationMatrix(X: RDD[Vector]): Matrix
+
+ /**
+ * Combine the two input RDD[Double]s into an RDD[Vector] and compute
the correlation using the
+ * correlation implementation for RDD[Vector]
+ */
+ def computeCorrelationWithMatrixImpl(x: RDD[Double], y: RDD[Double]):
Double = {
+ val mat: RDD[Vector] = x.zip(y).mapPartitions({ iter =>
+ iter.map {case(xi, yi) => new DenseVector(Array(xi, yi))}
+ }, preservesPartitioning = true)
+ computeCorrelationMatrix(mat)(0, 1)
+ }
+
+}
+
+/**
+ * Delegates computation to the specific correlation object based on the
input method name
+ *
+ * Currently supported correlations: pearson, spearman.
+ * After new correlation algorithms are added, please update the
documentation here and in
+ * Statistics.scala for the correlation APIs.
+ *
+ * Cases are ignored when doing method matching. We also allow initials,
e.g. "P" for "pearson", as
+ * long as initials are unique in the supported set of correlation
algorithms. In addition, a
--- End diff --
So the fact R has been supporting this feature for many years (and still
haven't deprecated it) is good precedence that among the well known
correlations, there aren't really any collisions. Since we require that only
well known and widely adopted algorithms be added to mllib, the likelihood of
collision is even smaller. (If someone's adding something in their own version
of spark, presumably they have already looked closely at the docs and been
warned).
Because we decided to go with strings for method specification instead of
enums for cross language uniformity, the method is only checked (and possibly
invalidated) at run time. Therefore, we do want to provide mechanisms for fault
tolerance and minimization in method specification. To address a related
comment, "spearman" and "spearmans" are both really common in referring to
Spearman's correlation (imagine the heartbreak of having your program fail
because of an extra "s" after hours of data manipulation computation!)
---
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.
---
