Created
October 5, 2016 00:22
-
-
Save srnghn/24e35ce23f504bcb2e8d5c0c060ad6d7 to your computer and use it in GitHub Desktop.
Pearson's R Correlation for Spark 2.0. Created after getting inconsistant results with Statistics.corr. The two scale columns to be evaluated are to be selected from a DataFrame, converted to class type Dataset[ScaleTuple] (defined in this code) and passed to the correlation function.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
// Create a class, ScaleTuple, to pass to the Pearson's R function so that columns can be referred to by specific names. | |
final case class ScaleTuple(var1: Double, var2: Double) | |
// Column names to use when converting to ScaleTuple | |
val colnames = Seq("var1", "var2") | |
/** | |
* Implementation of Pearson's R function: calculates r, the measurement of linear dependence between two variables | |
* Utilizes DataSet's 'agg' function | |
**/ | |
def getPearsonsR(scaleData: org.apache.spark.sql.Dataset[ScaleTuple]) = { | |
scaleData.createOrReplaceTempView("scaleD") | |
val avg = spark.sql("select avg(var1) as Avg1, avg(var2) as Avg2 from scaleD") | |
avg.createOrReplaceTempView("avg") | |
val dif = spark.sql("select A.var1, A.var2, (A.var1 - B.Avg1) as Diff1, (A.var2 - B.Avg2) as Diff2 from scaleD A join avg B") | |
dif.createOrReplaceTempView("dif") | |
val mult = spark.sql("select (Diff1 * Diff2) as Multiple, (Diff1 * Diff1) as Diff1Sq, (Diff2 * Diff2) as Diff2Sq from dif") | |
val totMult = mult.agg(org.apache.spark.sql.functions.sum("Multiple")). | |
first.get(0) match {case d: Double => d case l: Long => l.toDouble} | |
val totSqRt1 = mult.agg(sqrt(org.apache.spark.sql.functions.sum("Diff1Sq"))). | |
first.get(0) match {case d: Double => d case l: Long => l.toDouble} | |
val totSqRt2 = mult.agg(sqrt(org.apache.spark.sql.functions.sum("Diff2Sq"))). | |
first.get(0) match {case d: Double => d case l: Long => l.toDouble} | |
val r = totMult / (totSqRt1 * totSqRt2) | |
r | |
} | |
// Example of how to convert to ScaleTuple and call the Pearson's R function | |
val scaleTuple = spark.sql("select height, weight from metrics").toDF(colnames: _*).as[ScaleTuple] | |
val r = getPearsonsR(scaleTuple) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment