assign intervals

Author: nikhil-zlai

online/src/main/scala/ai/chronon/online/stats/AssignIntervals.scala

@@ -22,21 +22,21 @@ object AssignIntervals {
     val arr = Array.fill(bLen - 1)(0.0)
 
     // discard before b(0)
-    while(i < pLen && ptiles(i) < breaks(0)) {
+    while (i < pLen && ptiles(i) < breaks(0)) {
       i += 1
     }
 
-    while(bIndex < bLen) {
+    while (bIndex < bLen) {
       var result = 0.0
       val b = breaks(bIndex)
 
-      while(i < pLen && ptiles(i) < b) {
+      while (i < pLen && ptiles(i) < b) {
 
         // fraction of the equally spaced interval between p(i-1) to p(i) to assign to this break
         val fraction =
-          if(i == 0) 0.0
-          else if (ptiles(i) == ptiles(i-1)) 1.0
-          else (ptiles(i) - prev) / (ptiles(i) - ptiles(i-1))
+          if (i == 0) 0.0
+          else if (ptiles(i) == ptiles(i - 1)) 1.0
+          else (ptiles(i) - prev) / (ptiles(i) - ptiles(i - 1))
 
         result += fraction
         prev = ptiles(i)
@@ -46,7 +46,7 @@ object AssignIntervals {
       // residual fraction to assign
       val fraction =
         if (i <= 0 || i >= pLen) 0.0
-        else (b - prev) / (ptiles(i) - ptiles(i-1))
+        else (b - prev) / (ptiles(i) - ptiles(i - 1))
 
       result += fraction
       prev = b

online/src/main/scala/ai/chronon/online/stats/DistanceMetrics.scala

@@ -1,11 +1,8 @@
 package ai.chronon.online.stats
-
-import ai.chronon.api.ColorPrinter.ColorString
 import ai.chronon.api.Window
 
 import scala.math._
 
-
 object DistanceMetrics {
 
   // TODO move this to unit test
@@ -22,19 +19,18 @@ object DistanceMetrics {
     val hd = hellingerDistance(A, B)
     println(f"The Hellinger Distance between distributions A and B is: $hd%.5f")
 
-
     // format: off
     // aligned vertically for easier reasoning
     val ptiles = Array(   1,      4,   6,6,6,    8, 9      )
     val breaks = Array(0, 1, 2, 3,  5, 6,     7, 8, 9,   10)
     // format: on
 
     //val interval = 0.25
-    val expected = Array(0.0, 1.0/3.0 , 1.0/3.0, (1.0)/(3.0) + (1.0)/(2.0), (1.0)/(2.0), 2.5, 0.5, 1, 0)
+    val expected = Array(0.0, 1.0 / 3.0, 1.0 / 3.0, (1.0) / (3.0) + (1.0) / (2.0), (1.0) / (2.0), 2.5, 0.5, 1, 0)
 
     val result = AssignIntervals.on(ptiles = ptiles.map(_.toDouble), breaks = breaks.map(_.toDouble))
 
-    expected.zip(result).foreach{case (e, r) => println(s"exp: $e res: $r")}
+    expected.zip(result).foreach { case (e, r) => println(s"exp: $e res: $r") }
 
   }
 
@@ -43,15 +39,13 @@ object DistanceMetrics {
 
   case class Comparison[T](previous: T, current: T, timeDelta: Window)
 
-
-  def functionBuilder[T](binningFunc: Comparison[T] => Distributions, distanceFunc: Distributions => Double): Comparison[T] => Double = {
-    c =>
-      val dists = binningFunc(c)
-      val distance = distanceFunc(dists)
-      distance
+  def functionBuilder[T](binningFunc: Comparison[T] => Distributions,
+                         distanceFunc: Distributions => Double): Comparison[T] => Double = { c =>
+    val dists = binningFunc(c)
+    val distance = distanceFunc(dists)
+    distance
   }
 
-
   def hellingerDistance(p: Array[Double], q: Array[Double]): Double = {
     val (pProbs, qProbs) = computePDFs(p, q)
 
@@ -110,14 +104,14 @@ object DistanceMetrics {
     val interval: Double = 1.toDouble / (n - 1.0)
 
     def mass(i: Int, eh: Int): Mass = {
-      def indexMass(i: Int): Double = interval / (percentiles(i) - percentiles(i-1))
+      def indexMass(i: Int): Double = interval / (percentiles(i) - percentiles(i - 1))
       val isPointMass = eh > 1
       val m = (i, eh) match {
-        case (0, _) => 0.0 // before range
-        case (x, 0) if x>=n => 0.0 // after range
-        case (_, e) if e > 1 => (e - 1) * interval // point mass
-        case (x, 1) if x==n => indexMass(n-1) // exactly at end of range
-        case (x, _) => indexMass(x) // somewhere in between
+        case (0, _)           => 0.0 // before range
+        case (x, 0) if x >= n => 0.0 // after range
+        case (_, e) if e > 1  => (e - 1) * interval // point mass
+        case (x, 1) if x == n => indexMass(n - 1) // exactly at end of range
+        case (x, _)           => indexMass(x) // somewhere in between
       }
       Mass(m, isPointMass)
     }
@@ -137,8 +131,8 @@ object DistanceMetrics {
     require(p.length == q.length, s"Inputs are of different length ${p.length}, ${q.length}")
     var i = 0
     var result = 0.0
-    while(i < p.length) {
-      val inc = if (p(i)> 0 && q(i) > 0) p(i) * math.log(p(i) / q(i)) else 0
+    while (i < p.length) {
+      val inc = if (p(i) > 0 && q(i) > 0) p(i) * math.log(p(i) / q(i)) else 0
       result += inc
       i += 1
     }