first DTW method realization

2023-04-22 09:52:14 +03:00 · 2023-04-22 09:52:14 +03:00 · b8809d1c21
commit b8809d1c21
parent 134f265700
1 changed files with 125 additions and 0 deletions
--- a/kmath-stat/src/commonMain/kotlin/space/kscience/kmath/series/DynamicTimeWarping.kt
+++ b/kmath-stat/src/commonMain/kotlin/space/kscience/kmath/series/DynamicTimeWarping.kt
@ -0,0 +1,125 @@
+/*
+ * Copyright 2018-2023 KMath contributors.
+ * Use of this source code is governed by the Apache 2.0 license that can be found in the license/LICENSE.txt file.
+ */
+
+package space.kscience.kmath.series
+
+import space.kscience.kmath.structures.DoubleBuffer
+import kotlin.math.abs
+
+public const val LEFT_OFFSET : Int = -1
+public const val BOTTOM_OFFSET : Int = 1
+public const val DIAGONAL_OFFSET : Int = 0
+
+// TODO: Change container for alignMatrix to kmath special ND structure
+public data class DynamicTimeWarpingData(val totalCost : Double = 0.0,
+                                     val alignMatrix : Array<BooleanArray> = Array(0) {BooleanArray(0)}) {
+    override fun equals(other: Any?): Boolean {
+        if (this === other) return true
+        if (other == null || this::class != other::class) return false
+
+        other as DynamicTimeWarpingData
+
+        if (totalCost != other.totalCost) return false
+        if (!alignMatrix.contentDeepEquals(other.alignMatrix)) return false
+
+        return true
+    }
+
+    override fun hashCode(): Int {
+        var result = totalCost.hashCode()
+        result = 31 * result + alignMatrix.contentDeepHashCode()
+        return result
+    }
+}
+
+/**
+ * costMatrix calculates special matrix of costs alignment for two series.
+ * Formula: costMatrix[i, j] = euqlidNorm(series1(i), series2(j)) + min(costMatrix[i - 1, j],
+ *                                                                      costMatrix[i, j - 1],
+ *                                                                      costMatrix[i - 1, j - 1]).
+ * There is special cases for i = 0 or j = 0.
+ */
+
+public fun costMatrix(series1 : DoubleBuffer, series2 : DoubleBuffer) : Array<DoubleArray> {
+    val dtwMatrix: Array<DoubleArray> = Array(series1.size){ row ->
+        DoubleArray(series2.size) { col ->
+            abs(series1[row] - series2[col])
+        }
+    }
+
+    for (i in dtwMatrix.indices) {
+        for (j in dtwMatrix[i].indices) {
+            dtwMatrix[i][j] += when {
+                i == 0 && j == 0 -> 0.0
+                i == 0 -> dtwMatrix[i][j-1]
+                j == 0 -> dtwMatrix[i-1][j]
+                else -> minOf(
+                    dtwMatrix[i][j-1],
+                    dtwMatrix[i-1][j],
+                    dtwMatrix[i-1][j-1]
+                )
+            }
+        }
+    }
+    return dtwMatrix
+}
+
+/**
+ * PathIndices class for better code perceptibility.
+ * Special fun moveOption represent offset for indices. Arguments of this function
+ * is flags for bottom, diagonal or left offsets respectively.
+ */
+
+public data class PathIndices (var id_x: Int, var id_y: Int) {
+    public fun moveOption (direction: Int) {
+        when(direction) {
+            BOTTOM_OFFSET -> id_x--
+            DIAGONAL_OFFSET -> {
+                id_x--
+                id_y--
+            }
+            LEFT_OFFSET -> id_y--
+            else -> throw Exception("There is no such offset flag!")
+        }
+    }
+}
+
+/**
+ * Final DTW method realization. Returns alignment matrix
+ * for two series comparing and penalty for this alignment.
+ */
+
+public fun dynamicTimeWarping(series1 : DoubleBuffer, series2 : DoubleBuffer) : DynamicTimeWarpingData {
+    var cost = 0.0
+    var pathLength = 0
+    val costMatrix = costMatrix(series1, series2)
+    val alignMatrix : Array<BooleanArray> = Array(costMatrix.size) { BooleanArray(costMatrix.first().size) }
+    val indexes = PathIndices(alignMatrix.lastIndex,  alignMatrix.last().lastIndex)
+
+    with(indexes) {
+        alignMatrix[id_x][id_y] = true
+        cost += costMatrix[id_x][id_y]
+        pathLength++
+        while (id_x != 0 || id_y != 0) {
+            when {
+                id_x == 0 || costMatrix[id_x][id_y] == costMatrix[id_x][id_y - 1] + abs(series1[id_x] - series2[id_y]) -> {
+                    moveOption(LEFT_OFFSET)
+                }
+                id_y == 0 || costMatrix[id_x][id_y] == costMatrix[id_x - 1][id_y] + abs(series1[id_x] - series2[id_y]) -> {
+                    moveOption(BOTTOM_OFFSET)
+                }
+                costMatrix[id_x][id_y] == costMatrix[id_x - 1][id_y - 1] + abs(series1[id_x] - series2[id_y]) -> {
+                    moveOption(DIAGONAL_OFFSET)
+                }
+            }
+            alignMatrix[id_x][id_y] = true
+            cost += costMatrix[id_x][id_y]
+            pathLength++
+        }
+        cost /= pathLength
+    }
+    return DynamicTimeWarpingData(cost, alignMatrix)
+}
+