kscience/kmath
9
0
Fork 1
Code Issues Pull Requests 6 Packages Projects Releases 16 Wiki Activity

Variance test post-merge cleanup

Browse Source
This commit is contained in:
Alexander Nozik 2023-05-09 19:22:39 +03:00
parent 4dbcaca87c
commit 8eb25596a0
6 changed files with 40 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
.space.kts
View File

@ -1,3 +1,3 @@
job("Build") {
gradlew("openjdk:11", "build")
job("Build and run tests") {
gradlew("amazoncorretto:17-alpine", "build")
}

2
build.gradle.kts
View File

@ -15,7 +15,7 @@ allprojects {
}
group = "space.kscience"
version = "0.3.1-dev-RC"
version = "0.3.1"
}
subprojects {

2
kmath-ejml/src/main/kotlin/space/kscience/kmath/ejml/_generated.kt
View File

@ -19,9 +19,9 @@ import org.ejml.sparse.csc.factory.DecompositionFactory_DSCC
import org.ejml.sparse.csc.factory.DecompositionFactory_FSCC
import org.ejml.sparse.csc.factory.LinearSolverFactory_DSCC
import org.ejml.sparse.csc.factory.LinearSolverFactory_FSCC
import space.kscience.kmath.UnstableKMathAPI
import space.kscience.kmath.linear.*
import space.kscience.kmath.linear.Matrix
import space.kscience.kmath.UnstableKMathAPI
import space.kscience.kmath.nd.StructureFeature
import space.kscience.kmath.operations.DoubleField
import space.kscience.kmath.operations.FloatField

18
kmath-stat/src/commonMain/kotlin/space/kscience/kmath/distributions/NormalDistribution.kt
View File

@ -6,6 +6,7 @@
package space.kscience.kmath.distributions
import space.kscience.kmath.chains.Chain
import space.kscience.kmath.operations.DoubleField.pow
import space.kscience.kmath.random.RandomGenerator
import space.kscience.kmath.samplers.GaussianSampler
import space.kscience.kmath.samplers.InternalErf
@ -34,8 +35,23 @@ public class NormalDistribution(public val sampler: GaussianSampler) : Distribut
}
}
private companion object {
public companion object {
private val SQRT2 = sqrt(2.0)
/**
* Zelen & Severo approximation for the standard normal CDF.
* The error upper boundary by 7.5 * 10e-8.
*/
public fun zSNormalCDF(x: Double): Double {
val t = 1 / (1 + 0.2316419 * abs(x))
val sum = 0.319381530 * t -
0.356563782 * t.pow(2) +
1.781477937 * t.pow(3) -
1.821255978 * t.pow(4) +
1.330274429 * t.pow(5)
val temp = sum * exp(-abs(x).pow(2) / 2) / (2 * PI).pow(0.5)
return if (x >= 0) 1 - temp else temp
}
}
}

24
kmath-stat/src/commonMain/kotlin/space/kscience/kmath/distributions/zSNormalCdf.kt
View File

@ -1,24 +0,0 @@
/*
* 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.distributions
import space.kscience.kmath.operations.DoubleField.pow
import kotlin.math.PI
import kotlin.math.absoluteValue
import kotlin.math.exp
/**
* Zelen & Severo approximation for the standard normal CDF.
* The error is bounded by 7.5 * 10e-8.
* */
public fun zSNormalCDF(x: Double): Double {
val t = 1 / (1 + 0.2316419 * x.absoluteValue)
val summ = 0.319381530*t - 0.356563782*t.pow(2) + 1.781477937*t.pow(3) - 1.821255978*t.pow(4) + 1.330274429*t.pow(5)
val temp = summ * exp(-x.absoluteValue.pow(2) / 2) / (2 * PI).pow(0.5)
return if (x >= 0) 1 - temp else temp
}

28
kmath-stat/src/commonMain/kotlin/space/kscience/kmath/series/VarianceRatioTest.kt
View File

@ -5,7 +5,7 @@
package space.kscience.kmath.series
import space.kscience.kmath.distributions.zSNormalCDF
import space.kscience.kmath.distributions.NormalDistribution
import space.kscience.kmath.operations.DoubleField.pow
import space.kscience.kmath.operations.fold
import kotlin.math.absoluteValue
@ -14,8 +14,12 @@ import kotlin.math.absoluteValue
/**
* Container class for Variance Ratio Test result:
* ratio itself, corresponding Z-score, also it's p-value
* **/
public data class VarianceRatioTestResult(val varianceRatio: Double=1.0, val zScore: Double=0.0, val pValue: Double=0.5)
*/
public data class VarianceRatioTestResult(
val varianceRatio: Double = 1.0,
val zScore: Double = 0.0,
val pValue: Double = 0.5,
)
/**
@ -23,11 +27,17 @@ public data class VarianceRatioTestResult(val varianceRatio: Double=1.0, val zSc
* under Homoscedastic or Heteroscedstic assumptions
* with two-sided p-value test
* https://ssrn.com/abstract=346975
* **/
public fun SeriesAlgebra<Double, *, *, *>.varianceRatioTest(series: Series<Double>, shift: Int, homoscedastic: Boolean=true): VarianceRatioTestResult {
*
* @author https://github.com/mrFendel
*/
public fun SeriesAlgebra<Double, *, *, *>.varianceRatioTest(
series: Series<Double>,
shift: Int,
homoscedastic: Boolean = true,
): VarianceRatioTestResult {
require(shift > 1) {"Shift must be greater than one"}
require(shift < series.size) {"Shift must be smaller than sample size"}
require(shift > 1) { "Shift must be greater than one" }
require(shift < series.size) { "Shift must be smaller than sample size" }
val sum = { x: Double, y: Double -> x + y }
@ -46,7 +56,7 @@ public fun SeriesAlgebra<Double, *, *, *>.varianceRatioTest(series: Series<Doubl
val varianceAgg = demeanedSquaresAgg.fold(0.0, sum)
val varianceRatio =
varianceAgg * (series.size.toDouble() - 1) / variance / (series.size.toDouble() - shift.toDouble() + 1) / (1 - shift.toDouble()/series.size.toDouble()) / shift.toDouble()
varianceAgg * (series.size.toDouble() - 1) / variance / (series.size.toDouble() - shift.toDouble() + 1) / (1 - shift.toDouble() / series.size.toDouble()) / shift.toDouble()
// calculating asymptotic variance
@ -63,7 +73,7 @@ public fun SeriesAlgebra<Double, *, *, *>.varianceRatioTest(series: Series<Doubl
}
val zScore = (varianceRatio - 1) / phi.pow(0.5)
val pValue = 2*(1 - zSNormalCDF(zScore.absoluteValue))
val pValue = 2 * (1 - NormalDistribution.zSNormalCDF(zScore.absoluteValue))
return VarianceRatioTestResult(varianceRatio, zScore, pValue)
}