num4anova 0.0.13-java → 0.0.15-java

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 84f7f49327e10a3da88a363daa4d6d5dec1dd4da00827d6f8e5c60ee6db4bc08
-  data.tar.gz: 495495a96d218034014971770b654431bb0df61aec47f8f3171fe351285022c4
+  metadata.gz: 81c39fe39ac94774f6ce94343e91c968190c63c510c9742d0a4b1e33c6f1a399
+  data.tar.gz: 9ba93973bdf9ad7237dd02702b67cd493335659a9b9f9162976ecb703ae351db
 SHA512:
-  metadata.gz: 1b46cc7670940eb88343959c80bad55209a981388acdd823c3d9eeafa72a629d2319cfb49707f3f998d72f0d080480785129c948c8ebedb388693476be59add1
-  data.tar.gz: 83585b2c13c1f964b8bf92e507a005deb4d358a612e23b939cb9d9911175006a1378e16719c1b73073834aa510f245b4432f0aba9f5dfcdceb78ee2cf5638ba3
+  metadata.gz: c2d3ffd3cbf0bcb0c4955c64c587b48da9d878bb61ce5c3a2b547de4c6563bafbf6852aa032e378d318419e55d6583f4af6467a7456a598501d6acea5f2b9c91
+  data.tar.gz: 7d97f9780f7d68f601810ac5ba5bcda923050b80e11fbff53b78017ceb61d954a788bb62f9a1dc85bc79c7e4fd0be857c538d6647d7506f91c6100295891238d

data/CHANGELOG.md

@@ -2,6 +2,16 @@
 
 ## Unreleased
 
+## [0.0.15] - 2024-04-13
+
+### add
+- add fuction of bonferrono_test in kruskalwallis_test.
+
+## [0.0.14] - 2024-04-08
+
+### add
+- add fuction of bonferrono_test in NonParametrixTestLib.
+
 ## [0.0.13] - 2024-04-04
 
 ### chg

data/ext/num4anova/MultiComp.java

@@ -3,9 +3,17 @@ import java.util.Arrays;
 import java.util.List;
 import java.util.ArrayList;
 import org.apache.commons.math3.distribution.TDistribution;
+import org.apache.commons.math3.stat.inference.MannWhitneyUTest;
 
 import org.apache.commons.math3.util.Combinations;
 public class MultiComp {
+    /*********************************/
+    /* interface define              */
+    /*********************************/
+    private interface HypothesisTest {
+        double[][] calcTestStatistic(double[][] xi);
+        boolean[][] executeTest(double[][] statistic, double a);
+    }
     public static class ParametrixTest {
         private static ParametrixTest paramTest = new ParametrixTest();
         public static ParametrixTest getInstance() {
@@ -46,13 +54,6 @@ public class MultiComp {
             return hypoth.executeTest(statistic, a);
         }
         /*********************************/
-        /* interface define              */
-        /*********************************/
-        private interface HypothesisTest {
-            double[][] calcTestStatistic(double[][] xi);
-            boolean[][] executeTest(double[][] statistic, double a);
-        }
-        /*********************************/
         /* Class define                  */
         /*********************************/
         // turkey法
@@ -366,5 +367,55 @@ public class MultiComp {
             }
         }
     }
+    public static class NonParametrixTest {
+        private static NonParametrixTest nonParamTest = new NonParametrixTest();
+        public static NonParametrixTest getInstance() {
+            return nonParamTest;
+        }
+        public boolean[][] bonferronoTest(double[][] xi, double a) {
+            HypothesisTest hypoth = new BonferroniTest();
+
+            double[][] statistic = hypoth.calcTestStatistic(xi);
+            return hypoth.executeTest(statistic, a * 0.5);
+        }
+        /*********************************/
+        /* Class define                  */
+        /*********************************/
+        // ボンフェロー法
+        private class BonferroniTest implements HypothesisTest {
+            private int n = 0;
+            private int k = 0;
+            public double[][] calcTestStatistic(double[][] xi) {
+                n = xi.length;
+                double[][] statistic = new double[n][n];
+                //
+                Combinations c = new Combinations(n, 2);
+                List<int[]> al = new ArrayList<>();
+                for(int[] iterate : c) {
+                    al.add(iterate);
+                }
+                k = al.size();
+                //
+                MannWhitneyUTest utest = new MannWhitneyUTest();
+                for(int[] array : al) {
+                    int i = array[0];
+                    int j = array[1];
+
+                    statistic[i][j] = utest.mannWhitneyUTest(xi[i], xi[j]);
+                }
+                return statistic;
+            }
+            public boolean[][] executeTest(double[][] statistic, double a) {
+                boolean[][] ret = new boolean[n][n];
+
+                for(int i = 0; i < n; i++) {
+                    for(int j = i+1; j < n; j++) {
+                        ret[i][j] = (statistic[i][j] < a * k) ? true : false;
+                    }
+                }
+                return ret;
+            }
+        }        
+    }
 }
 

data/ext/num4anova/OneWayLayout.java

@@ -30,6 +30,11 @@ import org.apache.commons.math3.stat.descriptive.DescriptiveStatistics;
 import org.apache.commons.math3.distribution.ChiSquaredDistribution;
 import org.apache.commons.math3.distribution.FDistribution;
 import java.util.Map;
+
+import org.apache.commons.math3.stat.ranking.NaNStrategy;
+import org.apache.commons.math3.stat.ranking.NaturalRanking;
+import org.apache.commons.math3.stat.ranking.TiesStrategy;
+import java.util.stream.IntStream;
 public class OneWayLayout {
     private static OneWayLayout oneWay = new OneWayLayout();
     public static OneWayLayout getInstance() {
@@ -72,6 +77,12 @@ public class OneWayLayout {
         double statistic = oneway.calcTestStatistic(xi);
         return oneway.execute_test(statistic, a);
     }
+    public boolean kruskalWallisTest(double[][] xi, double a) {
+        OneWayAnovaTest oneway = new KruskalWallisTest();
+
+        double statistic = oneway.calcTestStatistic(xi);
+        return oneway.execute_test(statistic, a);
+    }
     /*********************************/
     /* interface define              */
     /*********************************/
@@ -380,5 +391,66 @@ public class OneWayLayout {
             return (statistic >= f) ? true : false;
         }
     }
-}
+    // クラスカル・ウォリス検定
+    private class KruskalWallisTest implements OneWayAnovaTest {
+        private NaturalRanking naturalRanking;
+        private int[] ni = null;
+        public KruskalWallisTest() {
+            naturalRanking = new NaturalRanking(NaNStrategy.FIXED, 
+                                                 TiesStrategy.AVERAGE);
+        }
+        public double calcTestStatistic(double[][] xi) {
+            double[] z = concatSample(xi);            // 全てのデータをつなぐ
+            double[] ranks = naturalRanking.rank(z);  // rankに順位値に変換
+            double[] sumRankXi = calcSumRankXi(ranks);
+            double kw = 0.0;
+            int n = z.length;
+
+            for(int i = 0; i < sumRankXi.length; i++) {
+                kw += sumRankXi[i] * sumRankXi[i] / ni[i];
+            }
+            return 12.0 / (n * (n + 1.0)) * kw - 3.0 * (n + 1.0);
+        }
+        public boolean execute_test(double statistic, double a) {
+            ChiSquaredDistribution chi2Dist = new ChiSquaredDistribution(ni.length - 1);
+            double r_val = chi2Dist.inverseCumulativeProbability(1.0 - a);
+
+            return (r_val < statistic) ? true : false;
+        }
+        private int[] calcNi(double[][] xi) {
+            int[] ni = new int[xi.length];
+
+            for(int i = 0; i < ni.length; i++) {
+                ni[i] = xi[i].length;
+            }
+            return ni;
+        }
+        private double[] concatSample(double[][] xi) {
+            DescriptiveStatistics stat = new DescriptiveStatistics();
+            ni = calcNi(xi);
+            Arrays.stream(ni).forEach(stat::addValue);
+            int n = Double.valueOf(stat.getSum()).intValue();
+
+            double[] z = new double[n];
+            int idx = 0;
+            for(int cnt = 0; cnt < xi.length; cnt++) {
+                System.arraycopy(xi[cnt], 0, z, idx, ni[cnt]);
+                idx += ni[cnt];
+            }
+            return z;
+        }
+        private double[] calcSumRankXi(double[] ranks) {
+            double[] sumRi = new double[ni.length];
+            int idx = 0;
+
+            for(int cnt = 0; cnt < ni.length; cnt++) {
+                sumRi[cnt] = IntStream.range(idx, idx + ni[cnt])
+                                    .mapToDouble(i -> ranks[i])
+                                    .sum();
+                idx += ni[cnt];
+            }
+            return sumRi;
+        }
+    }
+ }
 

data/lib/multicomp.rb

@@ -43,7 +43,7 @@ module MultiCompLib
             ret = @paramTest.turkeyTest(xi.to_java(Java::double[]), a)
             return ret.to_a
         end
-        # ボンフェロー二の不等式による多重比較
+        # ボンフェロー二の不等式による多重比較(T検定)
         #
         # @overload bonferrono_test(xi, a)
         #   @param [array]  xi データ(double[][])
@@ -74,5 +74,37 @@ module MultiCompLib
     end
     # ノンパラメトリック検定
     class NonParametrixTestLib
+        def initialize
+            @nonParamTest = MultiComp::NonParametrixTest.getInstance()
+        end
+        # ボンフェロー二の不等式による多重比較(マン・ホイットニーU検定)
+        #
+        # @overload bonferrono_test(xi, a)
+        #   @param [array]  xi データ(double[][])
+        #   @param [double] a         有意水準
+        #   @return [Array] 検定結果(boolean[][] true:棄却域内 false:棄却域外)
+        # @example
+        #    xi = [
+        #        [12.2, 18.8, 18.2],
+        #        [22.2, 20.5, 14.6],
+        #        [20.8, 19.5, 26.3],
+        #        [26.4, 32.5, 31.3],
+        #        [24.5, 21.2, 22.4],
+        #    ]
+        #    nonParaTest = MultiCompLib::NonParametrixTestLib.new
+        #    nonParaTest.bonferrono_test(xi, 0.05)
+        #    => 
+        #      [
+        #        [false, false, true,  true,  true],
+        #        [false, false, false, true,  true],
+        #        [false, false, false, true,  false],
+        #        [false, false, false, false, true],
+        #        [false, false, false, false, false],
+        #     ]
+        def bonferrono_test(xi, a)
+            ret = @nonParamTest.bonferronoTest(xi.to_java(Java::double[]), a)
+            return ret.to_a
+        end
+
     end
 end

data/lib/num4anova.rb

@@ -151,6 +151,25 @@ module Num4AnovaLib
         def replicate_test(xi, a)
             return @oneWay.replicateTest(xi.to_java(Java::double[]), a)
         end
+        # クラスカル・ウォリスの検定
+        #
+        # @overload kruskalwallis_test(xi, a)
+        #   @param [array]  xi データ(double[][])
+        #   @param [double] a         有意水準
+        #   @return [boolean] 検定結果(true:棄却域内 false:棄却域外)
+        #   xi = [
+        #       [12.2, 18.8, 18.2],
+        #       [22.2, 20.5, 14.6],
+        #       [20.8, 19.5, 26.3],
+        #       [26.4, 32.5, 31.3],
+        #       [24.5, 21.2, 22.4],
+        #   ]
+        #   oneWay = Num4AnovaLib::OneWayLayoutLib.new 
+        #   oneWay.kruskalwallis_test(xi, 0.05)
+        #   => true
+        def kruskalwallis_test(xi, a)
+            return @oneWay.kruskalWallisTest(xi.to_java(Java::double[]), a)
+        end
     end
 
     # 二元配置の分散分析

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: num4anova
 version: !ruby/object:Gem::Version
-  version: 0.0.13
+  version: 0.0.15
 platform: java
 authors:
 - siranovel
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2024-04-04 00:00:00.000000000 Z
+date: 2024-04-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake