增加分段回归

src/main/java/org/wlld/MatrixTools/MatrixOperation.java

@@ -243,7 +243,7 @@ public class MatrixOperation {
             double nub = 0;
             for (int i = 0; i < matrix.getX(); i++) {
                 for (int j = 0; j < matrix.getY(); j++) {
-                    nub = ArithUtil.add(Math.pow(matrix.getNumber(i, j), 2), nub);
+                    nub = Math.pow(matrix.getNumber(i, j), 2) + nub;
                 }
             }
             return Math.sqrt(nub);

src/main/java/org/wlld/regressionForest/Forest.java

@@ -1,12 +1,10 @@
 package org.wlld.regressionForest;
 
 import org.wlld.MatrixTools.Matrix;
+import org.wlld.MatrixTools.MatrixOperation;
 import org.wlld.tools.Frequency;
 
-import java.util.Arrays;
-import java.util.HashMap;
-import java.util.Map;
-import java.util.Random;
+import java.util.*;
 
 
 /**
@@ -24,13 +22,16 @@ public class Forest extends Frequency {
     private double resultVariance;//结果矩阵方差
     private double median;//结果矩阵中位数
     private double shrinkParameter;//方差收缩参数
-    private Matrix pc;//需要映射的基
+    private Matrix pc;//需要映射的基的集合
+    private Matrix pc1;//需要映射的基
     private double[] w;
-    private int cosSize = 10;//cos 分成几份
+    private boolean isOldG = true;//是否使用老基
+    private int oldGId = 0;//老基的id
 
-    public Forest(int featureSize, double shrinkParameter) {
+    public Forest(int featureSize, double shrinkParameter, Matrix pc) {
         this.featureSize = featureSize;
         this.shrinkParameter = shrinkParameter;
+        this.pc = pc;
         w = new double[featureSize];
     }
 
@@ -57,44 +58,22 @@ public class Forest extends Frequency {
         return equalNub;
     }
 
-    private void createG() throws Exception {//生成新基
-        double[] cg = new double[featureSize - 1];
-        Random random = new Random();
-        double sigma = 0;
-        for (int i = 0; i < featureSize - 1; i++) {
-            double rm = random.nextDouble();
-            cg[i] = rm;
-            sigma = sigma + Math.pow(rm, 2);
-        }
-        double cosOne = 1.0D / cosSize;
-        double[] ag = new double[cosSize - 1];
-        for (int i = 1; i < cosSize; i++) {
-            double cos = cosOne * i;
-            ag[i] = Math.sqrt(sigma / (1 / Math.pow(cos, 2) - 1));
-        }
-        int x = (cosSize - 1) * featureSize;
-        pc = new Matrix(x, featureSize);
-        for (int i = 0; i < featureSize; i++) {
-            Matrix matrix = new Matrix(ag.length, featureSize);
-            for (int j = 0; j < ag.length; j++) {
-                for (int k = 0; k < featureSize; k++) {
-                    if (k != i) {
-                        if (k < i) {
-                            matrix.setNub(j, k, cg[k]);
-                        } else {
-                            matrix.setNub(j, k, cg[k - 1]);
-                        }
+    private void findG() throws Exception {//寻找新的切入维度
+        // 先尝试从原有维度切入
+        int xSize = conditionMatrix.getX();
+        int ySize = conditionMatrix.getY();
+        Matrix matrix = new Matrix(xSize, ySize);
+        for (int i = 0; i < xSize; i++) {
+            for (int j = 0; j < ySize; j++) {
+                if (j < ySize - 1) {
+                    matrix.setNub(i, j, conditionMatrix.getNumber(i, j));
                 } else {
-                        matrix.setNub(j, k, ag[j]);
-                    }
-                }
+                    matrix.setNub(i, j, resultMatrix.getNumber(i, 0));
                 }
             }
         }
-
-    private void findG() throws Exception {//寻找新的切入维度
-        // 先尝试从原有维度切入
-        Map<Integer, Double> varMap = new HashMap<>();//保存原有维度方差
+        double maxOld = 0;
+        int type = 0;
         for (int i = 0; i < featureSize; i++) {
             double[] g = new double[conditionMatrix.getX()];
             for (int j = 0; j < g.length; j++) {
@@ -105,9 +84,41 @@ public class Forest extends Frequency {
                 }
             }
             double var = variance(g);//计算方差
-            varMap.put(i, var);
+            if (var > maxOld) {
+                maxOld = var;
+                type = i;
+            }
+        }
+        int x = pc.getX();
+        double max = 0;
+        for (int i = 0; i < x; i++) {
+            Matrix g = pc.getRow(i);
+            double gNorm = MatrixOperation.getNorm(g);
+            double[] var = new double[xSize];
+            for (int j = 0; j < xSize; j++) {
+                Matrix parameter = matrix.getRow(j);
+                double dist = transG(g, parameter, gNorm);
+                var[j] = dist;
+            }
+            double variance = variance(var);
+            if (variance > max) {
+                max = variance;
+                pc1 = g;
+            }
+        }
+        //找到非原始基最离散的新基
+        if (max > maxOld) {//使用新基
+            isOldG = false;
+        } else {//使用原有基
+            isOldG = true;
+            oldGId = type;
+        }
     }
 
+    private double transG(Matrix g, Matrix parameter, double gNorm) throws Exception {//将数据映射到新基
+        //先求内积
+        double innerProduct = MatrixOperation.innerProduct(g, parameter);
+        return innerProduct / gNorm;
     }
 
     public void cut() throws Exception {
@@ -123,8 +134,8 @@ public class Forest extends Frequency {
             //检测中位数median有多少个一样的值
             int equalNub = getEqualNub(median, dm);
             //System.out.println("equalNub==" + equalNub + ",y==" + y);
-            forestLeft = new Forest(featureSize, shrinkParameter);
-            forestRight = new Forest(featureSize, shrinkParameter);
+            forestLeft = new Forest(featureSize, shrinkParameter, pc);
+            forestRight = new Forest(featureSize, shrinkParameter, pc);
             Matrix conditionMatrixLeft = new Matrix(z + equalNub, featureSize);//条件矩阵左
             Matrix conditionMatrixRight = new Matrix(y - z - equalNub, featureSize);//条件矩阵右
             Matrix resultMatrixLeft = new Matrix(z + equalNub, 1);//结果矩阵左

src/main/java/org/wlld/regressionForest/RegressionForest.java

@@ -5,6 +5,9 @@ import org.wlld.MatrixTools.MatrixOperation;
 import org.wlld.tools.Frequency;
 
 import java.util.Arrays;
+import java.util.HashMap;
+import java.util.Map;
+import java.util.Random;
 
 /**
  * @param
@@ -22,6 +25,16 @@ public class RegressionForest extends Frequency {
     private double[] results;//结果数组
     private double min;//结果最小值
     private double max;//结果最大值
+    private Matrix pc;//需要映射的基
+    private int cosSize = 10;//cos 分成几份
+
+    public int getCosSize() {
+        return cosSize;
+    }
+
+    public void setCosSize(int cosSize) {
+        this.cosSize = cosSize;
+    }
 
     public RegressionForest(int size, int featureNub, double shrinkParameter) throws Exception {//初始化
         if (size > 0 && featureNub > 0) {
@@ -30,7 +43,8 @@ public class RegressionForest extends Frequency {
             results = new double[size];
             conditionMatrix = new Matrix(size, featureNub);
             resultMatrix = new Matrix(size, 1);
-            forest = new Forest(featureNub, shrinkParameter);
+            createG();
+            forest = new Forest(featureNub, shrinkParameter, pc);
             forest.setW(w);
             forest.setConditionMatrix(conditionMatrix);
             forest.setResultMatrix(resultMatrix);
@@ -91,6 +105,62 @@ public class RegressionForest extends Frequency {
         }
     }
 
+    private void createG() throws Exception {//生成新基
+        double[] cg = new double[featureNub - 1];
+        Random random = new Random();
+        double sigma = 0;
+        for (int i = 0; i < featureNub - 1; i++) {
+            double rm = random.nextDouble();
+            cg[i] = rm;
+            sigma = sigma + Math.pow(rm, 2);
+        }
+        double cosOne = 1.0D / cosSize;
+        double[] ag = new double[cosSize - 1];//装一个维度内所有角度的余弦值
+        for (int i = 1; i < cosSize; i++) {
+            double cos = cosOne * i;
+            ag[i] = Math.sqrt(sigma / (1 / Math.pow(cos, 2) - 1));
+        }
+        int x = (cosSize - 1) * featureNub;
+        pc = new Matrix(x, featureNub);
+        for (int i = 0; i < featureNub; i++) {//遍历所有的固定基
+            //以某个固定基摆动的所有新基集合的矩阵
+            Matrix matrix = new Matrix(ag.length, featureNub);
+            for (int j = 0; j < ag.length; j++) {
+                for (int k = 0; k < featureNub; k++) {
+                    if (k != i) {
+                        if (k < i) {
+                            matrix.setNub(j, k, cg[k]);
+                        } else {
+                            matrix.setNub(j, k, cg[k - 1]);
+                        }
+                    } else {
+                        matrix.setNub(j, k, ag[j]);
+                    }
+                }
+            }
+            //将一个固定基内摆动的新基都装到最大的集合内
+            int index = (cosSize - 1) * i;
+            push(pc, matrix, index);
+        }
+    }
+
+    //将两个矩阵从上到下进行合并
+    private void push(Matrix mother, Matrix son, int index) throws Exception {
+        if (mother.getY() == son.getY()) {
+            int x = index + son.getX();
+            int y = mother.getY();
+            int start = 0;
+            for (int i = index; i < x; i++) {
+                for (int j = 0; j < y; j++) {
+                    mother.setNub(i, j, son.getNumber(start, j));
+                }
+                start++;
+            }
+        } else {
+            throw new Exception("matrix Y is not equals");
+        }
+    }
+
     public void insertFeature(double[] feature, double result) throws Exception {//插入数据
         if (feature.length == featureNub - 1) {
             for (int i = 0; i < featureNub; i++) {