1 importjava.io.BufferedReader;2 importjava.io.File;3 importjava.io.FileReader;4 importjava.io.IOException;5

6 /**

7 * 线性回归8 *@authorCassieRyu9 *批量梯度下降法10 */

11 public classLinear {12

13 private double[][] trainData;//训练数据集

14 private int row; //训练集的行(样本数目)

15 private int column; //训练集的列(特征数目+2)第一行是人为添加的x0， 最后一列为y值

16

17 private double[] theta; //参数theta向量

18 private double alpha; //步长

19 private int iteration; //迭代次数20

21 //构造函数

22 public Linear(String fileName, double alpha, intite){23 int rowF=getRowFromFile(fileName);24 int columnF =getColumnFromFile(fileName);25

26 row =rowF;27 column = columnF+1; //为了计算方便，加上x0那行

28

29 trainData = new double[row][column];30 loadTrainData(fileName);31

32 this.alpha =alpha;33 this.iteration =ite;34

35 theta = new double[column-1];//减去y对应的那行

36 initializeTheta();//theta的初始化

37

38 trainedTheta();//训练后的theta值

39 }40

41 //返回训练集的样本数目row

42 public intgetRowFromFile(String fileName){43

44 int count=0;45 File file = newFile(fileName);46 BufferedReader br = null;47 try{48 br = new BufferedReader(newFileReader(file));49 String temp = null;50 while((temp = br.readLine())!=null){ //循环读取下一行

51 count++;52 }53 }catch(IOException e){54 e.printStackTrace();55 }finally{56 if(br!=null)57 try{58 br.close();59 }catch(IOException e1){60

61 }62 }63

64 returncount;65 }66

67 //返回训练集列数，不包含x0

68 public intgetColumnFromFile(String fileName){69

70 int count=0;71 File file = newFile(fileName);72 BufferedReader br = null;73 try{74 br = new BufferedReader(newFileReader(file));75 String temp = null;76 if((temp = br.readLine())!=null){77 String [] tempStr = temp.split(" ");//用空格将列分开

78 count = tempStr.length;//数组长度为列的数目

79 }80 }catch(IOException e){81 e.printStackTrace();82 }finally{83 if(br!=null)84 try{85 br.close();86 }catch(IOException e1){87

88 }89 }90

91 returncount;92 }93

94 //返回训练集

95 public voidloadTrainData(String fileName){96

97 //初始化x0为1

98 for(int i=0;i

101 File file = newFile(fileName);102 BufferedReader br = null;103 try{104 br = new BufferedReader(newFileReader(file));105 String temp = null;106 int count=0;107 while((temp = br.readLine())!=null){ //行循环

108 String [] tempStr = temp.split(" ");//用空格将列分开

109

110 for(int i=1;i

111 trainData[count][i] = Double.parseDouble(tempStr[i-1]);112 count++; //行号加1

113 }114 }catch(IOException e){115 e.printStackTrace();116 }finally{117 if(br!=null)118 try{119 br.close();120 }catch(IOException e1){121

122 }123 }124 }125

126 //初始化theta的值

127 public voidinitializeTheta(){128

129 for(int i=0;i

133 //训练theta的值

134 public voidtrainedTheta(){135

136 while((iteration--)>0){//迭代次数137

138 //每迭代一次需带入新的theta值重新计算一次h(xi)-y(i)

139 double[] temp = new double[row];140 temp=getDerivation(); //h(xi)-y(i)

141

142 for(int j=0;j

143 double []tep = new double[row];144 double result=0.0;145 for(int i=0;i

147 result+=tep[i];148 }149 theta[j]-= alpha*result;150 }151 }152 }153

154 //得到(theta(k)*X(ik)-Y(i))即(h(xj)-yj)

155 public double[] getDerivation(){156

157 double [] deff = new double[row];158

159 for(int i=0;i

166 //得到theta(k)*X(ik)

167 public double getHypothesisFunc(int i){ //i为具体的某一行

168

169 double result=0;170 for(int k=0;k

176 //打印训练集

177 public voidprintTrainData(){178

179 System.out.printf("\n训练集：\n");180

181 for(int i=0;i

184 for(int j=0;j

191 }192

193 //打印theta值

194 public voidprintTheta(){195

196 System.out.printf("Theta集：\n");197 for(int j=0;j

203 //预测过程，即将theta带入h函数

204 public double predict(double[] newData){205

206 double h=0.0;207 for(int i=0;i