java实现回归_线性、逻辑回归的java实现

线性回归和逻辑回归的实现大体一致，将其抽象出一个抽象类Regression，包含整体流程，其中有三个抽象函数，将在线性回归和逻辑回归中重写。

将样本设为Sample类，其中采用数组作为特征的存储形式。

1. 样本类Sample

1 public classSample {2

3 double[] features;4 int feaNum; //the number of sample's features

5 double value; //value of sample in regression

6 int label; //class of sample

7

8 public Sample(intnumber) {9 feaNum =number;10 features = new double[feaNum];11 }12

13 public voidoutSample() {14 System.out.println("The sample's features are:");15 for(int i = 0; i < feaNum; i++) {16 System.out.print(features[i] + " ");17 }18 System.out.println();19 System.out.println("The label is: " +label);20 System.out.println("The value is: " +value);21 }22 }

2. 抽象类Regression

public abstract classRegression {double[] theta; //parameters

int paraNum; //the number of parameters

double rate; //learning rate

Sample[] sam; //samples

int samNum; //the number of samples

double th; //threshold value

/*** initialize the samples

*@params : training set

*@paramnum : the number of training samples*/

public void Initialize(Sample[] s, intnum) {

samNum=num;

sam= newSample[samNum];for(int i = 0; i < samNum; i++) {

sam[i]=s[i];

}

}/*** initialize all parameters

*@parampara : theta

*@paramlearning_rate

*@paramthreshold*/

public void setPara(double[] para, double learning_rate