Pytorch与深度学习－02.回归预测

线性回归是最容易理解的一种预测方式。线性方程Y = aX+b，大家都认识。回归预测就是知道一堆X和Y的值，计算出最接近真实a和b的两个值。这个一般运用在连续的，线性的数据预测上。

这里我将举一个线性回归的例子。线性就必须是连续的数据，Y=aX+b，是最好理解的线性方程，我们随机给出一组连续的X值（100个），然后把这些数据代入Y=aX+b方程里求出100个Y，a的值我取2.5，b的值我取随机数。这样，我就得到一组经过“抖动”后的数据。并用这两组数据X和Y去训练，求出最接近的a`和b`。

然后我再用求出的a`和b`和原始数据X套入公式，求出Y，然后画出Y和Y`。比较两组数据差异。

>>>示例程序

import numpy as np
import torch
import matplotlib.pyplot as plt

def prepare_db():
    #生成100个点的X值
    train_X = np.linspace(-2*np.pi,2*np.pi,100)
    
    #根据train_db的值，生成相应的Y=aX+b值，并进行随机加减
    a=2.5
    train_Y = train_X*a+np.random.rand(100)
    
    train_X = train_X.reshape(-1,1)
    train_Y = train_Y.reshape(-1,1)
    
    return train_X,train_Y
#=============================================
class LinearRegression(torch.nn.Module):

    def __init__(self):
        super().__init__()
        self.linear = torch.nn.Linear(1,1)
        
    def forward(self,x):
        out = self.linear(x)
        return out
    
#==============================================
class Linear_Model():
    def __init__(self):
        self.learning_rate = 0.001
        self.epoches = 10000
        self.loss_function = torch.nn.MSELoss()
        self.create_model()
        
    def create_model(self):
        self.model = LinearRegression()
        self.optimizer = torch.optim.SGD(self.model.parameters(), lr=self.learning_rate)
    
    def train(self, train_X,train_Y, model_save_path="model.pth"):
     
        x = torch.tensor(train_X).float()
        y = torch.tensor(train_Y).float()
        
        for epoch in range(self.epoches):
            prediction = self.model(x)
            loss = self.loss_function(prediction, y)

            self.optimizer.zero_grad()
            loss.backward()
            self.optimizer.step()

            if epoch % 1000 == 0:
                print("epoch: {}, loss is: {}".format(epoch, loss.item()))
                
        torch.save(self.model.state_dict(), "linear.pth")

        
    def test(self,test_db,model_open_path='model.pth'):
        self.model.load_state_dict(torch.load(model_open_path))
        prediction = self.model(torch.tensor(test_db).float())
        return prediction.detach().numpy()

#==================================================
if __name__ == '__main__':
    
    train_X,train_Y = prepare_db()
    linear = Linear_Model()    
    linear.train(train_X,train_Y)
    ret = linear.test(train_X,"linear.pth")
    
    plt.plot(train_Y,'g')
    plt.plot(ret,'r')
    plt.show()

>>>运行结果