4. 线性回归-python实现

# 数据集生成
import numpy as np
import torch
import matplotlib.pyplot as plt
import torch.utils.data as Data
from torch import nn
from torch.nn import init

num_inputs = 2
num_outputs = 1
num_samples = 1000

weight_true = torch.tensor([[2], [-3.4]])
true_w = [2,-3.4]
bais_true = -4.2
features = torch.tensor(np.random.normal(0, 1, (num_samples, num_inputs)), dtype=torch.float)
labels = torch.mm(features, weight_true) + bais_true
#features = torch.tensor(np.random.normal(0, 1, (num_samples, num_inputs)), dtype=torch.float)
#labels = true_w[0] * features[:, 0] + true_w[1] * features[:, 1] + bais_true
labels += torch.tensor(np.random.normal(0, 0.01, size=labels.size()), dtype=torch.float)

# 加载数据
#batch_size = 50
#dataset = Data.TensorDataset(features, labels)
#dataiter = Data.DataLoader(dataset, batch_size, shuffle=True)

batch_size = 10
# 将训练数据的特征和标签组合
dataset = Data.TensorDataset(features, labels)
# 随机读取小批量
data_iter = Data.DataLoader(dataset, batch_size, shuffle=True)

# 模型
class LineNet(nn.Module):
    def __init__(self, n_feature):
        super(LineNet, self).__init__()
        self.linear = nn.Linear(n_feature, 1)
    # forward 定义前向传播
    def forward(self, x):
        y = self.linear(x)
        return y

net = LineNet(num_inputs)
print(net)

# 模型参数
init.normal_(net.linear.weight, mean=0, std=0.01)
init.constant_(net.linear.bias, val=0)  # 也可以直接修改bias的data: net[0].bias.data.fill_(0)

# 损失函数
loss = nn.MSELoss()
# 算法优化
optimzer = torch.optim.SGD(net.parameters(),lr=0.04)
# 训练
num_epochs = 100
for epoch in range(num_epochs):
    train_l = 0.0
    for X,y in data_iter:
        output = net(X)
        l = loss(output, y.view(-1, 1))
        optimzer.zero_grad() # 梯度清零，等价于net.zero_grad()
        l.backward()
        optimzer.step()

        train_l+=l
    print('epoch %d ,loss %.4f' % (epoch + 1, train_l))
    print(f'权重weight:{net.linear.weight}，偏差bais{net.linear.bias}')



plt.scatter(features[:, 0].numpy(), labels.numpy(), 10, 'r')
plt.scatter(features[:, 1].numpy(), labels.numpy(), 10, 'g')
plt.show()

运行结果：

 抛出问题：在做features数据集使用的是标准正态分布，但如果std != 1 就是拟合不出结果，如果那个高手知道。麻烦指导下，谢谢。