3.3.分类问题

# Logistic回归
import torch
from torch.autograd import Variable
import torch.nn as nn

with open('iris.csv','r') as f:
    data_list=f.readlines()
    data_list=[i.split('\n')[0] for i in data_list]
    data_list=[i.split(',') for i in data_list]
    x_data=[(float(i[0]),float(i[1]),float(i[2]),float(i[3])) for i in data_list]
    y_data=[(float(i[4])) for i in data_list]
    x_data=torch.Tensor(x_data)
    y_data=torch.Tensor(y_data).unsqueeze(1)
    if torch.cuda.is_available():
        x=Variable(x_data).cuda()
        y=Variable(y_data).cuda()
    else:
        x=Variable(x_data)
        y=Variable(y_data)

#定义Logistic回归的模型，以及二分类问题的损失函数和优化方法
class LogisticRegression(nn.Module):
    def __init__(self):
        super(LogisticRegression,self).__init__()
        self.lr=nn.Linear(4,1)
        self.sm=nn.Sigmoid()
    
    def forward(self,x):
        x=self.lr(x)
        x=self.sm(x)
        return x
if torch.cuda.is_available():
    logistic_model=LogisticRegression().cuda()
else:
    logistic_model=LogisticRegression()
    
criterion=nn.BCELoss()#二分类的损失函数
optimizer=torch.optim.SGD(logistic_model.parameters(),lr=1e-3,momentum=0.9)

#训练模型
for epoch in range(50000):
    #forward
    out=logistic_model(x)
    loss=criterion(out,y)
    print_loss=loss.data[0]
    #判断输出结果如果大于0.5就等于1，小于0.5就等于0
    mask=out.ge(0.5).float()
    correct=(mask==y).sum()
    acc=correct.data[0]/x.size(0)
    #backward
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()
    
    if(epoch+1)%1000==0:
        print('*'*10)
        print('epoch{}'.format(epoch+1))
        print('loss is {:.4f}'.format(print_loss))
        print('acc is {:.4f}'.format(acc))

**********
epoch1000
loss is 0.0667
acc is 1.0000
**********
epoch2000
loss is 0.0349
acc is 1.0000
**********
epoch3000
loss is 0.0239
acc is 1.0000
**********
epoch4000
loss is 0.0183
acc is 1.0000
**********
epoch5000
loss is 0.0149
acc is 1.0000
**********
epoch6000
loss is 0.0125
acc is 1.0000
**********
epoch7000
loss is 0.0109
acc is 1.0000
**********
epoch8000
loss is 0.0096
acc is 1.0000
**********
epoch9000
loss is 0.0086
acc is 1.0000
**********
epoch10000
loss is 0.0078
acc is 1.0000
**********
epoch11000
loss is 0.0072
acc is 1.0000
**********
epoch12000
loss is 0.0066
acc is 1.0000
**********
epoch13000
loss is 0.0062
acc is 1.0000
**********
epoch14000
loss is 0.0058
acc is 1.0000
**********
epoch15000
loss is 0.0054
acc is 1.0000
**********
epoch16000
loss is 0.0051
acc is 1.0000
**********
epoch17000
loss is 0.0048
acc is 1.0000
**********
epoch18000
loss is 0.0046
acc is 1.0000
**********
epoch19000
loss is 0.0044
acc is 1.0000
**********
epoch20000
loss is 0.0042
acc is 1.0000
**********
epoch21000
loss is 0.0040
acc is 1.0000
**********
epoch22000
loss is 0.0038
acc is 1.0000
**********
epoch23000
loss is 0.0037
acc is 1.0000
**********
epoch24000
loss is 0.0035
acc is 1.0000
**********
epoch25000
loss is 0.0034
acc is 1.0000
**********
epoch26000
loss is 0.0033
acc is 1.0000
**********
epoch27000
loss is 0.0032
acc is 1.0000
**********
epoch28000
loss is 0.0031
acc is 1.0000
**********
epoch29000
loss is 0.0030
acc is 1.0000
**********
epoch30000
loss is 0.0029
acc is 1.0000
**********
epoch31000
loss is 0.0028
acc is 1.0000
**********
epoch32000
loss is 0.0027
acc is 1.0000
**********
epoch33000
loss is 0.0026
acc is 1.0000
**********
epoch34000
loss is 0.0026
acc is 1.0000
**********
epoch35000
loss is 0.0025
acc is 1.0000
**********
epoch36000
loss is 0.0024
acc is 1.0000
**********
epoch37000
loss is 0.0024
acc is 1.0000
**********
epoch38000
loss is 0.0023
acc is 1.0000
**********
epoch39000
loss is 0.0023
acc is 1.0000
**********
epoch40000
loss is 0.0022
acc is 1.0000
**********
epoch41000
loss is 0.0022
acc is 1.0000
**********
epoch42000
loss is 0.0021
acc is 1.0000
**********
epoch43000
loss is 0.0021
acc is 1.0000
**********
epoch44000
loss is 0.0020
acc is 1.0000
**********
epoch45000
loss is 0.0020
acc is 1.0000
**********
epoch46000
loss is 0.0020
acc is 1.0000
**********
epoch47000
loss is 0.0019
acc is 1.0000
**********
epoch48000
loss is 0.0019
acc is 1.0000
**********
epoch49000
loss is 0.0018
acc is 1.0000
**********
epoch50000
loss is 0.0018
acc is 1.0000