本文为为🔗365天深度学习训练营内部文章

原作者：K同学啊

 一 导入数据

import numpy as np
import pandas as pd
import torch
from torch import nn
import torch.nn.functional as F
import seaborn as sns
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from torch.utils.data import TensorDataset,DataLoader
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings("ignore")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

'''
导入数据
'''
df = pd.read_csv('heart.csv')
print(df)

二 构建数据集 

'''
构建数据集
'''
# 1. 标准化
X = df.iloc[:,:-1]
y = df.iloc[:,-1]
sc = StandardScaler()
X = sc.fit_transform(X)

# 2.划分数据集
X = torch.tensor(np.array(X),dtype=torch.float32)
y = torch.tensor(np.array(y),dtype=torch.int64)
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.1,random_state=14)

# 3.构建数据加载器
train_dl = DataLoader(TensorDataset(X_train,y_train),batch_size=64,shuffle=False)
test_dl = DataLoader(TensorDataset(X_test,y_test),batch_size=64,shuffle=False)

 三 模型训练

'''
模型训练
'''
# 1.构建模型
class model_rnn(nn.Module):
    def __init__(self):
        super(model_rnn,self).__init__()
        self.rnn0 = nn.RNN(input_size=13,hidden_size=200,num_layers=1,batch_first=True)
        self.fc0 = nn.Linear(200,50)
        self.fc1 = nn.Linear(50,2)

    def forward(self,x):
        out,hidden1 = self.rnn0(x)
        out = self.fc0(out)
        out = self.fc1(out)
        return out

model = model_rnn()
print(model)

四 定义训练函数 

# 2.定义训练函数
# 训练循环
def train(dataloader, model, loss_fn, optimizer):
    size = len(dataloader.dataset)  # 训练集的大小，一共60000张图片
    num_batches = len(dataloader)  # 批次数目，1875（60000/32）

    train_loss, train_acc = 0, 0  # 初始化训练损失和正确率

    for X, y in dataloader:  # 获取图片及其标签
        X, y = X.to(device), y.to(device)

        # 计算预测误差
        pred = model(X)  # 网络输出
        loss = loss_fn(pred, y)  # 计算网络输出和真实值之间的差距，targets为真实值，计算二者差值即为损失

        # 反向传播
        optimizer.zero_grad()  # grad属性归零
        loss.backward()  # 反向传播
        optimizer.step()  # 每一步自动更新

        # 记录acc与loss
        train_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
        train_loss += loss.item()

    train_acc /= size
    train_loss /= num_batches

    return train_acc, train_loss

 五 定义测试函数

def t(dataloader, model, loss_fn):
    size = len(dataloader.dataset)  # 测试集的大小，一共10000张图片
    num_batches = len(dataloader)  # 批次数目，313（10000/32=312.5，向上取整）
    test_loss, test_acc = 0, 0

    # 当不进行训练时，停止梯度更新，节省计算内存消耗
    with torch.no_grad():
        for imgs, target in dataloader:
            imgs, target = imgs.to(device), target.to(device)

            # 计算loss
            target_pred = model(imgs)
            loss = loss_fn(target_pred, target)

            test_loss += loss.item()
            test_acc += (target_pred.argmax(1) == target).type(torch.float).sum().item()

    test_acc /= size
    test_loss /= num_batches

    return test_acc, test_loss

 六 训练

loss_fn = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-4 # 学习率
opt = torch.optim.Adam(model.parameters(),lr=learn_rate)
epochs = 50

train_loss = []
train_acc = []
test_loss = []
test_acc = []

for epoch in range(epochs):
    model.train()
    epoch_train_acc, epoch_train_loss = train(train_dl, model, loss_fn, opt)

    model.eval()
    epoch_test_acc, epoch_test_loss = t(test_dl, model, loss_fn)

    train_acc.append(epoch_train_acc)
    train_loss.append(epoch_train_loss)
    test_acc.append(epoch_test_acc)
    test_loss.append(epoch_test_loss)

    template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%，Test_loss:{:.3f}')
    print(template.format(epoch + 1, epoch_train_acc * 100, epoch_train_loss, epoch_test_acc * 100, epoch_test_loss))
print('Done')

Epoch: 1, Train_acc:46.7%, Train_loss:0.698, Test_acc:54.8%，Test_loss:0.689
Epoch: 2, Train_acc:59.6%, Train_loss:0.682, Test_acc:58.1%，Test_loss:0.682
Epoch: 3, Train_acc:71.3%, Train_loss:0.666, Test_acc:58.1%，Test_loss:0.676
Epoch: 4, Train_acc:77.2%, Train_loss:0.652, Test_acc:54.8%，Test_loss:0.670
Epoch: 5, Train_acc:81.6%, Train_loss:0.637, Test_acc:54.8%，Test_loss:0.665
Epoch: 6, Train_acc:82.4%, Train_loss:0.623, Test_acc:58.1%，Test_loss:0.659
Epoch: 7, Train_acc:80.9%, Train_loss:0.609, Test_acc:58.1%，Test_loss:0.654
Epoch: 8, Train_acc:81.2%, Train_loss:0.594, Test_acc:58.1%，Test_loss:0.649
Epoch: 9, Train_acc:81.6%, Train_loss:0.580, Test_acc:58.1%，Test_loss:0.644
Epoch:10, Train_acc:81.6%, Train_loss:0.565, Test_acc:58.1%，Test_loss:0.639
Epoch:11, Train_acc:82.0%, Train_loss:0.550, Test_acc:58.1%，Test_loss:0.634
Epoch:12, Train_acc:82.0%, Train_loss:0.535, Test_acc:58.1%，Test_loss:0.630
Epoch:13, Train_acc:82.0%, Train_loss:0.520, Test_acc:58.1%，Test_loss:0.625
Epoch:14, Train_acc:82.0%, Train_loss:0.504, Test_acc:58.1%，Test_loss:0.621
Epoch:15, Train_acc:82.0%, Train_loss:0.488, Test_acc:58.1%，Test_loss:0.617
Epoch:16, Train_acc:82.4%, Train_loss:0.473, Test_acc:58.1%，Test_loss:0.614
Epoch:17, Train_acc:83.1%, Train_loss:0.457, Test_acc:61.3%，Test_loss:0.611
Epoch:18, Train_acc:83.1%, Train_loss:0.442, Test_acc:61.3%，Test_loss:0.608
Epoch:19, Train_acc:83.8%, Train_loss:0.427, Test_acc:64.5%，Test_loss:0.605
Epoch:20, Train_acc:85.7%, Train_loss:0.412, Test_acc:61.3%，Test_loss:0.603
Epoch:21, Train_acc:86.4%, Train_loss:0.398, Test_acc:64.5%，Test_loss:0.601
Epoch:22, Train_acc:87.1%, Train_loss:0.383, Test_acc:64.5%，Test_loss:0.600
Epoch:23, Train_acc:87.9%, Train_loss:0.370, Test_acc:64.5%，Test_loss:0.600
Epoch:24, Train_acc:88.2%, Train_loss:0.357, Test_acc:67.7%，Test_loss:0.601
Epoch:25, Train_acc:88.2%, Train_loss:0.344, Test_acc:71.0%，Test_loss:0.602
Epoch:26, Train_acc:88.6%, Train_loss:0.332, Test_acc:71.0%，Test_loss:0.605
Epoch:27, Train_acc:89.3%, Train_loss:0.321, Test_acc:71.0%，Test_loss:0.609
Epoch:28, Train_acc:89.3%, Train_loss:0.310, Test_acc:67.7%，Test_loss:0.614
Epoch:29, Train_acc:89.3%, Train_loss:0.300, Test_acc:67.7%，Test_loss:0.620
Epoch:30, Train_acc:90.1%, Train_loss:0.290, Test_acc:67.7%，Test_loss:0.627
Epoch:31, Train_acc:90.1%, Train_loss:0.281, Test_acc:71.0%，Test_loss:0.635
Epoch:32, Train_acc:90.8%, Train_loss:0.272, Test_acc:71.0%，Test_loss:0.644
Epoch:33, Train_acc:90.8%, Train_loss:0.264, Test_acc:67.7%，Test_loss:0.654
Epoch:34, Train_acc:90.8%, Train_loss:0.257, Test_acc:64.5%，Test_loss:0.663
Epoch:35, Train_acc:90.8%, Train_loss:0.249, Test_acc:64.5%，Test_loss:0.673
Epoch:36, Train_acc:90.8%, Train_loss:0.243, Test_acc:64.5%，Test_loss:0.683
Epoch:37, Train_acc:91.9%, Train_loss:0.236, Test_acc:64.5%，Test_loss:0.693
Epoch:38, Train_acc:92.3%, Train_loss:0.230, Test_acc:64.5%，Test_loss:0.703
Epoch:39, Train_acc:92.3%, Train_loss:0.224, Test_acc:64.5%，Test_loss:0.714
Epoch:40, Train_acc:92.3%, Train_loss:0.218, Test_acc:64.5%，Test_loss:0.724
Epoch:41, Train_acc:92.3%, Train_loss:0.213, Test_acc:61.3%，Test_loss:0.736
Epoch:42, Train_acc:92.3%, Train_loss:0.207, Test_acc:61.3%，Test_loss:0.748
Epoch:43, Train_acc:93.4%, Train_loss:0.202, Test_acc:58.1%，Test_loss:0.760
Epoch:44, Train_acc:93.8%, Train_loss:0.197, Test_acc:58.1%，Test_loss:0.772
Epoch:45, Train_acc:93.8%, Train_loss:0.192, Test_acc:58.1%，Test_loss:0.784
Epoch:46, Train_acc:94.5%, Train_loss:0.187, Test_acc:58.1%，Test_loss:0.798
Epoch:47, Train_acc:94.9%, Train_loss:0.182, Test_acc:58.1%，Test_loss:0.812
Epoch:48, Train_acc:94.9%, Train_loss:0.177, Test_acc:58.1%，Test_loss:0.826
Epoch:49, Train_acc:95.2%, Train_loss:0.172, Test_acc:54.8%，Test_loss:0.842
Epoch:50, Train_acc:95.2%, Train_loss:0.167, Test_acc:54.8%，Test_loss:0.858
Done

七 结果可视化

1.准确率和损失值

plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False      # 用来正常显示负号
plt.rcParams['figure.dpi'] = 100        #分辨率

epochs_range = range(epochs)

plt.figure(figsize=(12, 3))
plt.subplot(1, 2, 1)

plt.plot(epochs_range, train_acc, label='Training Accuracy')
plt.plot(epochs_range, test_acc, label='Test Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Training Loss')
plt.plot(epochs_range, test_loss, label='Test Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()

2.混淆矩阵 

'''
绘制混淆矩阵
'''
print('=============输入数据shape为==============')
print('X_test.shape:',X_test.shape)
print('y_test.shape:',y_test.shape)

pred = model(X_test.to(device)).argmax(1).cpu().numpy()

print('\n==========输出数据shape为==============')
print('pred.shape:',pred.shape)

from sklearn.metrics import confusion_matrix,ConfusionMatrixDisplay

# 计算混淆矩阵
cm = confusion_matrix(y_test,pred)

plt.figure(figsize=(6,5))
plt.suptitle('')
sns.heatmap(cm,annot=True,fmt='d',cmap='Blues')
plt.xticks(fontsize=10)
plt.yticks(fontsize=10)
plt.title('Confusion Matrix',fontsize=12)
plt.xlabel('Pred Label',fontsize=10)
plt.ylabel('True Label',fontsize=10)
plt.tight_layout()
plt.show()

八 使用模型进行预测 

'''
使用模型进行预测
'''
test_X = X_test[0].reshape(1,-1)
pred = model(test_X.to(device)).argmax(1).item()
print('模型预测结果为：',pred)
print('=='*20)
print('0:不会患心脏病')
print('1:可能患心脏病')

模型预测结果为： 0
========================================
0:不会患心脏病
1:可能患心脏病