价格分类(神经网络)

# 1.导入依赖包
import timeimport torch
import torch.nn as nn
import torch.optim as optimfrom torch.utils.data import TensorDataset, DataLoader
from sklearn.model_selection import train_test_splitimport numpy as np
import pandas as pd
import matplotlib.pyplot as pltfrom torchsummary import summary# 2.构建数据集
def create_dataset():# 2.1 读取数据集data = pd.read_csv('dataset/手机价格预测.csv')# 2.2 获取特征值和目标值，类型转化  特征（Float）  标签（Long）x, y = data.iloc[:, :-1], data.iloc[:, -1]x, y = x.astype(np.float32), y.astype(np.int64)# 2.3 数据集划分x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2,random_state=2)# 2.4 数据转Tensortrain_dataset = TensorDataset(torch.from_numpy(x_train.values), torch.tensor(y_train.values))test_dataset = TensorDataset(torch.from_numpy(x_test.values), torch.tensor(y_test.values))return train_dataset, test_dataset, x_train.shape[1], len(np.unique(y))# 3. 构建模型
class PhonePriceModel(nn.Module):def __init__(self, input_dim, output_dim):super(PhonePriceModel, self).__init__()self.linear1 = nn.Linear(input_dim, 256)self.linear2 = nn.Linear(256, 1024)self.fc = nn.Linear(1024, output_dim)def forward(self, x):x = torch.relu(self.linear1(x))x = torch.relu(self.linear2(x))output = self.fc(x)# output = torch.softmax(self.fc(x), dim=-1)return output# 4.模型训练(225)
def train(model, train_dataset, num_epochs, batch_size):# 2 初始化参数  损失函数  优化器loss1 = nn.CrossEntropyLoss()# optimizer = optim.SGD(model.parameters(), lr=1e-4, momentum=0.9)optimizer = optim.Adam(model.parameters(), lr=1e-4, betas=(0.99, 0.99))start = time.time()# 2 2个遍历  epoch  dataloaderfor epoch in range(num_epochs):dataloader = DataLoader(train_dataset, shuffle=True, batch_size=batch_size)total_num = 0total_loss = 0.0for x, y in dataloader:# 5 前向传播  损失计算 梯度归零  反向传播 参数更新output = model(x)loss = loss1(output, y)optimizer.zero_grad()loss.backward()optimizer.step()total_num += 1  # 批次total_loss += loss.item()epoch += 1print(f'epoch:{epoch + 1:4d},loss:{total_loss / (total_num * epoch):.4f}, time:{time.time() - start:.2f}s')# 模型持久化torch.save(model.state_dict(), 'model/phone2.pth')# 5.模型预测评估
def test(model, test_dataset, input_dim, output_dim):# 3.导入数据dataloader = DataLoader(test_dataset, batch_size=8, shuffle=False)correct = 0# 4.遍历数据for x, y in dataloader:# 4.1 前向传播output = model(x)print(output)# 4.2 获取输出结果（类别）y_pred = torch.argmax(output, dim=1)# print(y_pred)  # 预测错误# 4.3 计算准确率Acccorrect += (y_pred == y).sum()print(correct.item())Acc = correct.item() / len(test_dataset)return Accif __name__ == '__main__':train_dataset, test_dataset, feature_num, label_num = create_dataset()# 1.实例化模型model = PhonePriceModel(feature_num, label_num)# 2.加载模型model.load_state_dict(torch.load('model/phone2.pth'))# 模型训练# train(model, train_dataset, num_epochs=50, batch_size=8)# 模型预测Acc = test(model, test_dataset, feature_num, label_num)print(f'Acc:{Acc:.5f}')