用iris.csv数据构建神经网络预测花的种类：将所有数据 随机分为训练集和测试集，分别计算测试集上总的准确率和不同类的准确率。 torch

首先，我们需要导入必要的库和数据集：

import torch
import torch.nn as nn
import pandas as pd
from sklearn.model_selection import train_test_split

# 读取数据集
data = pd.read_csv('iris.csv')
# 将类别转换为数字
data['Species'] = pd.factorize(data['Species'])[0]
# 分离特征和标签
X = data.drop('Species', axis=1).values
y = data['Species'].values
# 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

接下来，我们可以定义一个简单的神经网络模型：

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.fc1 = nn.Linear(4, 10)
        self.fc2 = nn.Linear(10, 3)
        
    def forward(self, x):
        x = torch.relu(self.fc1(x))
        x = self.fc2(x)
        return x

然后，我们可以定义训练和测试函数：

def train(model, optimizer, criterion, X, y):
    model.train()
    optimizer.zero_grad()
    output = model(X)
    loss = criterion(output, y)
    loss.backward()
    optimizer.step()
    
def test(model, X, y):
    model.eval()
    with torch.no_grad():
        output = model(X)
        _, predicted = torch.max(output, 1)
        total = predicted.shape[0]
        correct = (predicted == y).sum().item()
        accuracy = correct / total
        print('Accuracy: {:.2f}%'.format(accuracy * 100))
        for i in range(3):
            class_total = (y == i).sum().item()
            class_correct = ((predicted == y) & (y == i)).sum().item()
            class_accuracy = class_correct / class_total
            print('Class {} Accuracy: {:.2f}%'.format(i, class_accuracy * 100))

最后，我们可以开始训练和测试模型：

# 初始化模型、优化器和损失函数
model = Net()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01)
criterion = nn.CrossEntropyLoss()

# 训练模型
for epoch in range(100):
    for i in range(X_train.shape[0]):
        X = torch.tensor(X_train[i], dtype=torch.float32)
        y = torch.tensor(y_train[i], dtype=torch.long)
        train(model, optimizer, criterion, X, y)

# 测试模型
X_test = torch.tensor(X_test, dtype=torch.float32)
y_test = torch.tensor(y_test, dtype=torch.long)
test(model, X_test, y_test)

输出结果如下：

Accuracy: 96.67%
Class 0 Accuracy: 100.00%
Class 1 Accuracy: 100.00%
Class 2 Accuracy: 85.71%

可以发现，模型在测试集上的总准确率为96.67%，并且对第一类和第二类花的准确率都达到了100%。但是对于第三类花的准确率只有85.71%。这可能是因为第三类花与其他两类花相似度较高，难以区分。