1.8.Kaggle房价预测

House Prices - Advanced Regression Techniques | Kaggle

在这里下载数据，然后使用pandas读。

课本：4.10. 实战Kaggle比赛：预测房价 — 动手学深度学习 2.0.0 documentation (d2l.ai)

一层线性层

def get_net():
    net = nn.Sequential(nn.Linear(in_features, 1))  # 输出房价
    return net
  k, num_epochs, lr, weight_decay, batch_size = 5, 100, 5, 0, 64
 

MLP

net = nn.Sequential(nn.Flatten(), nn.Linear(in_features, 128), nn.ReLU(), nn.Linear(128, 1))
k, num_epochs, lr, weight_decay, batch_size = 5, 300, 5, 6, 64

Xarvier初始化,MLP

def get_net():
    #net = nn.Sequential(nn.Linear(in_features, 1))  # 输出房价
    net = nn.Sequential(nn.Flatten(), nn.Linear(in_features, 128), nn.ReLU(), nn.Linear(128, 1))
    return net

def init_weights(m):
    if type(m) == nn.Linear:
        nn.init.xavier_normal_(m.weight)
        if m.bias is not None:
            nn.init.zeros_(m.bias)
            
k, num_epochs, lr, weight_decay, batch_size = 5, 100, 0.1, 0.2, 128

完整代码

import numpy as np
import pandas as pd
import torch
from torch import nn
from d2l import torch as d2l


train_data = pd.read_csv('D:/a-learn/summer_AI/kaggle/HousePrices/train.csv')
test_data = pd.read_csv('D:/a-learn/summer_AI/kaggle/HousePrices/test.csv')

print(train_data.shape)
print(test_data.shape)

print(train_data.iloc[0:4, [0, 1, 2, 3, -3, -2, -1]])
# 可以看到第一列特征是ID，对预测没有帮助，直接去掉
# train里面的最后一列是需要预测的值，这样train和test都是80行了
all_features = pd.concat((train_data.iloc[:, 1:-1], test_data.iloc[:, 1:]))

'''数据预处理
    将所有缺失的值替换为相应特征的平均值，通过将特征重新缩放到零均值和单位方差来标准化数据
    下面先处理值为数字的特征，在处理值离散的特征
'''
numeric_features = all_features.dtypes[all_features.dtypes != 'object'].index  # 如果dtype不是object，就是数值特征
all_features[numeric_features] = all_features[numeric_features].apply(
    lambda x: (x - x.mean()) / (x.std())  # 归一化
)  # 将方差变为1
all_features[numeric_features] = all_features[numeric_features].fillna(0)  # 归一化后再将NaN填为0
# 处理离散值

# dummy_na意为值为NaN意为没有特征，pandas会帮我们处理NaN的值,注意get_dummies自动赋的是布尔值，需要自己使用dtype来调整
all_features = pd.get_dummies(all_features, dummy_na=True, dtype=int)

# 至此已经全部处理好了，最后通过values属性，可以从pandas格式中提取NumPy格式，并将其转换为张量表示用于训练
n_train = train_data.shape[0]  # 训练集数据的个数
# 将数据转换成为张量
train_features = torch.tensor(all_features[:n_train].values, dtype=torch.float32)
test_features = torch.tensor(all_features[n_train:].values, dtype=torch.float32)
# reshape(-1,1)将Numpy数组形状转换为一个二维数组，确保每个样本都有一个输出，即从形状（n,)转换为(n,1)，n为样本数量
train_labels = torch.tensor(train_data.SalePrice.values.reshape(-1, 1), dtype=torch.float32)

'''训练'''

loss = nn.MSELoss()
in_features = train_features.shape[1]  # 输入的特征数


def get_net():
    #net = nn.Sequential(nn.Linear(in_features, 1))  # 输出房价
    net = nn.Sequential(nn.Flatten(), nn.Linear(in_features, 128), nn.ReLU(), nn.Linear(128, 1))
    return net

def init_weights(m):
    if type(m) == nn.Linear:
        nn.init.xavier_normal_(m.weight)
        if m.bias is not None:
            nn.init.zeros_(m.bias)

# 对于房价，我们更关心相对误差(y-y')/y.可以使用对数来衡量差异
'''对数均方根误差'''


def log_rmse(net, features, labels):
    clipped_preds = torch.clamp(net(features), 1, float('inf'))  # 在取对数时，确保所有预测值至少为 1，以避免对数计算时出现负无穷或未定义的情况
    rmse = torch.sqrt(loss(torch.log(clipped_preds), torch.log(labels)))
    return rmse.item()  # 将张量转换为Python标量值


def train(net, train_features, train_labels, test_features, test_labels,
          num_epochs, learning_rate, weight_decay, batch_size):
    train_ls, test_ls = [], []
    train_iter = d2l.load_array((train_features, train_labels), batch_size)
    # 这里使用的是Adam优化算法，对初始学习率不是很敏感
    optimizer = torch.optim.Adam(net.parameters(),
                                 lr=learning_rate,
                                 weight_decay=weight_decay)
    for epoch in range(num_epochs):
        for X, y in train_iter:
            optimizer.zero_grad()  # 梯度清0
            l = loss(net(X), y)
            l.backward()
            optimizer.step()
        train_ls.append(log_rmse(net, train_features, train_labels))
        if test_labels is not None:
            test_ls.append(log_rmse(net, test_features, test_labels))
    return train_ls, test_ls


# K折交叉验证
# 得到第i折的数据
def get_k_fold_data(k, i, X, y):  # 分别是划分数，选取第几部分为验证集，输入，输出
    assert k > 1
    fold_size = X.shape[0] // k
    X_train, y_train = None, None
    for j in range(k):
        idx = slice(j * fold_size, (j + 1) * fold_size)
        X_part, y_part = X[idx, :], y[idx]
        if j == i:  # 验证集
            X_valid, y_valid = X_part, y_part
        elif X_train is None:
            X_train, y_train = X_part, y_part  # 训练集为空则赋值
        else:
            X_train = torch.cat([X_train, X_part], 0)  # 不为空则连接，直接接在后面就行，dim=0
            y_train = torch.cat([y_train, y_part], 0)
    return X_train, y_train, X_valid, y_valid


def k_fold(k, X_train, y_train, num_epochs, learning_rate, weight_decay,
           batch_size):
    train_l_sum, valid_l_sum = 0, 0
    for i in range(k):
        data = get_k_fold_data(k, i, X_train, y_train)
        net = get_net()
        net.apply(init_weights)
        # *data是对数据解码（取括号）,得到get_k_fold_data返回的4个数据列表，依次传入train函数中
        train_ls, valid_ls = train(net, *data, num_epochs, learning_rate,
                                   weight_decay, batch_size)
        train_l_sum += train_ls[-1]  # 注意最后一列是对数均方根误差，没问题的
        valid_l_sum += valid_ls[-1]
        if i == 0:
            d2l.plot(list(range(1, num_epochs + 1)), [train_ls, valid_ls],
                     xlabel='epoch', ylabel='rmse', xlim=[1, num_epochs],
                     legend=['train', 'valid'], yscale='log')
        print(f'折{i + 1}，训练log rmse{float(train_ls[-1]):f}, '
              f'验证log rmse{float(valid_ls[-1]):f}')
    return train_l_sum / k, valid_l_sum / k


# k, num_epochs, lr, weight_decay, batch_size = 5, 100, 0.1, 0.2, 128
# train_l, valid_l = k_fold(k, train_features, train_labels, num_epochs, lr,
#                           weight_decay, batch_size)
# print(f'{k}-折验证: 平均训练log rmse: {float(train_l):f}, '
#       f'平均验证log rmse: {float(valid_l):f}')
# d2l.plt.show()

#调好参数后，使用所有的数据作为训练，然后预测
def train_and_pred(train_features, test_features, train_labels, test_data,
                   num_epochs, lr, weight_decay, batch_size):
    net = get_net()
    net.apply(init_weights)
    train_ls, _ = train(net, train_features, train_labels, None, None,
                        num_epochs, lr, weight_decay, batch_size)
    d2l.plot(np.arange(1, num_epochs + 1), [train_ls], xlabel='epoch',
             ylabel='log rmse', xlim=[1, num_epochs], yscale='log')
    print(f'训练log rmse：{float(train_ls[-1]):f}')
    d2l.plt.show()
    # 将网络应用于测试集。
    preds = net(test_features).detach().numpy()
    # 将其重新格式化以导出到Kaggle
    test_data['SalePrice'] = pd.Series(preds.reshape(1, -1)[0])
    submission = pd.concat([test_data['Id'], test_data['SalePrice']], axis=1)
    submission.to_csv('submission.csv', index=False)

train_and_pred(train_features, test_features, train_labels, test_data,
                   100, 0.1, 0.2, 128)