pytorch学习8之kaggle房价预测

数据集在资源处可下载。

将数据集下载到本地以后，存在代码的/data路径下。

路径问题，可以通过下面语句进行查询，并判断

import hashlib
import os
import tarfile
import zipfile

import pandas as pd
import requests
import torch
from torch import nn
from d2l import torch as d2l
import sys

path = os.path.abspath(os.path.dirname(sys.argv[0]))
print("path", path)
train_data = pd.read_csv(path + "/data/kaggle_house_pred_train.csv")
test_data = pd.read_csv(path + "/data/kaggle_house_pred_test.csv")

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

# 查看数据集样式
print(train_data.iloc[0:4, [0, 1, 2, 3, -3, -2, -1]])

# 将第一列特征ID，即标签，删除,然后拼接训练集和测试集
all_feature = pd.concat((train_data.iloc[:, 1:-1], test_data.iloc[:, 1:]))

# 处理缺失值，将缺失值替换为相应特征值。通过将特征重新缩放到零均值和单位方差标准化数据特征
# 获取数值型特征的 index
numeric_features = all_feature.dtypes[all_feature.dtypes != 'object'].index
print(numeric_features)
# 将所有数值特征进行缩放处理。把数值列的某一数值减去全列的均值然后除以方差
all_feature[numeric_features] = all_feature[numeric_features].apply(
    lambda x: (x - x.mean()) / (x.std())
)
print(all_feature[numeric_features])
# 将没有采样的部分全部变为0
all_feature[numeric_features] = all_feature[numeric_features].fillna(0)

# 处理离散值。使用onehot编码替换
all_feature = pd.get_dummies(all_feature, dummy_na=True)
print(all_feature.shape)

# 从pd格式中提取np格式，并将其转换为张量模式
n_train = train_data.shape[0]
train_features = torch.tensor(all_feature[:n_train].values, dtype=torch.float32)
test_features = torch.tensor(all_feature[n_train:].values, dtype=torch.float32)
train_labels = torch.tensor(train_data.SalePrice.values.reshape(-1, 1), dtype=torch.float32)

print("train_feature", train_features.shape)
print("test_features", test_features.shape)
# 训练
loss = nn.MSELoss()
in_features = train_features.shape[1]


def get_net():
    # 单层线性回归
    net = nn.Sequential(nn.Linear(in_features, 1))
    return net


# 用价格预测的对数来衡量差异。（真实值-相对误差）/真实值
# 对比较大的正值做回归时，使用log
def log_rmse(net, features, labels):
    clipped_preds = torch.clamp(net(features), 1, float('inf'))
    # 将预测值做log，真值做log，然后回归
    rmse = torch.sqrt(loss(torch.log(clipped_preds), torch.log(labels)))
    return rmse.item()


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)
    # adam学习率的应用范围较为广
    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()
            l = loss(net(X), y)
            l.backward()
            optimizer.step()
        train_ls.append(log_rmse(net, train_features, train_labels))
        if test_ls.append(log_rmse(net, train_features, train_labels)):
            test_ls.append(log_rmse(net, test_features, test_labels))
    return train_ls, test_ls


# k折交叉验证
# 给定k折，切分训练集和验证集
def get_k_fold_data(k, i, X, y):
    assert k > 1
    # 每一折的大小为X的形状除以k
    fold_size = X.shape[0] // k
    X_train, y_train = None, None
    for j in range(k):
        # idx为第j折的样本
        idx = slice(j * fold_size, (j + 1) * fold_size)
        X_part, y_part = X[idx, :], y[idx]
        # 第i各部分作为验证集
        if j == i:
            X_vaild, y_vaild = X_part, y_part
        elif X_train is None:
            X_train, y_train = X_part, y_part
        else:
            X_train = torch.cat([X_train, X_part], 0)
            y_train = torch.cat([y_train, y_part], 0)
    # print("X_train.shape", X_train.shape)
    # print("y_train", y_train.shape)
    # print("X_vaild", X_vaild.shape)
    # print("y_vaild", y_vaild.shape)
    return X_train, y_train, X_vaild, y_vaild


# 做k折验证，返回训练和验证的K折预测平均值
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):
        print("i=", i + 1)
        # print("获取k折切分数据")
        data = get_k_fold_data(k, i, X_train, y_train)
        # print("获取网络")
        net = get_net()
        # print("预训练")
        train_ls, valid_ls = train(net, *data, num_epochs, learning_rate, weight_decay, batch_size)

        train_l_sum += train_ls[-1]
        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'fold {i + 1}, train log rmse {float(train_ls[-1])}'
              f'fold log rmse {float(valid_ls[-1]):f}')
    return train_l_sum / k, valid_l_sum / k


print("+++++++++++++++++++++++++K折交叉验证++++++++++++++++++++++++")
k, num_epochs, lr, weight_decay, batch_size = 5, 100, 5, 0, 64
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}')

import numpy as np
# 提交结果
def train_and_pred(train_features, test_feature, train_labels, test_data, num_epochs, lr, weight_decay, batch_size):
    net = get_net()
    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}')
    # 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,
               num_epochs, lr, weight_decay, batch_size)