mlp房價預測

mlp房價預測

跟着李沐的動手學深度學習，跟着寫了一遍房價預測的處理和預測，加了一些注釋，同時稍微改動了一些地方，把線性回歸改成了mlp
由於數據集比較小而且沒有缺失值，這里也沒有去做特征工程，如果特征量比較多的話，直接用pd.dummies()會出現很多無用特征，所以在特征比較多且數據量大的情況下還是要先做特征工程，刪去一些特征
之后有空再去做kaggle上的比賽

import hashlib
import os
import tarfile
import zipfile
import requests

DATA_HUB = dict()
DATA_URL = 'http://d2l-data.s3-accelerate.amazonaws.com/'

def download(name, cache_dir = os.path.join('..', 'data')):  
    """下載一個DATA_HUB中的文件，返回本地文件名。"""
    assert name in DATA_HUB, f"{name} 不存在於 {DATA_HUB}."
    url, sha1_hash = DATA_HUB[name]
    os.makedirs(cache_dir, exist_ok=True)
    fname = os.path.join(cache_dir, url.split('/')[-1])
    if os.path.exists(fname):
        sha1 = hashlib.sha1()
        with open(fname, 'rb') as f:
            while True:
                data = f.read(1048576)
                if not data:
                    break
                sha1.update(data)
        if sha1.hexdigest() == sha1_hash:
            return fname
    print(f'正在從{url}下載{fname}...')
    r = requests.get(url, stream=True, verify=True)
    with open(fname, 'wb') as f:
        f.write(r.content)
    return fname


%matplotlib inline
import numpy as np
import pandas as pd
import torch
from torch import nn
import matplotlib.pyplot as plt
# from d2l import torch as d2l
from torch.utils import data

DATA_HUB['kaggle_house_train'] = (  
    DATA_URL + 'kaggle_house_pred_train.csv',
    '585e9cc93e70b39160e7921475f9bcd7d31219ce')

DATA_HUB['kaggle_house_test'] = (  
    DATA_URL + 'kaggle_house_pred_test.csv',
    'fa19780a7b011d9b009e8bff8e99922a8ee2eb90')

train_data = pd.read_csv(download('kaggle_house_train'))
test_data = pd.read_csv(download('kaggle_house_test'))

print(train_data.shape)
print(test_data.shape)
print(train_data.iloc[:4,[0,1,2,-3,-2,-1]])
print(test_data.iloc[:4,[0,1,2,-3,-2,-1]])

# 把訓練集+測試集的特征放到一起，訓練集的第0列是ID要去除，最后一列是標簽
all_features = pd.concat((train_data.iloc[:, 1:-1], test_data.iloc[:, 1:]), axis = 0)
print(all_features.shape)

# 找出所有數值列
numeric_features = all_features.dtypes[all_features.dtypes != 'object'].index
# 對數值列用非nan值的均值填充nan
all_features[numeric_features] = all_features[numeric_features].apply(
    lambda x : x.fillna( value = x[[y is not np.nan for y in x]].mean() ) )
# 標准化所有數值列，變成均值為0，方差為1
all_features[numeric_features] = all_features[numeric_features].apply(
    lambda x : (x - x.mean()) / x.std())
# 用OneHot編碼替換離散值
all_features = pd.get_dummies(all_features, dummy_na = True)
print(all_features.shape)



# 轉化成torch.tensor類型
n_train = train_data.shape[0]
train_features = torch.tensor(all_features[:n_train].values,
                              dtype = torch.float32)
test_features = torch.tensor(all_features[n_train:].values,
                             dtype = torch.float32)
train_labels = torch.tensor(train_data.SalePrice.values.reshape(-1,1),
                            dtype = torch.float32)



# 定義訓練用的損失函數
loss = nn.MSELoss()
# 輸入特征數
in_features = train_features.shape[1]

# 線性回歸模型
# def get_net():
#    net = nn.Sequential(nn.Linear(in_features, 1))
#    return net

# mlp
def get_net():
    net = nn.Sequential(nn.Linear(in_features, 256), nn.ReLU(), nn.Dropout(0.5), nn.Linear(256, 1))
#     net = nn.Sequential(nn.Linear(in_features, 1))
    return net

# y的值比較大，所以都先取一個log，縮小范圍，再用均方根誤差
def log_rmse(net, features, labels):
    # torch.clamp(input, min, max, out=None) → Tensor
    # 將輸入input張量每個元素的夾緊到區間 [min,max]，並返回結果到一個新張量。
    clipped_preds = torch.clamp(net(features), 1, float('inf'))
    rmse = torch.sqrt(loss(torch.log(clipped_preds), torch.log(labels)))
    return rmse.item()

# 訓練函數
def train(net, train_features, train_labels, test_features, test_labels,
         num_epochs, learning_rate, weight_decay, batch_size):
    # 數據迭代器，用於每次得到隨機的一組batch
    train_iter = data.DataLoader(dataset = data.TensorDataset(train_features, train_labels),
                                batch_size = batch_size,
                                shuffle = True,
                                num_workers = 4,
                                drop_last = True)
    # 設置優化器， 這里用了Adam
    optimizer = torch.optim.Adam(net.parameters(), lr = learning_rate,
                                weight_decay = weight_decay)
    # 保存每一輪迭代之后的損失
    train_ls, test_ls = [], []
    # num_epochs輪訓練
    for epoch in range(num_epochs):
        # 變成train模式
        net.train()
        for X, y in train_iter:
            optimizer.zero_grad()
            l = loss(net(X), y)
            l.backward()
            optimizer.step()
        # 變成eval模式
        net.eval()
        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折交叉驗證，訓練數據在第i折，X: 特征， y: 標簽
def get_k_fold_data(k, i, X, y):
    # 要保證k>1
    assert k > 1
    fold_size = X.shape[0] // k
    X_train, y_train = None, None
    for j in range(k):
        # slice用於獲取一個切片對象 https://m.runoob.com/python/python-func-slice.html
        idx = slice(j * fold_size, (j + 1) * fold_size)
        X_part, y_part = X[idx,:], y[idx]
        if j == i:
            X_valid, y_valid = 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)
    return X_train, y_train, X_valid, y_valid


# k折交叉驗證
def k_fold(k, X_train, y_train, num_epochs, learning_rate, weight_decay, batch_size):
    # k折交叉驗證的平均訓練集損失和驗證集損失
    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()
        # *data用於把data解包成X_train, y_train, X_test, y_test
        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:
            plt.figure()
            plt.xlabel('epoch')
            plt.ylabel('rmse') 
            plt.xlim([1, num_epochs])
            plt.plot(list(range(1,num_epochs + 1)), train_ls, label = 'train')
            plt.yscale('log')
            plt.plot(list(range(1,num_epochs + 1)), valid_ls, label = 'valid')
            plt.legend()
            plt.show()
        print(f'fold {i+1}, train log rmse {float(train_ls[-1]):f}, valid log rmse {float(valid_ls[-1]):f}, ')
    # 取平均損失
    return train_l_sum / k, valid_l_sum / k


k, num_epochs, lr, weight_decay, batch_size = 5, 30, 0.05, 0.3, 64
train_l, valid_l = k_fold(k, train_features, train_labels, num_epochs, lr, weight_decay, batch_size)
print(f'{k}-折驗證：平均訓練log rmse: {float(train_l):f}, 平均驗證log rmse: {float(valid_l):f}')


def train_and_pred(train_features, test_features, 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)
    print(f'train log rmse {float(train_ls[-1]):f}')
    plt.figure()
    plt.xlabel('epoch')
    plt.ylabel('rmse') 
    plt.xlim([1, num_epochs])
    plt.plot(list(range(1,num_epochs + 1)), train_ls)
    plt.yscale('log')
    plt.show()
    # 轉換成eval模式
    net.eval()
    preds = net(test_features).detach().numpy()
    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)

k折交叉驗證時的誤差:

最后訓練得到的網絡的誤差: