李宏毅HW01——新冠疫情数据的预测

目的：熟悉熟悉pytorch

导入数据

!gdown --id '1kLSW_-cW2Huj7bh84YTdimGBOJaODiOS' --output covid.train.csv
!gdown --id '1iiI5qROrAhZn-o4FPqsE97bMzDEFvIdg' --output covid.test.csv

/Users/missbei/miniforge3/envs/NLP_search/lib/python3.8/site-packages/gdown/cli.py:127: FutureWarning: Option `--id` was deprecated in version 4.3.1 and will be removed in 5.0. You don't need to pass it anymore to use a file ID.
  warnings.warn(
Downloading...
From: https://drive.google.com/uc?id=1kLSW_-cW2Huj7bh84YTdimGBOJaODiOS
To: /Users/missbei/miniforge3/envs/NLP_search/Bilibili/Hung-Yi Lee/covid.train.csv
100%|██████████████████████████████████████| 2.49M/2.49M [00:00<00:00, 38.8MB/s]
/Users/missbei/miniforge3/envs/NLP_search/lib/python3.8/site-packages/gdown/cli.py:127: FutureWarning: Option `--id` was deprecated in version 4.3.1 and will be removed in 5.0. You don't need to pass it anymore to use a file ID.
  warnings.warn(
Downloading...
From: https://drive.google.com/uc?id=1iiI5qROrAhZn-o4FPqsE97bMzDEFvIdg
To: /Users/missbei/miniforge3/envs/NLP_search/Bilibili/Hung-Yi Lee/covid.test.csv
100%|████████████████████████████████████████| 993k/993k [00:00<00:00, 15.1MB/s]

import Package

import math
import numpy as np
import pandas as pd
import os
import csv

import torch
import torch.nn as nn
from torch.utils.data import Dataset, DataLoader, random_split

Some prepared code, no need to understand

from tqdm import tqdm
def same_seed(seed): 
    '''Fixes random number generator seeds for reproducibility.'''
    torch.backends.cudnn.deterministic = True
    torch.backends.cudnn.benchmark = False
    np.random.seed(seed)
    torch.manual_seed(seed)
    if torch.cuda.is_available():
        torch.cuda.manual_seed_all(seed)

def train_valid_split(data_set, valid_ratio, seed):
    '''Split provided training data into training set and validation set'''
    valid_set_size = int(valid_ratio * len(data_set)) 
    train_set_size = len(data_set) - valid_set_size
    train_set, valid_set = random_split(data_set, [train_set_size, valid_set_size], generator=torch.Generator().manual_seed(seed))
    return np.array(train_set), np.array(valid_set)

def predict(test_loader, model, device):
    model.eval() # Set your model to evaluation mode.
    preds = []
    for x in tqdm(test_loader):
        x = x.to(device)                        
        with torch.no_grad():                   
            pred = model(x)                     
            preds.append(pred.detach().cpu())   
    preds = torch.cat(preds, dim=0).numpy()  
    return preds

Dataset

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-SlVuXp5c-1661580334880)(attachment:%E6%88%AA%E5%B1%8F2022-08-26%20%E4%B8%8B%E5%8D%886.32.46.png)]

• 第0列是id
• 第1-37列是37个州的one-hot编码
• 38-41是COVID-like illness
• 42-49是Behavior Indicators
• 50-52是Mental Health Indicators
• 53是最后检测结果，阳不阳
• 后续是第2天-第5天的结果，都写成列

class COVID19Dataset(Dataset):
    '''
    if no y, the predict
    '''
    def __init__(self, x, y = None):
        if y is None:
            self.y = y
        else:
            self.y = torch.FloatTensor(y)
        self.x = torch.FloatTensor(x)
        
    def __getitem__(self, index):
        if self.y is None:
            return self.x[index]
        else:
            return self.x[index], self.y[index]
        
    def __len__(self):
        return len(self.x)

Neural Network Model

class NN_Model(nn.Module):
    def __init__(self, input_dim):
        super().__init__()              # 搞定多个继承关系的
        self.layers = nn.Sequential(
            nn.Linear(input_dim, 16),
            nn.ReLU(),
            nn.Linear(16, 8),
            nn.ReLU(),
            nn.Linear(8, 1)
        )
    
    def forward(self, x):
        x = self.layers(x)
        x = x.squeeze(1)                # (B, 1) ->(B)
        return x

Featurn Selection

def select_feat(train_data, valid_data, test_data, select_all = True):
    '''Selects useful features to perform regression'''
    y_train, y_valid = train_data[:, -1], valid_data[:, -1]
    raw_x_train, raw_x_valid, raw_x_test = train_data[:, :-1], valid_data[:, :-1], test_data
    
    if select_all:
        feat_idx = list(range(raw_x_train.shape[1]))
    # else: #根据传入参数自行筛选
    
    return raw_x_train[:, feat_idx], raw_x_valid[:, feat_idx], raw_x_test[:, feat_idx], y_train, y_valid

Configurations

device = 'cuda' if torch.cuda.is_available() else 'cpu'
config = {
   
    'seed' : 5201314,
    'select_all' : True,
    'valid_ratio' : 0.2,
    'n_epochs' : 3000,
    'batch_size': 256,
    'learning_rate' : 1e-5,
    'early_stop' : 400, #如果模型连续400个epoch没进步，那就停
    'save_path' : './models/model.ckpt' # save in hear       问题1:??没有s吗？？
}

Dataloader

# 让种子全部复用
same_seed(config['seed'])


# train_data.shape = (2699, 118)     id + 37states + 16feats*5days
# test_data.shape = (1078, 117)      without outcome case
train_data, test_data = pd.read_csv('./covid.train.csv').values, pd.read_csv('./covid.test.csv').values

# def train_valid_split(data_set, valid_ratio, seed)
train_data, valid_data = train_valid_split(train_data, config['valid_ratio'], config['seed'])

# 打印一下data size
print(f'''train_data size: {
     train_data.shape}
valid_data size: {
     valid_data.shape}
test_data size: {
     test_data.shape}''')

# select features    def select_feat(train_data, valid_data, test_data, select_all = True)
x_train, x_valid, x_test, y_train, y_valid = select_feat(train_data, valid_data, test_data, config['select_all'])

# 打印选择的featurn数量
print(f'''number of features: {
     x_train.shape[1]}''')

# 制作数据集
train_dataset = COVID19Dataset(x_train, y_train)
valid_dataset = COVID19Dataset(x_valid, y_valid)
test_dataset = COVID19Dataset(x_test)

# 使用dataloader
train_loader = DataLoader(train_dataset, batch_size = config['batch_size'], shuffle = True) # 问题2: pin——memory是什么？
valid_loader = DataLoader(valid_dataset, batch_size = config['batch_size'], shuffle = True)
test_loader = DataLoader(test_dataset, batch_size = config['batch_size'], shuffle = True)

train_data size: (2160, 118)
valid_data size: (539, 118)
test_data size: (1078, 117)
number of features: 117

DataLoader??

Training Loop

def trainer(train_loader, valid_loader, model, config, device):
    # Define your optimization algorithm. 
    # TODO: Please check https://pytorch.org/docs/stable/optim.html to get more available algorithms.
    # TODO: L2 regularization (optimizer(weight decay...) or implement by your self).
    
    criterion = nn.MSELoss(reduction = 'mean') # 损失函数
    optimizer = torch.optim.SGD(model.parameters(), lr = config['learning_rate'], momentum = 0.9)
    
    # 创建存放模型的目录，如果不存在的话
    if not os.path.isdir('./models'):
        os.mkdir('./models')
    
    n_epochs, best_loss, step, early_stop_count =