{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prerequisites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "!pip install numpy matplotlib pandas sklearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Python 3.7.9\n"
     ]
    }
   ],
   "source": [
    "!python -V"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "np.set_printoptions(suppress=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "导入原始 csv 数据"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = pd.read_csv('./data.csv', header=None)\n",
    "X = X.to_numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "原始数据大小 100 00 00 x 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000000, 300)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_1 = np.linspace(10, 250, 100, endpoint=True)\n",
    "layer_2 = np.linspace(0.5, 10, 100, endpoint=True)\n",
    "layer_3 = np.linspace(710, 1700, 100, endpoint=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "训练集一共有 100 00 00 条数据，预测 3 个输出 (实际每个模型每次只预测1个值，训练3个模型)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = np.zeros((X.shape[0], 3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "i = 0\n",
    "for l3 in layer_3:\n",
    "    for l2 in layer_2:\n",
    "        for l1 in layer_1:\n",
    "            y[i] = [l3, l2, l1]\n",
    "            i = i + 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000000, 3)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train test split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "划分训练集，测试集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.05, random_state=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from torch.utils.data import Dataset, TensorDataset, DataLoader\n",
    "from torch.utils.data.dataset import random_split"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "指定当前模型预测的值"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "property = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "转换数据为 pyrotch 类型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train_tensor = torch.from_numpy(X_train).float()\n",
    "y_train_tensor = torch.from_numpy(y_train[:, property]).float()\n",
    "\n",
    "x_test_tensor = torch.from_numpy(X_test).float()\n",
    "y_test_tensor = torch.from_numpy(y_test[:, property]).float()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Builds dataset with ALL data\n",
    "origin_train_dataset = TensorDataset(x_train_tensor, y_train_tensor)\n",
    "\n",
    "# Splits randomly into train and validation datasets\n",
    "train_dataset, val_dataset = random_split(origin_train_dataset, [int(x_train_tensor.shape[0] * 0.9), int(x_train_tensor.shape[0] * 0.1)])\n",
    "\n",
    "# Builds a loader for each dataset to perform mini-batch gradient descent\n",
    "train_loader = DataLoader(dataset=train_dataset, batch_size=2000)\n",
    "val_loader = DataLoader(dataset=val_dataset, batch_size=2000)\n",
    "\n",
    "test_dataset = TensorDataset(x_test_tensor, y_test_tensor)\n",
    "test_loader  = DataLoader(dataset=test_dataset, batch_size=2000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Net(nn.Module):\n",
    "\n",
    "    def __init__(self):\n",
    "        super(Net, self).__init__()\n",
    "        self.bn1 = nn.BatchNorm1d(X.shape[1])\n",
    "        self.fc1 = nn.Linear(X.shape[1], 100)\n",
    "        self.bn2 = nn.BatchNorm1d(100)\n",
    "        self.fc2 = nn.Linear(100, 50)\n",
    "        self.fc3 = nn.Linear(50, 10)\n",
    "        self.fc4 = nn.Linear(10, 1)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.bn1(x)\n",
    "        x = self.fc1(x)\n",
    "        x = torch.tanh(x)\n",
    "        x = self.bn2(x)\n",
    "        x = self.fc2(x)\n",
    "        x = torch.tanh(x)\n",
    "        x = self.fc3(x)\n",
    "        x = torch.relu(x)\n",
    "        x = self.fc4(x)\n",
    "        x = torch.relu(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.optim as optim"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "如果有 GPU 优先用 GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = 'cuda' if torch.cuda.is_available() else 'cpu'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_epochs = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_train_step(model, loss_fn, optimizer):\n",
    "    def train_step(x, y):\n",
    "        model.train()\n",
    "        yh = model(x)\n",
    "        yh = torch.reshape(yh, (-1,))\n",
    "        loss = loss_fn(y, yh)\n",
    "        loss.backward()\n",
    "        torch.nn.utils.clip_grad_norm_(model.parameters(), 0.25)\n",
    "        optimizer.step()\n",
    "        optimizer.zero_grad()\n",
    "        return loss.item()\n",
    "    return train_step"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Net().to(device)\n",
    "\n",
    "loss_fn = nn.MSELoss(reduction='mean')\n",
    "\n",
    "# optimizer = optim.SGD(model.parameters(), lr=0.01)\n",
    "optimizer = optim.Adam(model.parameters(), lr=0.001, weight_decay=0.001)\n",
    "\n",
    "train_step = make_train_step(model, loss_fn, optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Net(\n",
       "  (bn1): BatchNorm1d(300, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (fc1): Linear(in_features=300, out_features=100, bias=True)\n",
       "  (bn2): BatchNorm1d(100, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "  (fc2): Linear(in_features=100, out_features=50, bias=True)\n",
       "  (fc3): Linear(in_features=50, out_features=10, bias=True)\n",
       "  (fc4): Linear(in_features=10, out_features=1, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1] Training loss: 3.309\t Validation loss: 0.564\n",
      "[2] Training loss: 0.227\t Validation loss: 0.171\n",
      "[3] Training loss: 0.170\t Validation loss: 0.160\n",
      "[4] Training loss: 0.146\t Validation loss: 0.228\n",
      "[5] Training loss: 0.129\t Validation loss: 0.154\n",
      "[6] Training loss: 0.124\t Validation loss: 0.130\n",
      "[7] Training loss: 0.121\t Validation loss: 0.112\n",
      "[8] Training loss: 0.119\t Validation loss: 0.106\n",
      "[9] Training loss: 0.116\t Validation loss: 0.105\n",
      "[10] Training loss: 0.115\t Validation loss: 0.118\n",
      "[11] Training loss: 0.113\t Validation loss: 0.104\n",
      "[12] Training loss: 0.114\t Validation loss: 0.104\n",
      "[13] Training loss: 0.113\t Validation loss: 0.105\n",
      "[14] Training loss: 0.112\t Validation loss: 0.101\n",
      "[15] Training loss: 0.113\t Validation loss: 0.109\n",
      "[16] Training loss: 0.112\t Validation loss: 0.104\n",
      "[17] Training loss: 0.112\t Validation loss: 0.128\n",
      "[18] Training loss: 0.111\t Validation loss: 0.118\n",
      "[19] Training loss: 0.112\t Validation loss: 0.110\n",
      "[20] Training loss: 0.111\t Validation loss: 0.099\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-51-1e80e92b7064>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      7\u001b[0m         \u001b[0mx_batch\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mx_batch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mto\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdevice\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      8\u001b[0m         \u001b[0my_batch\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0my_batch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mto\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdevice\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 9\u001b[1;33m         \u001b[0mloss\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtrain_step\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx_batch\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_batch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     10\u001b[0m         \u001b[0mbatch_losses\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mloss\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[0mtraining_loss\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbatch_losses\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-48-5120fbea42f1>\u001b[0m in \u001b[0;36mtrain_step\u001b[1;34m(x, y)\u001b[0m\n\u001b[0;32m      2\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mtrain_step\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      3\u001b[0m         \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 4\u001b[1;33m         \u001b[0myh\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      5\u001b[0m         \u001b[0myh\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0myh\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m(\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      6\u001b[0m         \u001b[0mloss\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mloss_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0myh\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32md:\\anaconda3\\envs\\pytorch\\lib\\site-packages\\torch\\nn\\modules\\module.py\u001b[0m in \u001b[0;36m_call_impl\u001b[1;34m(self, *input, **kwargs)\u001b[0m\n\u001b[0;32m    725\u001b[0m             \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    726\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 727\u001b[1;33m             \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    728\u001b[0m         for hook in itertools.chain(\n\u001b[0;32m    729\u001b[0m                 \u001b[0m_global_forward_hooks\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-42-15bd87424a7c>\u001b[0m in \u001b[0;36mforward\u001b[1;34m(self, x)\u001b[0m\n\u001b[0;32m     12\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     13\u001b[0m         \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbn1\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 14\u001b[1;33m         \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfc1\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     15\u001b[0m         \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtanh\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     16\u001b[0m         \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbn2\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32md:\\anaconda3\\envs\\pytorch\\lib\\site-packages\\torch\\nn\\modules\\module.py\u001b[0m in \u001b[0;36m_call_impl\u001b[1;34m(self, *input, **kwargs)\u001b[0m\n\u001b[0;32m    725\u001b[0m             \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    726\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 727\u001b[1;33m             \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    728\u001b[0m         for hook in itertools.chain(\n\u001b[0;32m    729\u001b[0m                 \u001b[0m_global_forward_hooks\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32md:\\anaconda3\\envs\\pytorch\\lib\\site-packages\\torch\\nn\\modules\\linear.py\u001b[0m in \u001b[0;36mforward\u001b[1;34m(self, input)\u001b[0m\n\u001b[0;32m     91\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     92\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minput\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mTensor\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m->\u001b[0m \u001b[0mTensor\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 93\u001b[1;33m         \u001b[1;32mreturn\u001b[0m \u001b[0mF\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlinear\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mweight\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbias\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     94\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     95\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mextra_repr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m->\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32md:\\anaconda3\\envs\\pytorch\\lib\\site-packages\\torch\\nn\\functional.py\u001b[0m in \u001b[0;36mlinear\u001b[1;34m(input, weight, bias)\u001b[0m\n\u001b[0;32m   1688\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0minput\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdim\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m2\u001b[0m \u001b[1;32mand\u001b[0m \u001b[0mbias\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1689\u001b[0m         \u001b[1;31m# fused op is marginally faster\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1690\u001b[1;33m         \u001b[0mret\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0maddmm\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbias\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mweight\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1691\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1692\u001b[0m         \u001b[0moutput\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0minput\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmatmul\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mweight\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "training_losses = []\n",
    "validation_losses = []\n",
    "\n",
    "for epoch in range(n_epochs):\n",
    "    batch_losses = []\n",
    "    for x_batch, y_batch in train_loader:\n",
    "        x_batch = x_batch.to(device)\n",
    "        y_batch = y_batch.to(device)\n",
    "        loss = train_step(x_batch, y_batch)\n",
    "        batch_losses.append(loss)\n",
    "    training_loss = np.mean(batch_losses)\n",
    "    training_losses.append(training_loss)\n",
    "\n",
    "    with torch.no_grad():\n",
    "        val_losses = []\n",
    "        for x_val, y_val in val_loader:\n",
    "            x_val = x_val.to(device)\n",
    "            y_val = y_val.to(device)\n",
    "            model.eval()\n",
    "            yh = model(x_val)\n",
    "            yh = torch.reshape(yh, (-1,))\n",
    "            val_loss = loss_fn(y_val, yh).item()\n",
    "            val_losses.append(val_loss)\n",
    "        validation_loss = np.mean(val_losses)\n",
    "        validation_losses.append(validation_loss)\n",
    "\n",
    "    print(f\"[{epoch+1}] Training loss: {training_loss:.3f}\\t Validation loss: {validation_loss:.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mean_absolute_percentage_error(y_true, y_pred):\n",
    "    return torch.mean(torch.abs((y_true - y_pred) / y_true)) * 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_test_tensor = x_test_tensor.to(device)\n",
    "y_test_tensor = y_test_tensor.to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "用训练好的模型进行预测"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = model(x_test_tensor).squeeze()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "计算 loss function (MSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(0.0992, grad_fn=<MseLossBackward>)\n"
     ]
    }
   ],
   "source": [
    "test_loss = loss_fn(y_test_tensor, y_pred)\n",
    "print(test_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mean of absolute percentage error for C: 3.82%\n"
     ]
    }
   ],
   "source": [
    "print(f\"The mean of absolute percentage error for C: {mean_absolute_percentage_error(y_test_tensor.cpu(), y_pred.cpu()):.2f}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}