Lab_427_Zhaozhong_Xu8.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "566b32a9",
   "metadata": {
    "id": "566b32a9"
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import torch\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import glob\n",
    "import tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5de4b136",
   "metadata": {
    "id": "5de4b136"
   },
   "source": [
    "### Prepare Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "id": "54236276",
   "metadata": {
    "id": "54236276"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████| 30000/30000 [00:14<00:00, 2128.81it/s]\n"
     ]
    }
   ],
   "source": [
    "folder_path = '/Users/xuzhaozhong/Downloads/jets'\n",
    "x_train = []\n",
    "y_train = []\n",
    "\n",
    "for i in tqdm.trange(30000):\n",
    "    formatted_i = \"{:07}\".format(i)\n",
    "    file_path = f\"{folder_path}/event_{formatted_i}.npz\"  # Adjust the file extension if needed\n",
    "    with np.load(file_path) as event:\n",
    "        x_train.append(event['x'])\n",
    "        y_train.append(event['y'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "03b82cd9",
   "metadata": {
    "id": "03b82cd9"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████| 30000/30000 [00:00<00:00, 532310.36it/s]\n"
     ]
    }
   ],
   "source": [
    "x_train_padded = []\n",
    "\n",
    "for i in tqdm.trange(len(x_train)):\n",
    "    padded_item = np.zeros((200, 5))\n",
    "    num_rows = x_train[i].shape[0]\n",
    "\n",
    "    # Copy the original data into the padded_item\n",
    "    padded_item[:num_rows, :] = x_train[i]\n",
    "\n",
    "    x_train_padded.append(padded_item)\n",
    "\n",
    "x_train_padded_array = np.array(x_train_padded)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "690d2332",
   "metadata": {
    "id": "690d2332"
   },
   "outputs": [],
   "source": [
    "x_train = x_train_padded_array\n",
    "y_train = np.array(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "9a960b53",
   "metadata": {
    "id": "9a960b53"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(30000, 1000)\n"
     ]
    }
   ],
   "source": [
    "x_train = x_train.reshape(-1,200*5)\n",
    "print(x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "2457db52",
   "metadata": {
    "id": "2457db52"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(30000, 1, 200, 5)\n"
     ]
    }
   ],
   "source": [
    "# Scale the dataset according to standard scaling\n",
    "scaler = StandardScaler()\n",
    "x_train = scaler.fit_transform(x_train)\n",
    "x_train = x_train.reshape(-1, 1, 200, 5)\n",
    "print(x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "ca8477b1",
   "metadata": {
    "id": "ca8477b1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20000, 1, 200, 5) (20000, 5) (5000, 1, 200, 5)\n"
     ]
    }
   ],
   "source": [
    "x_test = x_train[25000:len(x_train)]\n",
    "y_test = y_train[25000:len(x_train)]\n",
    "x_validation=x_train[20000:25000]\n",
    "y_validation=y_train[20000:25000]\n",
    "x_train = x_train[:20000]\n",
    "y_train = y_train[:20000]\n",
    "print(x_train.shape,y_train.shape,x_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7db842d6",
   "metadata": {},
   "source": [
    "### Define Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "20f1211a",
   "metadata": {
    "id": "20f1211a"
   },
   "outputs": [],
   "source": [
    "class CNNModel(torch.nn.Module):\n",
    "    def __init__(self):\n",
    "        super(CNNModel, self).__init__()\n",
    "        self.cnn1 = torch.nn.Conv2d(in_channels=1,out_channels=32,kernel_size=9,stride=1,padding=4)\n",
    "        self.maxpool1 = torch.nn.MaxPool2d(kernel_size=2)\n",
    "        self.cnn2 = torch.nn.Conv2d(in_channels=32,out_channels=64,kernel_size=9,stride=1,padding=4)\n",
    "        self.maxpool2 = torch.nn.MaxPool2d(kernel_size=2)\n",
    "        self.fc1 = torch.nn.Linear(64*50*1,5)\n",
    "\n",
    "    def forward(self, x):\n",
    "        conv1_out = torch.nn.functional.relu(self.cnn1(x))\n",
    "        pool1_out = self.maxpool1(conv1_out)\n",
    "        \n",
    "        conv2_out = torch.nn.functional.relu(self.cnn2(pool1_out))\n",
    "        pool2_out = self.maxpool2(conv2_out)\n",
    "        \n",
    "        fc1_input = pool2_out.view(pool2_out.size(0),-1)\n",
    "        out = self.fc1(fc1_input)\n",
    "        return out\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7fc0fc6e",
   "metadata": {
    "id": "7fc0fc6e"
   },
   "source": [
    "### Define Hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "bc9251b9",
   "metadata": {
    "id": "bc9251b9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CNNModel(\n",
       "  (cnn1): Conv2d(1, 32, kernel_size=(9, 9), stride=(1, 1), padding=(4, 4))\n",
       "  (maxpool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  (cnn2): Conv2d(32, 64, kernel_size=(9, 9), stride=(1, 1), padding=(4, 4))\n",
       "  (maxpool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n",
       "  (fc1): Linear(in_features=3200, out_features=5, bias=True)\n",
       ")"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Initialize our neural network model with input and output dimensions\n",
    "model = CNNModel()\n",
    "device = torch.device( \"mps\"if torch.backends.mps.is_available else\"cpu\")\n",
    "#device = torch.device(\"cpu\")\n",
    "# Define the learning rate and epoch \n",
    "learning_rate = 0.001\n",
    "epochs = 50\n",
    "batchsize = 2000#(If you are using mini-batch gradient)\n",
    "\n",
    "# Define loss function and optimizer\n",
    "loss_func = torch.nn.CrossEntropyLoss()\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr = learning_rate,weight_decay=0.01)\n",
    "model.to(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba181b52",
   "metadata": {
    "id": "ba181b52"
   },
   "source": [
    "### Identify Tracked Values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "dca17de1",
   "metadata": {
    "id": "dca17de1"
   },
   "outputs": [],
   "source": [
    "validation_accuracy_list = []\n",
    "train_loss_list = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0d0b025",
   "metadata": {
    "id": "b0d0b025"
   },
   "source": [
    "### Train Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "id": "fd6a820c",
   "metadata": {
    "id": "fd6a820c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Averaged Training Loss for Epoch  0 :  1.4119691729545594\n",
      "Averaged Training Loss for Epoch  1 :  1.1541939616203307\n",
      "Averaged Training Loss for Epoch  2 :  1.0448987185955048\n",
      "Averaged Training Loss for Epoch  3 :  0.9749958515167236\n",
      "Averaged Training Loss for Epoch  4 :  0.9392392992973327\n",
      "Averaged Training Loss for Epoch  5 :  0.9144829630851745\n",
      "Averaged Training Loss for Epoch  6 :  0.8973813772201538\n",
      "Averaged Training Loss for Epoch  7 :  0.8851095914840699\n",
      "Averaged Training Loss for Epoch  8 :  0.876237541437149\n",
      "Averaged Training Loss for Epoch  9 :  0.8684121608734131\n",
      "Averaged Training Loss for Epoch  10 :  0.8605201601982116\n",
      "Averaged Training Loss for Epoch  11 :  0.8557437062263489\n",
      "Averaged Training Loss for Epoch  12 :  0.8510582983493805\n",
      "Averaged Training Loss for Epoch  13 :  0.8462850034236908\n",
      "Averaged Training Loss for Epoch  14 :  0.8423600614070892\n",
      "Averaged Training Loss for Epoch  15 :  0.837714421749115\n",
      "Averaged Training Loss for Epoch  16 :  0.8335491359233856\n",
      "Averaged Training Loss for Epoch  17 :  0.8297886848449707\n",
      "Averaged Training Loss for Epoch  18 :  0.8261475265026093\n",
      "Averaged Training Loss for Epoch  19 :  0.8235106229782104\n",
      "Averaged Training Loss for Epoch  20 :  0.8210442304611206\n",
      "Averaged Training Loss for Epoch  21 :  0.8187468469142913\n",
      "Averaged Training Loss for Epoch  22 :  0.8163223206996918\n",
      "Averaged Training Loss for Epoch  23 :  0.8135760366916657\n",
      "Averaged Training Loss for Epoch  24 :  0.811642849445343\n",
      "Averaged Training Loss for Epoch  25 :  0.8092813432216645\n",
      "Averaged Training Loss for Epoch  26 :  0.8072543323040009\n",
      "Averaged Training Loss for Epoch  27 :  0.8051989376544952\n",
      "Averaged Training Loss for Epoch  28 :  0.8029792666435241\n",
      "Averaged Training Loss for Epoch  29 :  0.8004387140274047\n",
      "Averaged Training Loss for Epoch  30 :  0.7984149992465973\n",
      "Averaged Training Loss for Epoch  31 :  0.7956917226314545\n",
      "Averaged Training Loss for Epoch  32 :  0.7925722002983093\n",
      "Averaged Training Loss for Epoch  33 :  0.7896265089511871\n",
      "Averaged Training Loss for Epoch  34 :  0.7873677849769593\n",
      "Averaged Training Loss for Epoch  35 :  0.7857326567173004\n",
      "Averaged Training Loss for Epoch  36 :  0.7845218360424042\n",
      "Averaged Training Loss for Epoch  37 :  0.7824663996696473\n",
      "Averaged Training Loss for Epoch  38 :  0.7798013687133789\n",
      "Averaged Training Loss for Epoch  39 :  0.7776390612125397\n",
      "Averaged Training Loss for Epoch  40 :  0.7757213115692139\n",
      "Averaged Training Loss for Epoch  41 :  0.7741061210632324\n",
      "Averaged Training Loss for Epoch  42 :  0.7735507786273956\n",
      "Averaged Training Loss for Epoch  43 :  0.771632331609726\n",
      "Averaged Training Loss for Epoch  44 :  0.7700551211833954\n",
      "Averaged Training Loss for Epoch  45 :  0.7686892867088317\n",
      "Averaged Training Loss for Epoch  46 :  0.7673898100852966\n",
      "Averaged Training Loss for Epoch  47 :  0.7666439771652221\n",
      "Averaged Training Loss for Epoch  48 :  0.7677781999111175\n",
      "Averaged Training Loss for Epoch  49 :  0.7637178838253021\n"
     ]
    }
   ],
   "source": [
    "import tqdm # For keeping track of our training progress in percentage\n",
    "\n",
    "# Convert the training, validation, testing dataset (NumPy arrays) into torch tensors\n",
    "\n",
    "train_inputs = torch.from_numpy(x_train).float().to(device)\n",
    "train_targets = torch.from_numpy(y_train).long().to(device) # Convert to 64-bit integer\n",
    "x_batch_train = torch.split(train_inputs, batchsize)\n",
    "y_batch_train = torch.split(train_targets, batchsize)\n",
    "batch_split_num = len(x_batch_train)\n",
    "\n",
    "validation_inputs = torch.from_numpy(x_validation).float().to(device)\n",
    "validation_targets = torch.from_numpy(y_validation).long().to(device)\n",
    "\n",
    "testing_inputs = torch.from_numpy(x_test).float().to(device)\n",
    "testing_targets = torch.from_numpy(y_test).long().to(device)\n",
    "\n",
    "# Training Loop ---------------------------------------------------------------------------------------\n",
    "\n",
    "iteration = 0\n",
    "for epoch in range(epochs):  \n",
    "    for i in range(batch_split_num):\n",
    "        optimizer.zero_grad() # Empty the gradient buffer so each learning event per epoch is separate\n",
    "\n",
    "        outputs = model(x_batch_train[i]) # Forward pass the inputs through the network to produce outputs \n",
    "        \n",
    "        y_batch_indices = torch.argmax(y_batch_train[i], dim=1)\n",
    "\n",
    "        loss = loss_func(outputs, y_batch_indices) # Compute the loss via comparing the output with expected targets\n",
    "    \n",
    "        train_loss_list.append(loss.item()) # Save the loss value to train_loss_list we defined\n",
    "    \n",
    "        loss.backward() # Compute how much changes to be made to weights/biases\n",
    "\n",
    "        optimizer.step() # Update the weights/biases according to learning rate\n",
    "        \n",
    "        outputs = model(validation_inputs)\n",
    "        \n",
    "        _, predicted = torch.max(outputs, 1)# Find the index  with the highest probability for each sample\n",
    "        \n",
    "        correct = (predicted == torch.argmax(validation_targets, dim=1)).sum().item()\n",
    "        \n",
    "        total = validation_targets.size(0)\n",
    "        \n",
    "        accuracy = correct / total\n",
    "        \n",
    "        validation_accuracy_list.append(accuracy)  \n",
    "\n",
    "    print(\"Averaged Training Loss for Epoch \", epoch,\": \", np.mean(train_loss_list[-batch_split_num:]))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "e8dac6ff",
   "metadata": {
    "id": "e8dac6ff"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.73\n"
     ]
    }
   ],
   "source": [
    "with torch.no_grad():\n",
    "    outputs = model(validation_inputs)\n",
    "    _, predicted = torch.max(outputs, 1)# Find the index  with the highest probability for each sample\n",
    "    correct = (predicted == torch.argmax(validation_targets, dim=1)).sum().item()\n",
    "    total = validation_targets.size(0)\n",
    "    accuracy = correct / total\n",
    "    print('Accuracy: {:.2f}'.format(accuracy))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "870d9f1e",
   "metadata": {
    "id": "870d9f1e"
   },
   "source": [
    "### Visualize and Evaluate Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "f79c8356",
   "metadata": {
    "id": "f79c8356"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1200x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Import seaborn for prettier plots\n",
    "\n",
    "import seaborn as sns\n",
    "# Visualize training loss\n",
    "\n",
    "plt.figure(figsize = (12, 6))\n",
    "\n",
    "# Visualize training loss with respect to iterations (1 iteration -> single batch)\n",
    "plt.subplot(2, 1, 1)\n",
    "plt.plot(train_loss_list, linewidth = 3)\n",
    "plt.ylabel(\"training loss\")\n",
    "plt.xlabel(\"iteration\")\n",
    "sns.despine()\n",
    "\n",
    "# Visualize validation accuracy with respect to epochs\n",
    "plt.subplot(2, 1, 2)\n",
    "plt.plot(validation_accuracy_list, linewidth = 3, color = 'gold')\n",
    "plt.ylabel(\"accuracy\")\n",
    "plt.xlabel(\"epoch\")\n",
    "sns.despine()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "e32be1a1",
   "metadata": {
    "id": "e32be1a1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing accuracy: 0.7328\n"
     ]
    }
   ],
   "source": [
    "# Compute the testing accuracy \n",
    "\n",
    "with torch.no_grad():        \n",
    "    outputs = model(testing_inputs)\n",
    "    \n",
    "    correct = (torch.argmax(outputs, dim=1) == torch.argmax(testing_targets,dim=1)).type(torch.FloatTensor).to(device)\n",
    "    \n",
    "    test_accuracy = correct.mean().item()\n",
    "    print('Testing accuracy: {:.4f}'.format(test_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "941669bc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of Gluon: 69.02%\n",
      "Accuracy of Light quark: 66.53%\n",
      "Accuracy of W boson: 78.45%\n",
      "Accuracy of Z boson: 72.31%\n",
      "Accuracy of Top quark: 79.73%\n",
      "The class that the model had the hardest time classifying is: Light quark\n"
     ]
    }
   ],
   "source": [
    "# Define class names for your problem\n",
    "class_names = ['Gluon', 'Light quark', 'W boson', 'Z boson', 'Top quark']\n",
    "\n",
    "# Initialize an empty list to store the accuracy for each class\n",
    "accuracy_1 = []\n",
    "\n",
    "# Calculate the accuracy for each class\n",
    "for i in range(5):\n",
    "    # Find the indices of the samples in the testing set that belong to class i\n",
    "    index = np.where(torch.argmax(testing_targets,dim=1).cpu().numpy() == i)[0]\n",
    "    # Compute the accuracy for class i\n",
    "    class_correct = (torch.argmax(outputs[index], dim=1) == i).type(torch.FloatTensor).to(device)\n",
    "    accuracy_1.append(class_correct.mean().item())\n",
    "    # Print out the accuracy for class i\n",
    "    print('Accuracy of {}: {:.2f}%'.format(class_names[i], accuracy_1[i] * 100))\n",
    "    \n",
    "# Find the class that the model had the hardest time classifying\n",
    "hardest_class = class_names[np.argmin(accuracy_1)]\n",
    "print('The class that the model had the hardest time classifying is: {}'.format(hardest_class))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "84df703b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Proportion of Gluon: 20.23%\n",
      "Proportion of Light quark: 19.92%\n",
      "Proportion of W boson: 20.05%\n",
      "Proportion of Z boson: 19.77%\n",
      "Proportion of Top quark: 20.03%\n"
     ]
    }
   ],
   "source": [
    "for i in range(5):\n",
    "    # Find the indices of the samples in the testing set that belong to class i\n",
    "    class_proportion = (torch.argmax(train_targets, dim=1) == i).type(torch.FloatTensor).to(device).mean().item()\n",
    "    # Print out the proportion for class i\n",
    "    print('Proportion of {}: {:.2f}%'.format(class_names[i], class_proportion * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0a2025c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}