cnn_classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Binary classification with CNN\n",
    "\n",
    "The paths utilized are defiend in the `path_module.py`\n",
    "\n",
    "Before this, the data must be processed and tokenized. \n",
    "1. `data_processing` notebook loads raw data and prepares training,validation and test dataset\n",
    "2. `data_tokenizer`notebook tokenizes the data using LiPCoT model\n",
    "\n",
    "### 1. Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from modules.bert_module import *\n",
    "from modules.path_module import *\n",
    "from modules.utils_module import *\n",
    "import os\n",
    "import joblib\n",
    "import torch\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Load tokenized data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data=joblib.load(os.path.join(tokenized_data_savepath,\"train_data.joblib\"))\n",
    "test_data=joblib.load(os.path.join(tokenized_data_savepath,\"test_data.joblib\"))\n",
    "val_data=joblib.load(os.path.join(tokenized_data_savepath,\"val_data.joblib\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Prepare data for CNN\n",
    "- We extract only the signal data and labels from dataset for CNN\n",
    "- We train using the training set and save the model that gives the best performance on the validation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_sigdata,train_label=prepare_cnn_data(val_data)\n",
    "val_sigdata,val_label=prepare_cnn_data(train_data)\n",
    "test_sigdata,test_label=prepare_cnn_data(test_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. CNN model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Paper: https://link.springer.com/article/10.1007/s00521-018-3689-5\n",
    "\n",
    "Architecture: https://link.springer.com/article/10.1007/s00521-018-3689-5/tables/1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "| Layers | Layer name    | Kernel size | No. of filters | Stride | Output shape | No. of trainable parameters | Regularization  |\n",
    "|--------|---------------|-------------|----------------|--------|--------------|-----------------------------|-----------------|\n",
    "| 0      | Input         | –           | –              | –      | 2500 x 59    | 0                           | –               |\n",
    "| 1      | 1D convolution| 20 x 1      | 5              | 1      | 2481 x 5     | 6000                        | –               |\n",
    "| 2      | Max-pooling   | 2 x 1       | 5              | 2      | 1240 x 5     | 0                           | –               |\n",
    "| 3      | 1D convolution| 10 x 1      | 10             | 1      | 1231 x 10    | 5000                        | –               |\n",
    "| 4      | Max-pooling   | 2 x 1       | 10             | 2      | 615 x 10     | 0                           | –               |\n",
    "| 5      | 1D convolution| 10 x 1      | 10             | 1      | 606 x 10     | 10000                       | –               |\n",
    "| 6      | Max-pooling   | 2 x 1       | 10             | 2      | 303 x 10     | 0                           | –               |\n",
    "| 7      | 1D convolution| 5 x 1       | 15             | 1      | 299 x 15     | 7500                        | –               |\n",
    "| 8      | Max-pooling   | 2 x 1       | 15             | 2      | 149 x 15     | 0                           | –               |\n",
    "| 10     | Dense         | –           | –              | –      | 20           | 44800                       | Dropout (0.5)   |\n",
    "| 12     | Dense         | –           | –              | –      | 10           | 210                         | Dropout (0.5)   |\n",
    "| 13     | Dense         | –           | –              | –      | 2            | 22                          | –               |\n",
    "|        | Total         |             |                |        |              | 68332                       |                 |\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "\n",
    "class CNN_Model(nn.Module):\n",
    "    def __init__(self,chan_total,signal_length):\n",
    "        super(CNN_Model, self).__init__()\n",
    "        self.conv1 = nn.Conv1d(in_channels=chan_total, out_channels=5, kernel_size=20, stride=1,padding=0)\n",
    "        self.pool = nn.AvgPool1d(kernel_size=2)\n",
    "        self.conv2=nn.Conv1d(in_channels=5, out_channels=10, kernel_size=10, stride=1,padding=0)\n",
    "        self.conv3=nn.Conv1d(in_channels=10, out_channels=10, kernel_size=10, stride=1,padding=0)\n",
    "        self.conv4=nn.Conv1d(in_channels=10, out_channels=15, kernel_size=5, stride=1,padding=0)\n",
    "        self.fc1 = nn.Linear(int(149*15), 20)\n",
    "        self.fc2 = nn.Linear(20, 10)\n",
    "        self.fc3 = nn.Linear(10, 2)\n",
    "        self.dropout = nn.Dropout(0.5)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        x = F.relu(self.conv1(x)) # layer 1\n",
    "        x = self.pool(x) # layer 2\n",
    "        x=F.relu(self.conv2(x)) # layer 3\n",
    "        x = self.pool(x) # layer 4\n",
    "        x=F.relu(self.conv3(x)) # layer 5\n",
    "        x = self.pool(x) # layer 6\n",
    "        x=F.relu(self.conv4(x)) # layer 7\n",
    "        x = self.pool(x) # layer 8\n",
    "        x = x.view(-1, int(149*15)) \n",
    "        x = self.dropout(x) # layer 9\n",
    "        x=self.fc1(x) # layer 10\n",
    "        x = self.dropout(x) # layer 11\n",
    "        x=self.fc2(x) # layer 12\n",
    "        x=self.fc3(x) # layer 13\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Initialize CNN model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running on: cpu\n",
      "Model created\n"
     ]
    }
   ],
   "source": [
    "import torch.optim as optim\n",
    "from torch import cuda\n",
    "\n",
    "# Set the random seed for reproducibility\n",
    "random_seed = 42\n",
    "random.seed(random_seed)\n",
    "torch.manual_seed(random_seed)\n",
    "torch.cuda.manual_seed(random_seed)\n",
    "np.random.seed(random_seed)\n",
    "\n",
    "# use GPU\n",
    "device='cpu'#'cuda' if cuda.is_available() else 'cpu'\n",
    "print(\"Running on: \"+device)\n",
    "cuda.empty_cache()\n",
    "\n",
    "# initialize model\n",
    "n_chan=train_data[0].data.shape[1]\n",
    "n_datalength=train_data[0].data.shape[0]\n",
    "model = CNN_Model(n_chan,n_datalength)\n",
    "model.to(device)\n",
    "print('Model created')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6. Configure Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size = 16#1 # could not find\n",
    "#num_epochs = 64\n",
    "num_epochs = 70 # from paper\n",
    "lrate=1e-4 #from paper\n",
    "patience=80 # no early stopping from paper\n",
    "# Define loss function and optimizer\n",
    "criterion = nn.BCELoss()\n",
    "optimizer = optim.Adam(model.parameters(), lr=lrate)\n",
    "cnn_modelfullpath=os.path.join(cnn_modelpath,'cnn.pt')\n",
    "\n",
    "training_arg={\n",
    "    'modelfullpath': cnn_modelfullpath,\n",
    "    'criterion': criterion,\n",
    "    'optimizer': optimizer,\n",
    "    'epochs': num_epochs,\n",
    "    'patience': patience,\n",
    "    'seed': random_seed\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7. Prepare data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "from torch.utils.data import DataLoader, TensorDataset\n",
    "\n",
    "train_dataset = TensorDataset(torch.stack(train_sigdata), torch.stack(train_label))\n",
    "val_dataset = TensorDataset(torch.stack(val_sigdata), torch.stack(val_label))\n",
    "test_dataset = TensorDataset(torch.stack(test_sigdata), torch.stack(test_label))\n",
    "\n",
    "train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
    "val_loader = DataLoader(val_dataset, batch_size=1, shuffle=False)\n",
    "test_loader = DataLoader(test_dataset, batch_size=1, shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 8. Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 1/70:   0%|          | 0/9 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 1/70: 100%|██████████| 9/9 [00:00<00:00, 33.02it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 47.8261\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 2/70: 100%|██████████| 9/9 [00:00<00:00, 36.88it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 48.7578\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 3/70: 100%|██████████| 9/9 [00:00<00:00, 34.08it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 51.8634\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 4/70: 100%|██████████| 9/9 [00:00<00:00, 34.80it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 54.6584\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 5/70: 100%|██████████| 9/9 [00:00<00:00, 35.34it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 57.4534\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 6/70: 100%|██████████| 9/9 [00:00<00:00, 36.19it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 58.3851\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 7/70: 100%|██████████| 9/9 [00:00<00:00, 34.94it/s]\n",
      "Epoch 8/70: 100%|██████████| 9/9 [00:00<00:00, 30.25it/s]\n",
      "Epoch 9/70: 100%|██████████| 9/9 [00:00<00:00, 34.81it/s]\n",
      "Epoch 10/70: 100%|██████████| 9/9 [00:00<00:00, 37.24it/s]\n",
      "Epoch 11/70: 100%|██████████| 9/9 [00:00<00:00, 34.73it/s]\n",
      "Epoch 12/70: 100%|██████████| 9/9 [00:00<00:00, 40.53it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 59.0062\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 13/70: 100%|██████████| 9/9 [00:00<00:00, 35.99it/s]\n",
      "Epoch 14/70: 100%|██████████| 9/9 [00:00<00:00, 34.01it/s]\n",
      "Epoch 15/70: 100%|██████████| 9/9 [00:00<00:00, 39.46it/s]\n",
      "Epoch 16/70: 100%|██████████| 9/9 [00:00<00:00, 36.88it/s]\n",
      "Epoch 17/70: 100%|██████████| 9/9 [00:00<00:00, 43.25it/s]\n",
      "Epoch 18/70: 100%|██████████| 9/9 [00:00<00:00, 39.38it/s]\n",
      "Epoch 19/70: 100%|██████████| 9/9 [00:00<00:00, 37.97it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 60.5590\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 20/70: 100%|██████████| 9/9 [00:00<00:00, 41.36it/s]\n",
      "Epoch 21/70: 100%|██████████| 9/9 [00:00<00:00, 41.85it/s]\n",
      "Epoch 22/70: 100%|██████████| 9/9 [00:00<00:00, 43.88it/s]\n",
      "Epoch 23/70: 100%|██████████| 9/9 [00:00<00:00, 47.86it/s]\n",
      "Epoch 24/70: 100%|██████████| 9/9 [00:00<00:00, 43.25it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 62.7329\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 25/70: 100%|██████████| 9/9 [00:00<00:00, 46.14it/s]\n",
      "Epoch 26/70: 100%|██████████| 9/9 [00:00<00:00, 42.84it/s]\n",
      "Epoch 27/70: 100%|██████████| 9/9 [00:00<00:00, 45.21it/s]\n",
      "Epoch 28/70: 100%|██████████| 9/9 [00:00<00:00, 44.61it/s]\n",
      "Epoch 29/70: 100%|██████████| 9/9 [00:00<00:00, 46.38it/s]\n",
      "Epoch 30/70: 100%|██████████| 9/9 [00:00<00:00, 45.89it/s]\n",
      "Epoch 31/70: 100%|██████████| 9/9 [00:00<00:00, 44.11it/s]\n",
      "Epoch 32/70: 100%|██████████| 9/9 [00:00<00:00, 46.62it/s]\n",
      "Epoch 33/70: 100%|██████████| 9/9 [00:00<00:00, 46.86it/s]\n",
      "Epoch 34/70: 100%|██████████| 9/9 [00:00<00:00, 44.11it/s]\n",
      "Epoch 35/70: 100%|██████████| 9/9 [00:00<00:00, 47.10it/s]\n",
      "Epoch 36/70: 100%|██████████| 9/9 [00:00<00:00, 38.28it/s]\n",
      "Epoch 37/70: 100%|██████████| 9/9 [00:00<00:00, 44.54it/s]\n",
      "Epoch 38/70: 100%|██████████| 9/9 [00:00<00:00, 43.26it/s]\n",
      "Epoch 39/70: 100%|██████████| 9/9 [00:00<00:00, 40.71it/s]\n",
      "Epoch 40/70: 100%|██████████| 9/9 [00:00<00:00, 42.24it/s]\n",
      "Epoch 41/70: 100%|██████████| 9/9 [00:00<00:00, 43.14it/s]\n",
      "Epoch 42/70: 100%|██████████| 9/9 [00:00<00:00, 42.44it/s]\n",
      "Epoch 43/70: 100%|██████████| 9/9 [00:00<00:00, 39.29it/s]\n",
      "Epoch 44/70: 100%|██████████| 9/9 [00:00<00:00, 36.13it/s]\n",
      "Epoch 45/70: 100%|██████████| 9/9 [00:00<00:00, 38.45it/s]\n",
      "Epoch 46/70: 100%|██████████| 9/9 [00:00<00:00, 41.45it/s]\n",
      "Epoch 47/70: 100%|██████████| 9/9 [00:00<00:00, 44.49it/s]\n",
      "Epoch 48/70: 100%|██████████| 9/9 [00:00<00:00, 47.85it/s]\n",
      "Epoch 49/70: 100%|██████████| 9/9 [00:00<00:00, 45.90it/s]\n",
      "Epoch 50/70: 100%|██████████| 9/9 [00:00<00:00, 44.76it/s]\n",
      "Epoch 51/70: 100%|██████████| 9/9 [00:00<00:00, 42.54it/s]\n",
      "Epoch 52/70: 100%|██████████| 9/9 [00:00<00:00, 41.65it/s]\n",
      "Epoch 53/70: 100%|██████████| 9/9 [00:00<00:00, 44.98it/s]\n",
      "Epoch 54/70: 100%|██████████| 9/9 [00:00<00:00, 46.36it/s]\n",
      "Epoch 55/70: 100%|██████████| 9/9 [00:00<00:00, 41.28it/s]\n",
      "Epoch 56/70: 100%|██████████| 9/9 [00:00<00:00, 46.38it/s]\n",
      "Epoch 57/70: 100%|██████████| 9/9 [00:00<00:00, 45.16it/s]\n",
      "Epoch 58/70: 100%|██████████| 9/9 [00:00<00:00, 46.14it/s]\n",
      "Epoch 59/70: 100%|██████████| 9/9 [00:00<00:00, 43.25it/s]\n",
      "Epoch 60/70: 100%|██████████| 9/9 [00:00<00:00, 43.26it/s]\n",
      "Epoch 61/70: 100%|██████████| 9/9 [00:00<00:00, 45.89it/s]\n",
      "Epoch 62/70: 100%|██████████| 9/9 [00:00<00:00, 43.68it/s]\n",
      "Epoch 63/70: 100%|██████████| 9/9 [00:00<00:00, 45.21it/s]\n",
      "Epoch 64/70: 100%|██████████| 9/9 [00:00<00:00, 45.22it/s]\n",
      "Epoch 65/70: 100%|██████████| 9/9 [00:00<00:00, 45.67it/s]\n",
      "Epoch 66/70: 100%|██████████| 9/9 [00:00<00:00, 45.45it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model state saved with validation accuracy: 63.3540\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 67/70: 100%|██████████| 9/9 [00:00<00:00, 45.21it/s]\n",
      "Epoch 68/70: 100%|██████████| 9/9 [00:00<00:00, 45.89it/s]\n",
      "Epoch 69/70: 100%|██████████| 9/9 [00:00<00:00, 44.99it/s]\n",
      "Epoch 70/70: 100%|██████████| 9/9 [00:00<00:00, 43.88it/s]\n"
     ]
    }
   ],
   "source": [
    "results= model_trainer(model,train_loader,val_loader,training_arg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 9. Inspect training process"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(results['epoch_loss'])\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('Avg loss')\n",
    "plt.show()\n",
    "\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(results['training_acc'],label='Training data')\n",
    "plt.plot(results['validation_acc'],label='Validation data')\n",
    "plt.legend()\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 10. Evaluate model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fine tune model loaded\n",
      "Train Binary Accuracy: 100.0000\n",
      "Confusion Matrix:\n",
      "[[72  0]\n",
      " [ 0 66]]\n",
      "Accuracy: 100.0\n",
      "Precision: 100.0\n",
      "Sensitivity (Recall): 100.0\n",
      "Specificity: 100.0\n",
      "F1 Score: 1.0\n",
      "AUC Score: 1.0\n",
      "\n",
      "\n",
      "Validation Binary Accuracy: 63.3540\n",
      "Confusion Matrix:\n",
      "[[101  67]\n",
      " [ 51 103]]\n",
      "Accuracy: 63.35403726708074\n",
      "Precision: 60.588235294117645\n",
      "Sensitivity (Recall): 66.88311688311688\n",
      "Specificity: 60.11904761904761\n",
      "F1 Score: 0.6358024691358024\n",
      "AUC Score: 0.6413497217068647\n",
      "\n",
      "\n",
      "Test Binary Accuracy: 51.0870\n",
      "Confusion Matrix:\n",
      "[[26 22]\n",
      " [23 21]]\n",
      "Accuracy: 51.08695652173913\n",
      "Precision: 48.837209302325576\n",
      "Sensitivity (Recall): 47.72727272727273\n",
      "Specificity: 54.166666666666664\n",
      "F1 Score: 0.4827586206896552\n",
      "AUC Score: 0.5392992424242424\n",
      "Completed\n"
     ]
    }
   ],
   "source": [
    "best_model_state=torch.load(cnn_modelfullpath, map_location=device)\n",
    "model.load_state_dict(best_model_state,strict=True)\n",
    "model.eval()\n",
    "print('Fine tune model loaded')\n",
    "\n",
    "tr_accuracy, tr_est,tr_labels,tr_scores= evaluate_binary_accuracy(model, train_loader)\n",
    "print(f'Train Binary Accuracy: {tr_accuracy:.4f}')\n",
    "perf_metrics(tr_labels, tr_est,tr_scores)\n",
    "print('\\n')\n",
    "v_accuracy,val_est,val_labels, val_scores = evaluate_binary_accuracy(model, val_loader)\n",
    "print(f'Validation Binary Accuracy: {v_accuracy:.4f}')\n",
    "perf_metrics(val_labels, val_est,val_scores)\n",
    "print('\\n')\n",
    "test_accuracy,test_est,test_labels, test_scores = evaluate_binary_accuracy(model, test_loader)\n",
    "print(f'Test Binary Accuracy: {test_accuracy:.4f}')\n",
    "perf_metrics(test_labels, test_est,test_scores)\n",
    "\n",
    "print('Completed')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mybertV2",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}