eval_cnn2d.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import os\n",
    "import json\n",
    "\n",
    "from tensorflow.keras.models import load_model\n",
    "from data_utils import prepare_data, normalize_min_max_v2\n",
    "from sklearn.metrics import precision_score, recall_score, f1_score, confusion_matrix\n",
    "\n",
    "def f1_metric(val_1, val2):\n",
    "    if val_1 + val2 == 0:\n",
    "        return 0  # To avoid division by zero\n",
    "    return 2 * (val_1 * val2) / (val_1 + val2)\n",
    "\n",
    "PATH = 'results/cnn2d_models'\n",
    "\n",
    "spectogram_map = {\n",
    "    256: [8, 64, 128, 250],\n",
    "    512: [0, 256, 511],\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation (Activation)      (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 57.499998807907104\n",
      "retest_acc 57.499998807907104\n",
      "256_8 Test-Retest Model Metrics:\n",
      "Accuracy: 57.499998807907104\n",
      "Precision: 0.4958333333333333\n",
      "Recall: 0.575\n",
      "F1 Score: 0.5266666666666666\n",
      "\n",
      "256_8 Retest-Test Model Metrics:\n",
      "Accuracy: 57.499998807907104\n",
      "Precision: 0.5883333333333333\n",
      "Recall: 0.575\n",
      "F1 Score: 0.5495238095238095\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_4 (Conv2D)            (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_6 (Activation)    (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_7 (Activation)    (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2 (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_8 (Activation)    (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 50.0\n",
      "retest_acc 57.499998807907104\n",
      "256_64 Test-Retest Model Metrics:\n",
      "Accuracy: 50.0\n",
      "Precision: 0.41380952380952374\n",
      "Recall: 0.5\n",
      "F1 Score: 0.4249206349206349\n",
      "\n",
      "256_64 Retest-Test Model Metrics:\n",
      "Accuracy: 57.499998807907104\n",
      "Precision: 0.5270833333333333\n",
      "Recall: 0.575\n",
      "F1 Score: 0.5166666666666667\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_4\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_8 (Conv2D)            (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_12 (Activation)   (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_8 (MaxPooling2 (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_9 (Conv2D)            (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_13 (Activation)   (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_9 (MaxPooling2 (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_4 (Flatten)          (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_14 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 69.9999988079071\n",
      "retest_acc 67.5000011920929\n",
      "256_128 Test-Retest Model Metrics:\n",
      "Accuracy: 69.9999988079071\n",
      "Precision: 0.75\n",
      "Recall: 0.7\n",
      "F1 Score: 0.695\n",
      "\n",
      "256_128 Retest-Test Model Metrics:\n",
      "Accuracy: 67.5000011920929\n",
      "Precision: 0.6283333333333333\n",
      "Recall: 0.675\n",
      "F1 Score: 0.6169047619047618\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_6\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_12 (Conv2D)           (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_18 (Activation)   (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_12 (MaxPooling (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_13 (Conv2D)           (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_19 (Activation)   (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_13 (MaxPooling (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_6 (Flatten)          (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_20 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 62.5\n",
      "retest_acc 69.9999988079071\n",
      "256_250 Test-Retest Model Metrics:\n",
      "Accuracy: 62.5\n",
      "Precision: 0.5791666666666666\n",
      "Recall: 0.625\n",
      "F1 Score: 0.5816666666666667\n",
      "\n",
      "256_250 Retest-Test Model Metrics:\n",
      "Accuracy: 69.9999988079071\n",
      "Precision: 0.7891666666666667\n",
      "Recall: 0.7\n",
      "F1 Score: 0.6976190476190476\n",
      "[57.499998807907104, 50.0, 69.9999988079071, 62.5]\n",
      "[57.499998807907104, 57.499998807907104, 67.5000011920929, 69.9999988079071]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_8\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_16 (Conv2D)           (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_24 (Activation)   (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_16 (MaxPooling (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_17 (Conv2D)           (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_25 (Activation)   (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_17 (MaxPooling (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_8 (Flatten)          (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_16 (Dense)             (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_26 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_17 (Dense)             (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 64.99999761581421\n",
      "retest_acc 77.49999761581421\n",
      "512_0 Test-Retest Model Metrics:\n",
      "Accuracy: 64.99999761581421\n",
      "Precision: 0.7083333333333334\n",
      "Recall: 0.65\n",
      "F1 Score: 0.6366666666666665\n",
      "\n",
      "512_0 Retest-Test Model Metrics:\n",
      "Accuracy: 77.49999761581421\n",
      "Precision: 0.775\n",
      "Recall: 0.775\n",
      "F1 Score: 0.7449999999999999\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_10\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_20 (Conv2D)           (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_30 (Activation)   (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_20 (MaxPooling (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_21 (Conv2D)           (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_31 (Activation)   (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_21 (MaxPooling (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_10 (Flatten)         (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_20 (Dense)             (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_32 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_10 (Dropout)         (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 77.49999761581421\n",
      "retest_acc 69.9999988079071\n",
      "512_256 Test-Retest Model Metrics:\n",
      "Accuracy: 77.49999761581421\n",
      "Precision: 0.75\n",
      "Recall: 0.775\n",
      "F1 Score: 0.7416666666666666\n",
      "\n",
      "512_256 Retest-Test Model Metrics:\n",
      "Accuracy: 69.9999988079071\n",
      "Precision: 0.6783333333333333\n",
      "Recall: 0.7\n",
      "F1 Score: 0.6552380952380952\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_12\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_24 (Conv2D)           (None, 32, 32, 20)        520       \n",
      "_________________________________________________________________\n",
      "activation_36 (Activation)   (None, 32, 32, 20)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_24 (MaxPooling (None, 16, 16, 20)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_25 (Conv2D)           (None, 16, 16, 40)        20040     \n",
      "_________________________________________________________________\n",
      "activation_37 (Activation)   (None, 16, 16, 40)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_25 (MaxPooling (None, 8, 8, 40)          0         \n",
      "_________________________________________________________________\n",
      "flatten_12 (Flatten)         (None, 2560)              0         \n",
      "_________________________________________________________________\n",
      "dense_24 (Dense)             (None, 512)               1311232   \n",
      "_________________________________________________________________\n",
      "activation_38 (Activation)   (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dropout_12 (Dropout)         (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_25 (Dense)             (None, 20)                10260     \n",
      "=================================================================\n",
      "Total params: 1,342,052\n",
      "Trainable params: 1,342,052\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "test_acc 44.999998807907104\n",
      "retest_acc 40.00000059604645\n",
      "512_511 Test-Retest Model Metrics:\n",
      "Accuracy: 44.999998807907104\n",
      "Precision: 0.3392857142857143\n",
      "Recall: 0.45\n",
      "F1 Score: 0.36484126984126986\n",
      "\n",
      "512_511 Retest-Test Model Metrics:\n",
      "Accuracy: 40.00000059604645\n",
      "Precision: 0.42547619047619045\n",
      "Recall: 0.4\n",
      "F1 Score: 0.36444444444444446\n",
      "[64.99999761581421, 77.49999761581421, 44.999998807907104]\n",
      "[77.49999761581421, 69.9999988079071, 40.00000059604645]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "/home/user/yahya/other/listener_cls_ffr/venv/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "ename": "",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31mThe Kernel crashed while executing code in the the current cell or a previous cell. Please review the code in the cell(s) to identify a possible cause of the failure. Click <a href='https://aka.ms/vscodeJupyterKernelCrash'>here</a> for more info. View Jupyter <a href='command:jupyter.viewOutput'>log</a> for further details."
     ]
    }
   ],
   "source": [
    "\n",
    "for window_size, overlap_list in spectogram_map.items():\n",
    "    tr_metrics = []\n",
    "    rt_metrics = []\n",
    "    tr_names = []\n",
    "    for overlap in overlap_list:\n",
    "        tr_exp = f'cnn2d_{window_size}_{overlap}_test_retest'\n",
    "        rt_exp = f'cnn2d_{window_size}_{overlap}_retest_test'\n",
    "        # Specify the path to the saved h5 model file\n",
    "        test_retest_path = f'{PATH}/{tr_exp}.h5'\n",
    "        retest_test_path = f'{PATH}/{rt_exp}.h5'\n",
    "\n",
    "        # Load the model\n",
    "        test_retest_model = load_model(test_retest_path)\n",
    "        retest_test_model = load_model(retest_test_path)\n",
    "        \n",
    "        print(test_retest_model.summary())\n",
    "\n",
    "        NPY_DATA_DIR = \"npy_datasets\"\n",
    "\n",
    "        test_dataset = np.load(\n",
    "            f\"{NPY_DATA_DIR}/test_{window_size}_{overlap}_png.npy\", allow_pickle=True\n",
    "        )\n",
    "        retest_dataset = np.load(\n",
    "            f\"{NPY_DATA_DIR}/retest_{window_size}_{overlap}_png.npy\", allow_pickle=True\n",
    "        )\n",
    "\n",
    "        X_test, y_test = prepare_data(test_dataset)\n",
    "        X_retest, y_retest = prepare_data(retest_dataset)\n",
    "\n",
    "        X_test = normalize_min_max_v2(X_test, 0, 1)\n",
    "        X_retest = normalize_min_max_v2(X_retest, 0, 1)\n",
    "\n",
    "        # reshape only when grayscale\n",
    "        X_test = X_test.reshape(X_test.shape[0], X_test.shape[1], X_test.shape[2], 1)\n",
    "        X_retest = X_retest.reshape(X_retest.shape[0], X_retest.shape[1], X_retest.shape[2], 1)\n",
    "\n",
    "        _, test_acc = test_retest_model.evaluate(X_retest, y_retest, verbose=0)\n",
    "        _, retest_acc = retest_test_model.evaluate(X_test, y_test, verbose=0)\n",
    "        \n",
    "        test_acc = test_acc * 100\n",
    "        retest_acc = retest_acc * 100\n",
    "\n",
    "        tr_metrics.append(test_acc)\n",
    "        rt_metrics.append(retest_acc)\n",
    "        tr_names.append(f'{overlap}')\n",
    "\n",
    "        print('test_acc', test_acc)\n",
    "        print('retest_acc', retest_acc)\n",
    "\n",
    "        test_pred = test_retest_model.predict(X_retest)\n",
    "        test_pred_classes = np.argmax(test_pred, axis=1)\n",
    "        y_test = np.argmax(y_test, axis=1)\n",
    "        test_confusion_mtx = confusion_matrix(y_test, test_pred_classes)\n",
    "        \n",
    "        # Precision, recall, and F1 score for test_retest_model\n",
    "        precision_test = precision_score(y_test, test_pred_classes, average='weighted')\n",
    "        recall_test = recall_score(y_test, test_pred_classes, average='weighted')\n",
    "        f1_score_test = f1_score(y_test, test_pred_classes, average='weighted')\n",
    "\n",
    "        print(f\"{window_size}_{overlap} Test-Retest Model Metrics:\")\n",
    "        print(f\"Accuracy: {test_acc}\")\n",
    "        print(f\"Precision: {precision_test}\")\n",
    "        print(f\"Recall: {recall_test}\")\n",
    "        print(f\"F1 Score: {f1_score_test}\")\n",
    "\n",
    "        retest_pred = retest_test_model.predict(X_test)\n",
    "        retest_pred_classes = np.argmax(retest_pred, axis=1)\n",
    "        y_retest = np.argmax(y_retest, axis=1)\n",
    "        retest_confusion_mtx = confusion_matrix(y_retest, retest_pred_classes)\n",
    "        \n",
    "        # Precision, recall, and F1 score for retest_test_model\n",
    "        precision_retest = precision_score(y_retest, retest_pred_classes, average='weighted')\n",
    "        recall_retest = recall_score(y_retest, retest_pred_classes, average='weighted')\n",
    "        f1_score_retest = f1_score(y_retest, retest_pred_classes, average='weighted')\n",
    "\n",
    "        print(f\"\\n{window_size}_{overlap} Retest-Test Model Metrics:\")\n",
    "        print(f\"Accuracy: {retest_acc}\")\n",
    "        print(f\"Precision: {precision_retest}\")\n",
    "        print(f\"Recall: {recall_retest}\")\n",
    "        print(f\"F1 Score: {f1_score_retest}\")\n",
    "\n",
    "        # sns.heatmap(test_confusion_mtx, annot=True, fmt='d')\n",
    "        # plt.title(f'{window_size}_{overlap} - Test/Retest - {test_acc:.3f}%')\n",
    "        # plt.xlabel('Predicted')\n",
    "        # plt.ylabel('True')\n",
    "        # plt.savefig(f'{PATH}/plots/cnn2d_{window_size}_{overlap}_test_retest.png')\n",
    "        # plt.show()\n",
    "\n",
    "        # sns.heatmap(test_confusion_mtx, annot=True, fmt='d')\n",
    "        # plt.title(f'{window_size}_{overlap} - Retest/Test - {retest_acc:.3f}%')\n",
    "        # plt.xlabel('Predicted')\n",
    "        # plt.ylabel('True')\n",
    "        # plt.savefig(f'{PATH}/plots/cnn2d_{window_size}_{overlap}_retest_test.png')\n",
    "        # plt.show()\n",
    "        \n",
    "    combined_f1 = []\n",
    "    for i in range(len(tr_metrics)):\n",
    "        combined_f1.append(f1_metric(tr_metrics[i], rt_metrics[i]))\n",
    "    \n",
    "    print(tr_metrics)\n",
    "    print(rt_metrics)\n",
    "\n",
    "    n = len(tr_metrics)  # Number of data pairs\n",
    "    ind = np.arange(n)  # The x locations for the groups\n",
    "    width = 0.25        # Width of the bars\n",
    "\n",
    "    fig, ax = plt.subplots(figsize=(10, 5))\n",
    "    x_bars = ax.bar(ind - width, tr_metrics, width, label='Test/Retest', color='blue')\n",
    "    y_bars = ax.bar(ind, rt_metrics, width, label='Retest/Test', color='orange')\n",
    "    z_bars = ax.bar(ind + width, combined_f1, width, label='Combined-F1', color='green')\n",
    "    \n",
    "    for bar in x_bars + y_bars + z_bars:\n",
    "        height = bar.get_height()\n",
    "        ax.annotate(f'{height:.0f}',  # Adjust format as needed\n",
    "                    xy=(bar.get_x() + bar.get_width() / 2, height),\n",
    "                    xytext=(0, 3),  # Offset\n",
    "                    textcoords='offset points',\n",
    "                    ha='center', va='bottom')\n",
    "\n",
    "    ax.set_xlabel('overlap')\n",
    "    ax.set_ylabel('accuracy (%)')\n",
    "    ax.set_title(f'2D CNN Accuracy - window_size={window_size}')\n",
    "    ax.set_xticks(ind)\n",
    "    ax.set_xticklabels(tr_names)\n",
    "    ax.set_ylim([0, 100])\n",
    "    ax.legend(fontsize='small', loc='lower right')\n",
    "\n",
    "    plt.savefig(f'{PATH}/plots/cnn2d_{window_size}_accuracy.png')\n",
    "    plt.show()\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "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": "3.7.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}