08_other_tips.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "3195be58",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-09-04T19:11:08.965057Z",
     "start_time": "2022-09-04T19:11:06.995541Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import keras\n",
    "import tensorflow as tf\n",
    "\n",
    "def reset_seed():\n",
    "    tf.random.set_seed(0)\n",
    "    np.random.seed(0)\n",
    "\n",
    "from sklearn import metrics\n",
    "\n",
    "def classification_results(y, yhat):\n",
    "    acc = metrics.accuracy_score(y, yhat)\n",
    "    mcc = metrics.matthews_corrcoef(y, yhat)\n",
    "    f1_weighted = metrics.f1_score(y, yhat, average=\"weighted\")\n",
    "    return acc, mcc, f1_weighted\n",
    "\n",
    "\n",
    "reset_seed()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15015028",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-09-04T13:07:52.163408Z",
     "start_time": "2022-09-04T13:07:52.159554Z"
    }
   },
   "source": [
    "# TIP 1 - General Tip: Use the datetime in your favor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d12f86a5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-09-04T19:11:09.474798Z",
     "start_time": "2022-09-04T19:11:08.966535Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pid</th>\n",
       "      <th>time</th>\n",
       "      <th>id</th>\n",
       "      <th>sort</th>\n",
       "      <th>sleep_phase</th>\n",
       "      <th>act_0</th>\n",
       "      <th>act_1</th>\n",
       "      <th>act_2</th>\n",
       "      <th>act_3</th>\n",
       "      <th>act_4</th>\n",
       "      <th>...</th>\n",
       "      <th>hr_1</th>\n",
       "      <th>hr_2</th>\n",
       "      <th>hr_3</th>\n",
       "      <th>hr_4</th>\n",
       "      <th>hr_5</th>\n",
       "      <th>hr_6</th>\n",
       "      <th>hr_7</th>\n",
       "      <th>hr_8</th>\n",
       "      <th>hr_9</th>\n",
       "      <th>hr_10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>29</td>\n",
       "      <td>(0, 0)</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>59</td>\n",
       "      <td>(0, 1)</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>76.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>89</td>\n",
       "      <td>(0, 2)</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>78.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>119</td>\n",
       "      <td>(0, 3)</td>\n",
       "      <td>3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>78.0</td>\n",
       "      <td>73.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>149</td>\n",
       "      <td>(0, 4)</td>\n",
       "      <td>4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>71.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>78.0</td>\n",
       "      <td>73.0</td>\n",
       "      <td>80.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pid  time      id  sort  sleep_phase  act_0  act_1  act_2  act_3  act_4  \\\n",
       "0    0    29  (0, 0)     0          0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "1    0    59  (0, 1)     1          0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "2    0    89  (0, 2)     2          0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "3    0   119  (0, 3)     3          0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "4    0   149  (0, 4)     4          0.0    0.0    0.0    0.0    0.0    0.0   \n",
       "\n",
       "   ...  hr_1  hr_2  hr_3  hr_4  hr_5  hr_6  hr_7  hr_8  hr_9  hr_10  \n",
       "0  ...   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0   71.0  \n",
       "1  ...   0.0   0.0   0.0   0.0   0.0   0.0   0.0   0.0  71.0   76.0  \n",
       "2  ...   0.0   0.0   0.0   0.0   0.0   0.0   0.0  71.0  76.0   78.0  \n",
       "3  ...   0.0   0.0   0.0   0.0   0.0   0.0  71.0  76.0  78.0   73.0  \n",
       "4  ...   0.0   0.0   0.0   0.0   0.0  71.0  76.0  78.0  73.0   80.0  \n",
       "\n",
       "[5 rows x 27 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(\"./datasets/df_raw_features.tar.gz\")\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ece473a3",
   "metadata": {},
   "source": [
    "Thanks to: <b> https://www.tensorflow.org/tutorials/structured_data/time_series </b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b0e95bb4",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 100 entries, 0 to 99\n",
      "Data columns (total 2 columns):\n",
      " #   Column   Non-Null Count  Dtype         \n",
      "---  ------   --------------  -----         \n",
      " 0   Time     100 non-null    datetime64[ns]\n",
      " 1   Numbers  100 non-null    int64         \n",
      "dtypes: datetime64[ns](1), int64(1)\n",
      "memory usage: 1.7 KB\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/47/3qykw18x31n87lxnd4kh3rtm0000gn/T/ipykernel_83526/1330924337.py:4: DeprecationWarning: This function is deprecated. Please call randint(0, 100 + 1) instead\n",
      "  numbers = np.random.random_integers(low=0, high=100, size=100)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Time</th>\n",
       "      <th>Numbers</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2022-09-16 00:00:00</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2022-09-16 01:00:00</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2022-09-16 02:00:00</td>\n",
       "      <td>64</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2022-09-16 03:00:00</td>\n",
       "      <td>67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2022-09-16 04:00:00</td>\n",
       "      <td>67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>2022-09-19 23:00:00</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>2022-09-20 00:00:00</td>\n",
       "      <td>23</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>2022-09-20 01:00:00</td>\n",
       "      <td>79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>2022-09-20 02:00:00</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>2022-09-20 03:00:00</td>\n",
       "      <td>85</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Time  Numbers\n",
       "0  2022-09-16 00:00:00       44\n",
       "1  2022-09-16 01:00:00       47\n",
       "2  2022-09-16 02:00:00       64\n",
       "3  2022-09-16 03:00:00       67\n",
       "4  2022-09-16 04:00:00       67\n",
       "..                 ...      ...\n",
       "95 2022-09-19 23:00:00       58\n",
       "96 2022-09-20 00:00:00       23\n",
       "97 2022-09-20 01:00:00       79\n",
       "98 2022-09-20 02:00:00       13\n",
       "99 2022-09-20 03:00:00       85\n",
       "\n",
       "[100 rows x 2 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nrows = 100\n",
    "\n",
    "time = pd.date_range(\"2022-09-16\", periods=nrows, freq=\"H\")\n",
    "numbers = np.random.random_integers(low=0, high=100, size=100)\n",
    "\n",
    "df = pd.DataFrame({\"Time\": time, \"Numbers\": numbers})\n",
    "df.info()\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "4a28c6b1",
   "metadata": {},
   "outputs": [],
   "source": [
    "day = 24*60*60\n",
    "timestamp_s = time.map(pd.Timestamp.timestamp)\n",
    "\n",
    "df['Day sin'] = np.sin(timestamp_s * (2 * np.pi / day))\n",
    "df['Day cos'] = np.cos(timestamp_s * (2 * np.pi / day))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b67e0c15",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Time</th>\n",
       "      <th>Numbers</th>\n",
       "      <th>Day sin</th>\n",
       "      <th>Day cos</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2022-09-16 00:00:00</td>\n",
       "      <td>44</td>\n",
       "      <td>-5.574892e-12</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2022-09-16 01:00:00</td>\n",
       "      <td>47</td>\n",
       "      <td>2.588190e-01</td>\n",
       "      <td>0.965926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2022-09-16 02:00:00</td>\n",
       "      <td>64</td>\n",
       "      <td>5.000000e-01</td>\n",
       "      <td>0.866025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2022-09-16 03:00:00</td>\n",
       "      <td>67</td>\n",
       "      <td>7.071068e-01</td>\n",
       "      <td>0.707107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2022-09-16 04:00:00</td>\n",
       "      <td>67</td>\n",
       "      <td>8.660254e-01</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>2022-09-19 23:00:00</td>\n",
       "      <td>58</td>\n",
       "      <td>-2.588190e-01</td>\n",
       "      <td>0.965926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>2022-09-20 00:00:00</td>\n",
       "      <td>23</td>\n",
       "      <td>-2.932653e-12</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>2022-09-20 01:00:00</td>\n",
       "      <td>79</td>\n",
       "      <td>2.588190e-01</td>\n",
       "      <td>0.965926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>2022-09-20 02:00:00</td>\n",
       "      <td>13</td>\n",
       "      <td>5.000000e-01</td>\n",
       "      <td>0.866025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99</th>\n",
       "      <td>2022-09-20 03:00:00</td>\n",
       "      <td>85</td>\n",
       "      <td>7.071068e-01</td>\n",
       "      <td>0.707107</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>100 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Time  Numbers       Day sin   Day cos\n",
       "0  2022-09-16 00:00:00       44 -5.574892e-12  1.000000\n",
       "1  2022-09-16 01:00:00       47  2.588190e-01  0.965926\n",
       "2  2022-09-16 02:00:00       64  5.000000e-01  0.866025\n",
       "3  2022-09-16 03:00:00       67  7.071068e-01  0.707107\n",
       "4  2022-09-16 04:00:00       67  8.660254e-01  0.500000\n",
       "..                 ...      ...           ...       ...\n",
       "95 2022-09-19 23:00:00       58 -2.588190e-01  0.965926\n",
       "96 2022-09-20 00:00:00       23 -2.932653e-12  1.000000\n",
       "97 2022-09-20 01:00:00       79  2.588190e-01  0.965926\n",
       "98 2022-09-20 02:00:00       13  5.000000e-01  0.866025\n",
       "99 2022-09-20 03:00:00       85  7.071068e-01  0.707107\n",
       "\n",
       "[100 rows x 4 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "88cc3e83",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[[\"Day sin\", \"Day cos\"]].plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d44d0e7b",
   "metadata": {},
   "source": [
    "# TIP 2 - Network tunning"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dfa5f00f",
   "metadata": {},
   "source": [
    "#### Tune your Network\n",
    "- Optmizer (https://keras.io/api/optimizers/)\n",
    "- Learning Rate (keras.optimizers.Adam(learning_rate=0.01))\n",
    "- Grid Search\n",
    "- Advanced Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "dc8d9c7e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# To be used with NB 7\n",
    "def bilstm_model(cnn_d=32, lstm_d=16):\n",
    "    \n",
    "    model = tf.keras.models.Sequential()\n",
    "    model.add(tf.keras.layers.Masking(mask_value=MASKING_VALUE))\n",
    "        \n",
    "    model.add(tf.keras.layers.Dropout(0.01))\n",
    "\n",
    "    model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(lstm_d, return_sequences=True)))\n",
    "    model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(lstm_d, return_sequences=True)))\n",
    "    \n",
    "    model.add(tf.keras.layers.Dense(64, activation=\"relu\"))\n",
    "    model.add(tf.keras.layers.Dense(16, activation=\"relu\"))\n",
    "    model.add(tf.keras.layers.Dense(1, activation=\"sigmoid\", name='output'))\n",
    "    \n",
    "    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    return model\n",
    "\n",
    "bilstm_model = bilstm_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c00c2246",
   "metadata": {},
   "source": [
    "# More Advanced networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ba50fb98",
   "metadata": {},
   "outputs": [],
   "source": [
    "# To be used with NB 7\n",
    "def bilstm_model(cnn_d=32, lstm_d=16):\n",
    "    \n",
    "    model = tf.keras.models.Sequential()\n",
    "    model.add(tf.keras.layers.Masking(mask_value=MASKING_VALUE))\n",
    "        \n",
    "    model.add(tf.keras.layers.Dropout(0.01))\n",
    "\n",
    "    model.add(tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(lstm_d, return_sequences=True)))\n",
    "\n",
    "    model.add(tf.keras.layers.Dense(64, activation=\"relu\"))\n",
    "    model.add(tf.keras.layers.Dense(16, activation=\"relu\"))\n",
    "    model.add(tf.keras.layers.Dense(1, activation=\"sigmoid\", name='output'))\n",
    "    \n",
    "    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    return model\n",
    "\n",
    "bilstm_model = bilstm_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "6cfc672c",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# To be used with NB 5 and 6\n",
    "def my_little_resnet_model(input_shape, blocks=3, cnn_d=32, stride=2):\n",
    "    \n",
    "    def residual_block(X, cnn_d, stride):\n",
    "        \n",
    "        out = keras.layers.Conv1D(cnn_d, stride, padding='same')(X)\n",
    "        out = keras.layers.ReLU()(out)\n",
    "        out = keras.layers.Conv1D(cnn_d, stride, padding='same')(out)\n",
    "        out = keras.layers.add([X, out])    # Adds the previous layer residual to the current output\n",
    "        out = keras.layers.ReLU()(out)\n",
    "        out = keras.layers.MaxPool1D(strides=2, padding=\"same\")(out)\n",
    "        return out\n",
    "    \n",
    "    inputs = keras.layers.Input(input_shape)\n",
    "    X = keras.layers.Conv1D(cnn_d, stride)(inputs)\n",
    "    for _ in range(blocks):\n",
    "        X = residual_block(X, cnn_d, stride)\n",
    "    \n",
    "    X = keras.layers.Flatten()(X)\n",
    "    X = keras.layers.Dense(32, activation='relu')(X)\n",
    "    X = keras.layers.Dense(8, activation='relu')(X)\n",
    "    output = keras.layers.Dense(1, activation='sigmoid')(X)\n",
    "\n",
    "    model = keras.Model(inputs=inputs, outputs=output)\n",
    "    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    \n",
    "    return model\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9ecfaeac",
   "metadata": {},
   "outputs": [],
   "source": [
    "# From: https://github.com/cauchyturing/UCR_Time_Series_Classification_Deep_Learning_Baseline/blob/master/ResNet.py\n",
    "# To be used with NB 5 and 6\n",
    "def build_resnet(input_shape, n_feature_maps):\n",
    "    \n",
    "    print ('build conv_x')\n",
    "    x = keras.layers.Input(shape=(input_shape))\n",
    "    conv_x = keras.layers.BatchNormalization()(x)\n",
    "    conv_x = keras.layers.Conv1D(n_feature_maps, 8, 1, padding='same')(conv_x)\n",
    "    conv_x = keras.layers.BatchNormalization()(conv_x)\n",
    "    conv_x = keras.layers.Activation('relu')(conv_x)\n",
    "     \n",
    "    print ('build conv_y')\n",
    "    conv_y = keras.layers.Conv1D(n_feature_maps, 5, 1, padding='same')(conv_x)\n",
    "    conv_y = keras.layers.BatchNormalization()(conv_y)\n",
    "    conv_y = keras.layers.Activation('relu')(conv_y)\n",
    "     \n",
    "    print ('build conv_z')\n",
    "    conv_z = keras.layers.Conv1D(n_feature_maps, 3, 1, padding='same')(conv_y)\n",
    "    conv_z = keras.layers.BatchNormalization()(conv_z)\n",
    "     \n",
    "    is_expand_channels = not (input_shape[-1] == n_feature_maps)\n",
    "    if is_expand_channels:\n",
    "        shortcut_y = keras.layers.Conv1D(n_feature_maps, 1, 1,padding='same')(x)\n",
    "        shortcut_y = keras.layers.BatchNormalization()(shortcut_y)\n",
    "    else:\n",
    "        shortcut_y = keras.layers.BatchNormalization()(x)\n",
    "    print ('Merging skip connection')\n",
    "    y = keras.layers.Add()([shortcut_y, conv_z])\n",
    "    y = keras.layers.Activation('relu')(y)\n",
    "     \n",
    "    print ('build conv_x')\n",
    "    x1 = y\n",
    "    conv_x = keras.layers.Conv1D(n_feature_maps*2, 8, 1, padding='same')(x1)\n",
    "    conv_x = keras.layers.BatchNormalization()(conv_x)\n",
    "    conv_x = keras.layers.Activation('relu')(conv_x)\n",
    "     \n",
    "    print ('build conv_y')\n",
    "    conv_y = keras.layers.Conv1D(n_feature_maps*2, 5, 1, padding='same')(conv_x)\n",
    "    conv_y = keras.layers.BatchNormalization()(conv_y)\n",
    "    conv_y = keras.layers.Activation('relu')(conv_y)\n",
    "     \n",
    "    print ('build conv_z')\n",
    "    conv_z = keras.layers.Conv1D(n_feature_maps*2, 3, 1, padding='same')(conv_y)\n",
    "    conv_z = keras.layers.BatchNormalization()(conv_z)\n",
    "     \n",
    "    is_expand_channels = not (input_shape[-1] == n_feature_maps*2)\n",
    "    if is_expand_channels:\n",
    "        shortcut_y = keras.layers.Conv1D(n_feature_maps*2, 1, 1,padding='same')(x1)\n",
    "        shortcut_y = keras.layers.BatchNormalization()(shortcut_y)\n",
    "    else:\n",
    "        shortcut_y = keras.layers.BatchNormalization()(x1)\n",
    "    print ('Merging skip connection')\n",
    "    y = keras.layers.Add()([shortcut_y, conv_z])\n",
    "    y = keras.layers.Activation('relu')(y)\n",
    "     \n",
    "    print ('build conv_x')\n",
    "    x1 = y\n",
    "    conv_x = keras.layers.Conv1D(n_feature_maps*2, 8, 1, padding='same')(x1)\n",
    "    conv_x = keras.layers.BatchNormalization()(conv_x)\n",
    "    conv_x = keras.layers.Activation('relu')(conv_x)\n",
    "     \n",
    "    print ('build conv_y')\n",
    "    conv_y = keras.layers.Conv1D(n_feature_maps*2, 5, 1, padding='same')(conv_x)\n",
    "    conv_y = keras.layers.BatchNormalization()(conv_y)\n",
    "    conv_y = keras.layers.Activation('relu')(conv_y)\n",
    "     \n",
    "    print ('build conv_z')\n",
    "    conv_z = keras.layers.Conv1D(n_feature_maps*2, 3, 1, padding='same')(conv_y)\n",
    "    conv_z = keras.layers.BatchNormalization()(conv_z)\n",
    "\n",
    "    is_expand_channels = not (input_shape[-1] == n_feature_maps*2)\n",
    "    if is_expand_channels:\n",
    "        shortcut_y = keras.layers.Conv1D(n_feature_maps*2, 1, 1,padding='same')(x1)\n",
    "        shortcut_y = keras.layers.BatchNormalization()(shortcut_y)\n",
    "    else:\n",
    "        shortcut_y = keras.layers.BatchNormalization()(x1)\n",
    "    print ('Merging skip connection')\n",
    "    y = keras.layers.Add()([shortcut_y, conv_z])\n",
    "    y = keras.layers.Activation('relu')(y)\n",
    "     \n",
    "    full = keras.layers.GlobalAveragePooling1D()(y)\n",
    "    out = keras.layers.Dense(1, activation='sigmoid')(full)\n",
    "    print ('        -- model was built.')\n",
    "    \n",
    "    model = keras.Model(inputs=x, outputs=out)\n",
    "    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "    \n",
    "    return model\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "be89f795",
   "metadata": {},
   "outputs": [],
   "source": [
    "# WaveNet: https://www.deepmind.com/blog/wavenet-a-generative-model-for-raw-audio\n",
    "\n",
    "def WaveNetResidualConv1D(num_filters, kernel_size, stacked_layer):\n",
    "\n",
    "    def build_residual_block(l_input):\n",
    "        resid_input = l_input\n",
    "        for dilation_rate in [2**i for i in range(stacked_layer)]:\n",
    "            l_sigmoid_conv1d = Conv1D(\n",
    "              num_filters, kernel_size, dilation_rate=dilation_rate,\n",
    "              padding='same', activation='sigmoid')(l_input)\n",
    "            l_tanh_conv1d = Conv1D(\n",
    "             num_filters, kernel_size, dilation_rate=dilation_rate,\n",
    "             padding='same', activation='mish')(l_input)\n",
    "            l_input = Multiply()([l_sigmoid_conv1d, l_tanh_conv1d])\n",
    "            l_input = Conv1D(num_filters, 1, padding='same')(l_input)\n",
    "            resid_input = Add()([resid_input ,l_input])\n",
    "        return resid_input\n",
    "    return build_residual_block\n",
    "\n",
    "def Classifier(shape_):\n",
    "    num_filters_ = 16\n",
    "    kernel_size_ = 3\n",
    "    stacked_layers_ = [12, 8, 4, 1]\n",
    "    l_input = Input(shape=(shape_))\n",
    "    x = Conv1D(num_filters_, 1, padding='same')(l_input)\n",
    "    x = WaveNetResidualConv1D(num_filters_, kernel_size_, stacked_layers_[0])(x)\n",
    "    x = Conv1D(num_filters_*2, 1, padding='same')(x)\n",
    "    x = WaveNetResidualConv1D(num_filters_*2, kernel_size_, stacked_layers_[1])(x)\n",
    "    x = Conv1D(num_filters_*4, 1, padding='same')(x)\n",
    "    x = WaveNetResidualConv1D(num_filters_*4, kernel_size_, stacked_layers_[2])(x)\n",
    "    x = Conv1D(num_filters_*8, 1, padding='same')(x)\n",
    "    x = WaveNetResidualConv1D(num_filters_*8, kernel_size_, stacked_layers_[3])(x)\n",
    "    l_output = Dense(11, activation='softmax')(x)\n",
    "    model = models.Model(inputs=[l_input], outputs=[l_output])\n",
    "    opt = Adam(lr=LR)\n",
    "    opt = tfa.optimizers.SWA(opt)\n",
    "    model.compile(loss=losses.CategoricalCrossentropy(), optimizer=opt, metrics=['accuracy'])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30ae8143",
   "metadata": {},
   "source": [
    "# Approaches based on spectograms (NB 7)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6cc572b7",
   "metadata": {},
   "source": [
    "\n",
    "Image from [dolby.io](https://dolby.io/blog/beginners-guide-to-visualizing-audio-as-a-spectogram-in-python/):\n",
    "\n",
    "<img src=\"images/python-spectogram-annotated.png\" width=\"2000\" height=\"200\">\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7cbdeb0e",
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'X_train' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "Cell \u001b[0;32mIn [2], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m Xcut \u001b[38;5;241m=\u001b[39m \u001b[43mX_train\u001b[49m[\u001b[38;5;241m1\u001b[39m][:\u001b[38;5;241m400\u001b[39m, \u001b[38;5;241m0\u001b[39m]\n\u001b[1;32m      3\u001b[0m powerSpectrum, frequenciesFound, time, imageAxis \u001b[38;5;241m=\u001b[39m plt\u001b[38;5;241m.\u001b[39mspecgram(Xcut, Fs\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m1\u001b[39m\u001b[38;5;241m/\u001b[39m\u001b[38;5;241m30.\u001b[39m)\n\u001b[1;32m      4\u001b[0m plt\u001b[38;5;241m.\u001b[39mshow()\n",
      "\u001b[0;31mNameError\u001b[0m: name 'X_train' is not defined"
     ]
    }
   ],
   "source": [
    "Xcut = X_train[1][:400, 0]\n",
    "\n",
    "powerSpectrum, frequenciesFound, time, imageAxis = plt.specgram(Xcut, Fs=1/30.)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1076e708",
   "metadata": {},
   "source": [
    "# Viz of ML results\n",
    "\n",
    "In case you are curious to know how good your model ended up being"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6dfbb8bc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# For NB 5 and 6\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_eval_one_pid(df, test_range, nnmodel, x_test, pid, act_col=\"act_10\", hr_col=\"hr_10\"):\n",
    "    \n",
    "        \n",
    "    df_held_out_test = df[df[\"pid\"].isin(test_range)][[\"pid\", \"time\", \"sleep_phase\", \"sleep\",  act_col, hr_col]].copy()\n",
    "    df_held_out_test[\"yhat\"] = nnmodel.predict(x_test).round()\n",
    "\n",
    "    # Plot image for this one pid\n",
    "    fig, (ax1, ax2, ax3, ax4) = plt.subplots(4, 1, sharex=True)\n",
    "    \n",
    "    df1 = df_held_out_test[df_held_out_test[\"pid\"] == pid].copy()\n",
    "    \n",
    "    df1[\"sleep\"] = df1[\"sleep\"].astype(int)\n",
    "    df1[[\"time\", act_col]].plot(x=\"time\", y=act_col, ax=ax1, color=\"green\")\n",
    "    df1[[\"time\", hr_col]].plot(x=\"time\", y=hr_col, ax=ax2, color=\"red\")\n",
    "    df1[[\"time\", \"sleep\"]].plot(x=\"time\", y=\"sleep\",  ax=ax3, color=\"blue\")\n",
    "    df1[[\"time\", \"yhat\"]].plot(x=\"time\", y=\"yhat\",  ax=ax4, color=\"purple\")\n",
    "    \n",
    "    ax1.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax2.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax3.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax4.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    plt.show()    \n",
    "\n",
    "\n",
    "plot_eval_one_pid(df, range(50, 100), cnnlstm_model, X_test, pid=51, act_col=\"act\", hr_col=\"hr\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ca55e145",
   "metadata": {},
   "outputs": [],
   "source": [
    "# For NB 7\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_eval_one_pid(df, test_range, maxdim, nnmodel, x_test, y_test, masking_value, pid):\n",
    "    \n",
    "    df_held_out_test = df[df[\"pid\"].isin(test_range)][[\"pid\", \"time\", \"sleep_phase\", \"sleep\", \"act_10\", \"hr_10\"]].copy()\n",
    "    df_held_out_test = df_held_out_test.groupby(\"pid\").head(maxdim).reset_index(drop=True)\n",
    "    \n",
    "    p = nnmodel.predict(x_test).round()\n",
    "    df_held_out_test[\"yhat\"] = p[y_test != masking_value].round()\n",
    "    \n",
    "    # Plot image for this one pid\n",
    "    fig, (ax1, ax2, ax3, ax4) = plt.subplots(4, 1, sharex=True)\n",
    "    \n",
    "    df1 = df_held_out_test[df_held_out_test[\"pid\"] == pid].copy()\n",
    "    \n",
    "    df1[\"sleep\"] = df1[\"sleep\"].astype(int)\n",
    "    df1[[\"time\", \"act_10\"]].plot(x=\"time\", y=\"act_10\", ax=ax1, color=\"green\")\n",
    "    df1[[\"time\", \"hr_10\"]].plot(x=\"time\", y=\"hr_10\", ax=ax2, color=\"red\")\n",
    "    df1[[\"time\", \"sleep\"]].plot(x=\"time\", y=\"sleep\",  ax=ax3, color=\"blue\")\n",
    "    df1[[\"time\", \"yhat\"]].plot(x=\"time\", y=\"yhat\",  ax=ax4, color=\"purple\")\n",
    "    \n",
    "    ax1.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax2.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax3.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    ax4.legend(loc='center left', bbox_to_anchor=(1.0, 0.5))\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "plot_eval_one_pid(df, range(50, 100), MAX_DIMENSION, bilstm_model,\n",
    "             X_test.astype(float), Y_test, MASKING_VALUE, pid=51)"
   ]
  }
 ],
 "metadata": {
  "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}