2023/machine_learning/sklearn_gravityspy_glitch_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "17d57049-f41f-462f-8aaf-9c663c7884e6",
   "metadata": {},
   "source": [
    "# Gravity Spy Machine Learning Classifications of LIGO Glitches from Observing Runs O1, O2, O3a, and O3b"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2aa85c7a-1f3a-48e5-b07b-5724486f8573",
   "metadata": {},
   "source": [
    "这个数据集名为 \"Gravity Spy Machine Learning Classifications of LIGO Glitches from Observing Runs O1, O2, O3a, and O3b\"，由Glanzer等人在2021年发布 (https://zenodo.org/record/5649212)。\n",
    "\n",
    "数据集包含了LIGO观测运行O1, O2, O3a和O3b期间，Gravity Spy机器学习模型对LIGO故障的所有分类。\n",
    "\n",
    "在这个教程中，我们将使用其中的 H1_O1 和 L1_O1 数据集。这些数据集包含了Omicron触发管道识别的所有噪声事件的分类，其中Omicron识别的信噪比超过7.5，噪声事件的峰值频率在10 Hz和2048 Hz之间。\n",
    "\n",
    "每个数据集中的列包括：\n",
    "\n",
    "- Omicron 流水线的信号元数据（如event_time, ifo, peak_time, peak_time_ns, start_time, start_time_ns, duration, peak_frequency, central_freq, bandwidth, channel, amplitude, snr, q_value）。\n",
    "- 每个 glitch 在数据集中的唯一标识符（gravityspy_id）。\n",
    "- 机器学习对每个故障属于特定Gravity Spy类的置信度（如1400Ripples, 1080Lines, Air_Compressor, Blip, Chirp, Extremely_Loud, Helix, Koi_Fish, Light_Modulation, Low_Frequency_Burst, Low_Frequency_Lines, No_Glitch, None_of_the_Above, Paired_Doves, Power_Line, Repeating_Blips, Scattered_Light, Scratchy, Tomte, Violin_Mode, Wandering_Line, Whistle）。\n",
    "- 机器学习预测的每个 glitch 的标签和分类的置信度（ml_label, ml_confidence）。\n",
    "- 每个 glitch 的公开可用的 [Omega scan](https://en.wikipedia.org/wiki/Constant-Q_transform) 的链接（url1, url2, url3, url4）。\n",
    "\n",
    "我们将使用这些数据来搭建和训练分类模型。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47b531bb-c55f-4423-84af-db2751f7229e",
   "metadata": {},
   "source": [
    "# Download data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a266edb5-14c2-49d9-a2da-568e4f589790",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# !wget https://zenodo.org/records/5649212/files/H1_O1.csv -P data/\n",
    "# !wget https://zenodo.org/records/5649212/files/L1_O1.csv -P data/"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dc1ea4b5-cc87-4b16-a2ca-40b799d567cd",
   "metadata": {},
   "source": [
    "# Load libs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "bb3dec0a-6d58-4645-a277-8eb5c87d210a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm\n",
    "\n",
    "# !pip install gwpy\n",
    "from gwpy.time import to_gps, from_gps\n",
    "\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.feature_selection import SelectFromModel\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import GridSearchCV, ShuffleSplit, ParameterGrid, StratifiedKFold"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "11acb4d7-405d-4fda-b56d-e05369642ead",
   "metadata": {},
   "source": [
    "\n",
    "# Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "9383c315-198b-4f00-a429-af0ead5f9bd9",
   "metadata": {
    "tags": []
   },
   "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>event_time</th>\n",
       "      <th>ifo</th>\n",
       "      <th>peak_time</th>\n",
       "      <th>peak_time_ns</th>\n",
       "      <th>start_time</th>\n",
       "      <th>start_time_ns</th>\n",
       "      <th>duration</th>\n",
       "      <th>peak_frequency</th>\n",
       "      <th>central_freq</th>\n",
       "      <th>bandwidth</th>\n",
       "      <th>snr</th>\n",
       "      <th>amplitude</th>\n",
       "      <th>ml_label</th>\n",
       "      <th>ml_confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.133747e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1133747480</td>\n",
       "      <td>875000000</td>\n",
       "      <td>1133747479</td>\n",
       "      <td>625000000</td>\n",
       "      <td>3.12500</td>\n",
       "      <td>23.161011</td>\n",
       "      <td>3083.450439</td>\n",
       "      <td>6125.699219</td>\n",
       "      <td>12.31222</td>\n",
       "      <td>9.112720e-22</td>\n",
       "      <td>Scattered_Light</td>\n",
       "      <td>0.999992</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.133139e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1133138521</td>\n",
       "      <td>980957031</td>\n",
       "      <td>1133138521</td>\n",
       "      <td>937500000</td>\n",
       "      <td>0.12500</td>\n",
       "      <td>262.064606</td>\n",
       "      <td>1576.186646</td>\n",
       "      <td>3022.359619</td>\n",
       "      <td>15.72605</td>\n",
       "      <td>2.482060e-22</td>\n",
       "      <td>Blip</td>\n",
       "      <td>0.999553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.131777e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1131777198</td>\n",
       "      <td>383789062</td>\n",
       "      <td>1131777196</td>\n",
       "      <td>445312023</td>\n",
       "      <td>3.05469</td>\n",
       "      <td>111.128151</td>\n",
       "      <td>281.727661</td>\n",
       "      <td>546.868347</td>\n",
       "      <td>8.31163</td>\n",
       "      <td>1.555730e-22</td>\n",
       "      <td>Blip</td>\n",
       "      <td>0.993591</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.133107e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1133106912</td>\n",
       "      <td>958983898</td>\n",
       "      <td>1133106912</td>\n",
       "      <td>937500000</td>\n",
       "      <td>0.12500</td>\n",
       "      <td>89.676369</td>\n",
       "      <td>201.923874</td>\n",
       "      <td>319.184143</td>\n",
       "      <td>11.29298</td>\n",
       "      <td>3.454280e-22</td>\n",
       "      <td>Blip</td>\n",
       "      <td>0.998819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.132660e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1132659628</td>\n",
       "      <td>687500000</td>\n",
       "      <td>1132659627</td>\n",
       "      <td>500000000</td>\n",
       "      <td>2.50000</td>\n",
       "      <td>10.771410</td>\n",
       "      <td>3876.945312</td>\n",
       "      <td>7741.605469</td>\n",
       "      <td>9.74269</td>\n",
       "      <td>1.326820e-19</td>\n",
       "      <td>Low_Frequency_Lines</td>\n",
       "      <td>0.984899</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25845</th>\n",
       "      <td>1.137220e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1137219636</td>\n",
       "      <td>326172113</td>\n",
       "      <td>1137219636</td>\n",
       "      <td>250000000</td>\n",
       "      <td>0.10547</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>1802.779419</td>\n",
       "      <td>1581.367554</td>\n",
       "      <td>12.79415</td>\n",
       "      <td>5.758980e-22</td>\n",
       "      <td>1400Ripples</td>\n",
       "      <td>0.999278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25846</th>\n",
       "      <td>1.130257e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1130256563</td>\n",
       "      <td>31250000</td>\n",
       "      <td>1130256562</td>\n",
       "      <td>875000000</td>\n",
       "      <td>0.50000</td>\n",
       "      <td>64.891693</td>\n",
       "      <td>64.202530</td>\n",
       "      <td>8.083960</td>\n",
       "      <td>8.34971</td>\n",
       "      <td>1.164350e-22</td>\n",
       "      <td>Power_Line</td>\n",
       "      <td>0.999904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25847</th>\n",
       "      <td>1.131414e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1131413524</td>\n",
       "      <td>687500000</td>\n",
       "      <td>1131413524</td>\n",
       "      <td>0</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>65.531433</td>\n",
       "      <td>66.881851</td>\n",
       "      <td>5.131030</td>\n",
       "      <td>7.70147</td>\n",
       "      <td>9.419780e-23</td>\n",
       "      <td>Power_Line</td>\n",
       "      <td>0.999980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25848</th>\n",
       "      <td>1.129983e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1129982991</td>\n",
       "      <td>312500000</td>\n",
       "      <td>1129982988</td>\n",
       "      <td>750000000</td>\n",
       "      <td>3.50000</td>\n",
       "      <td>11.519660</td>\n",
       "      <td>3979.047363</td>\n",
       "      <td>7942.870605</td>\n",
       "      <td>14.02844</td>\n",
       "      <td>7.782930e-19</td>\n",
       "      <td>Low_Frequency_Lines</td>\n",
       "      <td>0.992579</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25849</th>\n",
       "      <td>1.132738e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1132737675</td>\n",
       "      <td>93750000</td>\n",
       "      <td>1132737674</td>\n",
       "      <td>750000000</td>\n",
       "      <td>0.81250</td>\n",
       "      <td>83.005531</td>\n",
       "      <td>2352.000977</td>\n",
       "      <td>4540.329102</td>\n",
       "      <td>8.13509</td>\n",
       "      <td>9.043470e-23</td>\n",
       "      <td>No_Glitch</td>\n",
       "      <td>0.986779</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25850 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         event_time ifo   peak_time  peak_time_ns  start_time  start_time_ns  \\\n",
       "0      1.133747e+09  H1  1133747480     875000000  1133747479      625000000   \n",
       "1      1.133139e+09  H1  1133138521     980957031  1133138521      937500000   \n",
       "2      1.131777e+09  H1  1131777198     383789062  1131777196      445312023   \n",
       "3      1.133107e+09  H1  1133106912     958983898  1133106912      937500000   \n",
       "4      1.132660e+09  H1  1132659628     687500000  1132659627      500000000   \n",
       "...             ...  ..         ...           ...         ...            ...   \n",
       "25845  1.137220e+09  L1  1137219636     326172113  1137219636      250000000   \n",
       "25846  1.130257e+09  L1  1130256563      31250000  1130256562      875000000   \n",
       "25847  1.131414e+09  L1  1131413524     687500000  1131413524              0   \n",
       "25848  1.129983e+09  L1  1129982991     312500000  1129982988      750000000   \n",
       "25849  1.132738e+09  L1  1132737675      93750000  1132737674      750000000   \n",
       "\n",
       "       duration  peak_frequency  central_freq    bandwidth       snr  \\\n",
       "0       3.12500       23.161011   3083.450439  6125.699219  12.31222   \n",
       "1       0.12500      262.064606   1576.186646  3022.359619  15.72605   \n",
       "2       3.05469      111.128151    281.727661   546.868347   8.31163   \n",
       "3       0.12500       89.676369    201.923874   319.184143  11.29298   \n",
       "4       2.50000       10.771410   3876.945312  7741.605469   9.74269   \n",
       "...         ...             ...           ...          ...       ...   \n",
       "25845   0.10547     1527.022095   1802.779419  1581.367554  12.79415   \n",
       "25846   0.50000       64.891693     64.202530     8.083960   8.34971   \n",
       "25847   1.00000       65.531433     66.881851     5.131030   7.70147   \n",
       "25848   3.50000       11.519660   3979.047363  7942.870605  14.02844   \n",
       "25849   0.81250       83.005531   2352.000977  4540.329102   8.13509   \n",
       "\n",
       "          amplitude             ml_label  ml_confidence  \n",
       "0      9.112720e-22      Scattered_Light       0.999992  \n",
       "1      2.482060e-22                 Blip       0.999553  \n",
       "2      1.555730e-22                 Blip       0.993591  \n",
       "3      3.454280e-22                 Blip       0.998819  \n",
       "4      1.326820e-19  Low_Frequency_Lines       0.984899  \n",
       "...             ...                  ...            ...  \n",
       "25845  5.758980e-22          1400Ripples       0.999278  \n",
       "25846  1.164350e-22           Power_Line       0.999904  \n",
       "25847  9.419780e-23           Power_Line       0.999980  \n",
       "25848  7.782930e-19  Low_Frequency_Lines       0.992579  \n",
       "25849  9.043470e-23            No_Glitch       0.986779  \n",
       "\n",
       "[25850 rows x 14 columns]"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.concat([\n",
    "    pd.read_csv('data/H1_O1.csv'),\n",
    "    pd.read_csv('data/L1_O1.csv')\n",
    "])\n",
    "\n",
    "cols = ['event_time', 'ifo', 'peak_time', 'peak_time_ns', 'start_time', 'start_time_ns',\n",
    "        'duration', 'peak_frequency', 'central_freq', 'bandwidth', 'snr', \n",
    "        'amplitude', 'ml_label', 'ml_confidence']\n",
    "df = df[cols]\n",
    "df = df[df['ml_confidence'] >= 0.95]\n",
    "df = df.reset_index().drop('index', axis=1)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "dbfbde1f-4b62-4422-b13a-82d63f681b3a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 25850 entries, 0 to 25849\n",
      "Data columns (total 14 columns):\n",
      " #   Column          Non-Null Count  Dtype  \n",
      "---  ------          --------------  -----  \n",
      " 0   event_time      25850 non-null  float64\n",
      " 1   ifo             25850 non-null  object \n",
      " 2   peak_time       25850 non-null  int64  \n",
      " 3   peak_time_ns    25850 non-null  int64  \n",
      " 4   start_time      25850 non-null  int64  \n",
      " 5   start_time_ns   25850 non-null  int64  \n",
      " 6   duration        25850 non-null  float64\n",
      " 7   peak_frequency  25850 non-null  float64\n",
      " 8   central_freq    25850 non-null  float64\n",
      " 9   bandwidth       25850 non-null  float64\n",
      " 10  snr             25850 non-null  float64\n",
      " 11  amplitude       25850 non-null  float64\n",
      " 12  ml_label        25850 non-null  object \n",
      " 13  ml_confidence   25850 non-null  float64\n",
      "dtypes: float64(8), int64(4), object(2)\n",
      "memory usage: 2.8+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "e552e011-e85a-487e-9d7d-408ee0c4b220",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ml_label\n",
       "Low_Frequency_Lines    6592\n",
       "Low_Frequency_Burst    4823\n",
       "Scattered_Light        3020\n",
       "Blip                   2690\n",
       "Power_Line             1888\n",
       "Scratchy               1673\n",
       "No_Glitch              1173\n",
       "1400Ripples            1159\n",
       "Koi_Fish                768\n",
       "Light_Modulation        484\n",
       "Repeating_Blips         314\n",
       "Whistle                 310\n",
       "Extremely_Loud          275\n",
       "Helix                   241\n",
       "Violin_Mode             125\n",
       "Paired_Doves             80\n",
       "Air_Compressor           77\n",
       "Chirp                    59\n",
       "Tomte                    36\n",
       "Wandering_Line           36\n",
       "None_of_the_Above        25\n",
       "1080Lines                 2\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.ml_label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "c2780eaa-0f9d-4ff9-8053-ad45241acbc2",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(24650, 14)"
      ]
     },
     "execution_count": 162,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_labels = [label for label in pd.unique(df.ml_label) if label not in ['None_of_the_Above', '1080Lines', 'No_Glitch']]\n",
    "df = pd.concat([df[df.ml_label == label]\n",
    "                         for label in target_labels])\n",
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "e812ceea-a3a1-4cf5-9828-a8d97eaf4ef9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ml_label\n",
       "Low_Frequency_Lines    6592\n",
       "Low_Frequency_Burst    4823\n",
       "Scattered_Light        3020\n",
       "Blip                   2690\n",
       "Power_Line             1888\n",
       "Scratchy               1673\n",
       "1400Ripples            1159\n",
       "Koi_Fish                768\n",
       "Light_Modulation        484\n",
       "Repeating_Blips         314\n",
       "Whistle                 310\n",
       "Extremely_Loud          275\n",
       "Helix                   241\n",
       "Violin_Mode             125\n",
       "Paired_Doves             80\n",
       "Air_Compressor           77\n",
       "Chirp                    59\n",
       "Tomte                    36\n",
       "Wandering_Line           36\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.ml_label.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9213754b-6def-4d98-b7ef-11f927e7f993",
   "metadata": {},
   "source": [
    "# Data prepararing + Feature engineering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "d9243375-5129-4e15-9c04-cfc491235746",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 17.6 s, sys: 522 ms, total: 18.1 s\n",
      "Wall time: 17.4 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# 创建 ID\n",
    "df.loc[:, 'id'] = df.event_time.map(lambda x: from_gps(x).strftime('g%y%m%d_%H%M%S+')+str(int(from_gps(x).microsecond/1000)).zfill(3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "id": "9c0ce6f2-14a0-430c-a0cd-2685ace76f22",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "dcdc388a-4906-4475-b9e3-299b49aa5440",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 17 s, sys: 480 ms, total: 17.5 s\n",
      "Wall time: 16.9 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# 对时间特征处理，取出关键信息\n",
    "df.loc[:, 'peak_time_fullgps'] = df.peak_time + 1e-9*df.peak_time_ns\n",
    "df.loc[:, 'start_time_fullgps'] = df.start_time + 1e-9*df.start_time_ns\n",
    "df.loc[:, 'peak_time'] = df.peak_time_fullgps.map(lambda x: from_gps(x))\n",
    "df.loc[:, 'start_time'] = df.start_time_fullgps.map(lambda x: from_gps(x))\n",
    "\n",
    "for target_time in ['peak_time', 'start_time']:\n",
    "    df.loc[:, f'{target_time}_day'] = df[f'{target_time}'].dt.day\n",
    "    df.loc[:, f'{target_time}_year'] = df[f'{target_time}'].dt.year\n",
    "    df.loc[:, f'{target_time}_month'] = df[f'{target_time}'].dt.month\n",
    "    df.loc[:, f'{target_time}_hour'] = df[f'{target_time}'].dt.hour\n",
    "    df.loc[:, f'{target_time}_minute'] = df[f'{target_time}'].dt.minute\n",
    "    df.loc[:, f'{target_time}_second'] = df[f'{target_time}'].dt.second\n",
    "    df.loc[:, f'{target_time}_microsecond'] = df[f'{target_time}'].dt.microsecond\n",
    "    df.loc[:, f'{target_time}_day_of_week'] = df[f'{target_time}'].dt.day_of_week\n",
    "    df.loc[:, f'{target_time}_day_of_year'] = df[f'{target_time}'].dt.day_of_year\n",
    "    df.loc[:, f'{target_time}_weekday'] = df[f'{target_time}'].dt.weekday"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "id": "46722da1-c157-4965-a683-0238cb27b07f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# 对标签类别进行编码\n",
    "class_le = LabelEncoder()\n",
    "df.loc[:, 'ml_label'] = class_le.fit_transform(df['ml_label'].values)\n",
    "df.loc[:, 'label'] = df.ml_label.astype('int')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "590e6d01-f677-4b19-a408-2e6df2a2b915",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 24650 entries, 0 to 25845\n",
      "Data columns (total 32 columns):\n",
      " #   Column                  Non-Null Count  Dtype  \n",
      "---  ------                  --------------  -----  \n",
      " 0   duration                24650 non-null  float64\n",
      " 1   peak_frequency          24650 non-null  float64\n",
      " 2   central_freq            24650 non-null  float64\n",
      " 3   bandwidth               24650 non-null  float64\n",
      " 4   snr                     24650 non-null  float64\n",
      " 5   amplitude               24650 non-null  float64\n",
      " 6   id                      24650 non-null  object \n",
      " 7   peak_time_fullgps       24650 non-null  float64\n",
      " 8   start_time_fullgps      24650 non-null  float64\n",
      " 9   peak_time_day           24650 non-null  int32  \n",
      " 10  peak_time_year          24650 non-null  int32  \n",
      " 11  peak_time_month         24650 non-null  int32  \n",
      " 12  peak_time_hour          24650 non-null  int32  \n",
      " 13  peak_time_minute        24650 non-null  int32  \n",
      " 14  peak_time_second        24650 non-null  int32  \n",
      " 15  peak_time_microsecond   24650 non-null  int32  \n",
      " 16  peak_time_day_of_week   24650 non-null  int32  \n",
      " 17  peak_time_day_of_year   24650 non-null  int32  \n",
      " 18  peak_time_weekday       24650 non-null  int32  \n",
      " 19  start_time_day          24650 non-null  int32  \n",
      " 20  start_time_year         24650 non-null  int32  \n",
      " 21  start_time_month        24650 non-null  int32  \n",
      " 22  start_time_hour         24650 non-null  int32  \n",
      " 23  start_time_minute       24650 non-null  int32  \n",
      " 24  start_time_second       24650 non-null  int32  \n",
      " 25  start_time_microsecond  24650 non-null  int32  \n",
      " 26  start_time_day_of_week  24650 non-null  int32  \n",
      " 27  start_time_day_of_year  24650 non-null  int32  \n",
      " 28  start_time_weekday      24650 non-null  int32  \n",
      " 29  label                   24650 non-null  int64  \n",
      " 30  ifo_H1                  24650 non-null  float64\n",
      " 31  ifo_L1                  24650 non-null  float64\n",
      "dtypes: float64(10), int32(20), int64(1), object(1)\n",
      "memory usage: 4.3+ MB\n"
     ]
    }
   ],
   "source": [
    "# 去掉不需要的特征列；把 ifo 特征进行独热向量编码\n",
    "df = df.drop(['event_time', 'peak_time', 'peak_time_ns', 'start_time', 'start_time_ns', 'ml_confidence', 'ml_label'], axis=1)\n",
    "df = pd.concat([df, pd.get_dummies(df[['ifo']], dtype='float')], axis=1).drop('ifo', axis=1)\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d07533c1-a625-4e57-b698-b96b90484c51",
   "metadata": {},
   "source": [
    "# Train data + test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "770ebfb3-3b63-4900-b6f0-7b11e32280f4",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((17255, 30), (17255,), (7395, 30), (7395,))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns = df.drop(['label', 'id'], axis=1).columns.tolist()\n",
    "X, y = df.drop(['label', 'id'], axis=1), df.label\n",
    "\n",
    "# 切分训练集和测试集\n",
    "X_train, X_test, y_train, y_test = \\\n",
    "    train_test_split(X, y, test_size=0.3, random_state=0)\n",
    "X_train.shape, y_train.shape, X_test.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "3d22dc7e-ada2-402c-adb2-5d840e8fe253",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# 标准化\n",
    "stdsc = StandardScaler()\n",
    "X_train_std = stdsc.fit_transform(X_train)\n",
    "X_test_std = stdsc.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9056124a-618a-44f9-b5b7-f09180d17f47",
   "metadata": {},
   "source": [
    "# Model: Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "0bc93ab4-6216-4310-9a1c-eaeedcbad564",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 0.7035641842944074\n",
      "Test accuracy: 0.7207572684246112\n"
     ]
    }
   ],
   "source": [
    "# 小试牛刀\n",
    "lr = LogisticRegression(penalty='l1', solver='liblinear', C=0.1)\n",
    "lr.fit(X_train_std, y_train)\n",
    "print('Training accuracy:', lr.score(X_train_std, y_train))\n",
    "print('Test accuracy:', lr.score(X_test_std, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3cb262f9-4377-4b11-832e-bd8bc7af07bb",
   "metadata": {},
   "source": [
    "# Model: Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "997572f5-176d-4c71-8f11-6631f8c0961f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 15min 17s, sys: 17.4 s, total: 15min 35s\n",
      "Wall time: 41.2 s\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {color: black;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" checked><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "# 随机森林玩一下！看看特征的重要性\n",
    "forest = RandomForestClassifier(n_estimators=10000,\n",
    "                                random_state=0,\n",
    "                                n_jobs=-1)\n",
    "forest.fit(X_train_std, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "f38111d0-a264-41b6-93b8-3515372d067e",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 1.0\n",
      "Test accuracy: 0.9245436105476673\n"
     ]
    }
   ],
   "source": [
    "# 准确率直接干翻逻辑回归！但是显然过拟合严重。。。\n",
    "print('Training accuracy:', forest.score(X_train_std, y_train))\n",
    "print('Test accuracy:', forest.score(X_test_std, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "62497b1e-4550-4540-9fd5-21420d3a5982",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 1) peak_frequency                 0.251434\n",
      " 2) amplitude                      0.142417\n",
      " 3) duration                       0.100266\n",
      " 4) snr                            0.084439\n",
      " 5) start_time_fullgps             0.050502\n",
      " 6) peak_time_fullgps              0.050324\n",
      " 7) central_freq                   0.041764\n",
      " 8) bandwidth                      0.041271\n",
      " 9) peak_time_day_of_year          0.022656\n",
      "10) start_time_day_of_year         0.022510\n",
      "11) ifo_H1                         0.018443\n",
      "12) ifo_L1                         0.018280\n",
      "13) peak_time_day                  0.013818\n",
      "14) start_time_day                 0.013731\n",
      "15) peak_time_microsecond          0.013493\n",
      "16) start_time_hour                0.011752\n",
      "17) peak_time_hour                 0.011705\n",
      "18) peak_time_minute               0.010069\n",
      "19) start_time_minute              0.010039\n",
      "20) peak_time_second               0.009586\n",
      "21) start_time_second              0.009572\n",
      "22) start_time_microsecond         0.008237\n",
      "23) start_time_day_of_week         0.006520\n",
      "24) peak_time_day_of_week          0.006514\n",
      "25) peak_time_weekday              0.006504\n",
      "26) start_time_weekday             0.006482\n",
      "27) peak_time_month                0.006303\n",
      "28) start_time_month               0.006166\n",
      "29) peak_time_year                 0.002625\n",
      "30) start_time_year                0.002579\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "feat_labels = columns\n",
    "importances = forest.feature_importances_\n",
    "\n",
    "indices = np.argsort(importances)[::-1]\n",
    "\n",
    "for f in range(X_train.shape[1]):\n",
    "    print(\"%2d) %-*s %f\" % (f + 1, 30, \n",
    "                            feat_labels[indices[f]], \n",
    "                            importances[indices[f]]))\n",
    "    \n",
    "plt.title('Feature Importances')\n",
    "plt.bar(range(X_train.shape[1]), \n",
    "        importances[indices],\n",
    "        color='lightblue', \n",
    "        align='center')\n",
    "\n",
    "plt.xticks(range(X_train.shape[1]), \n",
    "           np.array(feat_labels)[indices], rotation=90)\n",
    "plt.xlim([-1, X_train.shape[1]])\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "3830c3f4-b218-45ea-9eb0-0c919836a09d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(17255, 8)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 挑选出明显重要性强的 8 个特征\n",
    "sfm = SelectFromModel(forest, threshold=0.03, prefit=True)\n",
    "X_train_std_selected = sfm.transform(X_train_std)\n",
    "X_test_std_selected = sfm.transform(X_test_std)\n",
    "\n",
    "X_train_std_selected.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "id": "d7fdefbd-d441-4b91-b5af-3c5e0b8dad8a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 13min 44s, sys: 15.6 s, total: 14min\n",
      "Wall time: 41.2 s\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-8 {color: black;}#sk-container-id-8 pre{padding: 0;}#sk-container-id-8 div.sk-toggleable {background-color: white;}#sk-container-id-8 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-8 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-8 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-8 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-8 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-8 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-8 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-8 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-8 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-8 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-8 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-8 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-8 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-8 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-8 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-8 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-8 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-8 div.sk-item {position: relative;z-index: 1;}#sk-container-id-8 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-8 div.sk-item::before, #sk-container-id-8 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-8 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-8 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-8 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-8 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-8 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-8 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-8 div.sk-label-container {text-align: center;}#sk-container-id-8 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-8 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-8\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-8\" type=\"checkbox\" checked><label for=\"sk-estimator-id-8\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "# 用这 8 个重要性强的特征再重新训练随机森林模型\n",
    "forest_selected = RandomForestClassifier(n_estimators=10000,\n",
    "                                random_state=0,\n",
    "                                n_jobs=-1)\n",
    "forest_selected.fit(X_train_std_selected, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "e53545a4-f677-4a3e-a474-5c5170f68c62",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 1.0\n",
      "Test accuracy: 0.9225152129817444\n"
     ]
    }
   ],
   "source": [
    "# 结果仅用这 8 个特征训练的模型，性能依旧杠杠的！\n",
    "print('Training accuracy:', forest_selected.score(X_train_std_selected, y_train))\n",
    "print('Test accuracy:', forest_selected.score(X_test_std_selected, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "7a510272-00c6-45e4-b8f3-f66a4cd2024d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_combined_std_selected = np.concatenate([X_train_std_selected, X_test_std_selected])\n",
    "y_combined = np.concatenate([y_train, y_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "6101c06a-725c-46ad-b2dd-8977ad13c934",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9157160243407707\n",
      "CPU times: user 266 ms, sys: 28 s, total: 28.3 s\n",
      "Wall time: 5min 29s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# 再交叉验证一下\n",
    "shuffle_split = ShuffleSplit(test_size=.5, train_size=.5, n_splits=5)\n",
    "model = RandomForestClassifier(n_estimators=10000, random_state=0,)\n",
    "result = cross_val_score(model, X_combined_std_selected, y_combined, cv=shuffle_split, n_jobs=-1)\n",
    "print(result.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "3c5eab2f-1171-4ef2-9f45-9b415c6d7f3f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9214604462474645\n",
      "CPU times: user 426 ms, sys: 26.5 s, total: 26.9 s\n",
      "Wall time: 8min 28s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# 再交叉验证一下\n",
    "stratified_split = StratifiedKFold(n_splits=5)\n",
    "model = RandomForestClassifier(n_estimators=10000, random_state=0,)\n",
    "result = cross_val_score(model, X_combined_std_selected, y_combined, cv=stratified_split, n_jobs=-1)\n",
    "print(result.mean())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "54db153e-2bd1-4bcb-ab65-0b6a93e4bd39",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "b086767c-1051-43d0-8f0a-6a21c03d9d1d",
   "metadata": {},
   "source": [
    "# Model: SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "5aef5aa5-25b8-4770-9cdb-f12a7c56b830",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of training set: (17255, 8)   shape of test set: (7395, 8)\n"
     ]
    }
   ],
   "source": [
    "# 接下来对 8 个特征下的训练数据，我们就在强大的 SVM 模型下调参，处理一下过拟合问题，看性能是不是还能再提升？\n",
    "print(f\"Shape of training set: {X_train_std_selected.shape}   shape of test set: {X_test_std_selected.shape}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "bdbb1834-b7c4-4f48-b00c-ef1614f8f8b2",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/6 [00:00<?, ?it/s]\n",
      "gamma=0.001:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=0.001:  17%|█▋        | 1/6 [00:20<01:40, 20.15s/it]\u001b[A\n",
      "gamma=0.001:  33%|███▎      | 2/6 [00:40<01:21, 20.28s/it]\u001b[A\n",
      "gamma=0.001:  50%|█████     | 3/6 [01:00<01:00, 20.08s/it]\u001b[A\n",
      "gamma=0.001:  67%|██████▋   | 4/6 [01:17<00:37, 18.89s/it]\u001b[A\n",
      "gamma=0.001:  83%|████████▎ | 5/6 [01:31<00:17, 17.12s/it]\u001b[A\n",
      "gamma=0.001: 100%|██████████| 6/6 [01:42<00:00, 17.11s/it]\u001b[A\n",
      " 17%|█▋        | 1/6 [01:42<08:33, 102.68s/it]\n",
      "gamma=0.01:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=0.01:  17%|█▋        | 1/6 [00:20<01:41, 20.37s/it]\u001b[A\n",
      "gamma=0.01:  33%|███▎      | 2/6 [00:40<01:20, 20.10s/it]\u001b[A\n",
      "gamma=0.01:  50%|█████     | 3/6 [00:56<00:55, 18.47s/it]\u001b[A\n",
      "gamma=0.01:  67%|██████▋   | 4/6 [01:10<00:33, 16.55s/it]\u001b[A\n",
      "gamma=0.01:  83%|████████▎ | 5/6 [01:21<00:14, 14.42s/it]\u001b[A\n",
      "gamma=0.01: 100%|██████████| 6/6 [01:30<00:00, 15.05s/it]\u001b[A\n",
      " 33%|███▎      | 2/6 [03:12<06:21, 95.40s/it] \n",
      "gamma=0.1:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=0.1:  17%|█▋        | 1/6 [00:19<01:39, 19.98s/it]\u001b[A\n",
      "gamma=0.1:  33%|███▎      | 2/6 [00:37<01:13, 18.35s/it]\u001b[A\n",
      "gamma=0.1:  50%|█████     | 3/6 [00:50<00:48, 16.17s/it]\u001b[A\n",
      "gamma=0.1:  67%|██████▋   | 4/6 [01:00<00:27, 13.81s/it]\u001b[A\n",
      "gamma=0.1:  83%|████████▎ | 5/6 [01:09<00:11, 11.77s/it]\u001b[A\n",
      "gamma=0.1: 100%|██████████| 6/6 [01:17<00:00, 12.89s/it]\u001b[A\n",
      " 50%|█████     | 3/6 [04:30<04:21, 87.16s/it]\n",
      "gamma=1:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=1:  17%|█▋        | 1/6 [00:19<01:38, 19.66s/it]\u001b[A\n",
      "gamma=1:  33%|███▎      | 2/6 [00:37<01:15, 18.78s/it]\u001b[A\n",
      "gamma=1:  50%|█████     | 3/6 [00:51<00:49, 16.50s/it]\u001b[A\n",
      "gamma=1:  67%|██████▋   | 4/6 [01:02<00:28, 14.32s/it]\u001b[A\n",
      "gamma=1:  83%|████████▎ | 5/6 [01:12<00:12, 12.73s/it]\u001b[A\n",
      "gamma=1: 100%|██████████| 6/6 [01:26<00:00, 14.36s/it]\u001b[A\n",
      " 67%|██████▋   | 4/6 [05:56<02:53, 86.76s/it]\n",
      "gamma=10:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=10:  17%|█▋        | 1/6 [00:19<01:38, 19.70s/it]\u001b[A\n",
      "gamma=10:  33%|███▎      | 2/6 [00:45<01:33, 23.43s/it]\u001b[A\n",
      "gamma=10:  50%|█████     | 3/6 [01:15<01:18, 26.12s/it]\u001b[A\n",
      "gamma=10:  67%|██████▋   | 4/6 [01:46<00:56, 28.20s/it]\u001b[A\n",
      "gamma=10:  83%|████████▎ | 5/6 [02:20<00:30, 30.19s/it]\u001b[A\n",
      "gamma=10: 100%|██████████| 6/6 [02:54<00:00, 29.04s/it]\u001b[A\n",
      " 83%|████████▎ | 5/6 [08:50<01:58, 118.31s/it]\n",
      "gamma=100:   0%|          | 0/6 [00:00<?, ?it/s]\u001b[A\n",
      "gamma=100:  17%|█▋        | 1/6 [00:25<02:08, 25.75s/it]\u001b[A\n",
      "gamma=100:  33%|███▎      | 2/6 [01:12<02:31, 37.96s/it]\u001b[A\n",
      "gamma=100:  50%|█████     | 3/6 [02:10<02:22, 47.38s/it]\u001b[A\n",
      "gamma=100:  67%|██████▋   | 4/6 [03:15<01:48, 54.38s/it]\u001b[A\n",
      "gamma=100:  83%|████████▎ | 5/6 [04:24<00:59, 59.39s/it]\u001b[A\n",
      "gamma=100: 100%|██████████| 6/6 [05:33<00:00, 55.55s/it]\u001b[A\n",
      "100%|██████████| 6/6 [14:24<00:00, 144.00s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best score:  0.8563894523326572\n",
      "best parameters:  {'C': 100, 'gamma': 0.1}\n",
      "CPU times: user 14min 21s, sys: 3.01 s, total: 14min 24s\n",
      "Wall time: 14min 24s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "best_score = 0\n",
    "\n",
    "for gamma in tqdm([0.001, 0.01, 0.1, 1, 10, 100]):\n",
    "    for C in tqdm([0.001, 0.01, 0.1, 1, 10, 100], desc=f'gamma={gamma}'):\n",
    "        # for each combination of parameters\n",
    "        # train an SVC\n",
    "        svm_selected = SVC(gamma=gamma, C=C)\n",
    "        svm_selected.fit(X_train_std_selected, y_train)\n",
    "        # evaluate the SVC on the test set \n",
    "        score = svm_selected.score(X_test_std_selected, y_test)\n",
    "        # if we got a better score, store the score and parameters\n",
    "        if score > best_score:\n",
    "            best_score = score\n",
    "            best_parameters = {'C': C, 'gamma': gamma}\n",
    "\n",
    "print(\"best score: \", best_score)\n",
    "print(\"best parameters: \", best_parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "949a46fc-65fc-42ff-ad9c-4325817efd77",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-3 {color: black;}#sk-container-id-3 pre{padding: 0;}#sk-container-id-3 div.sk-toggleable {background-color: white;}#sk-container-id-3 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-3 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-3 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-3 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-3 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-3 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-3 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-3 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-3 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-3 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-3 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-3 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-3 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-3 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-3 div.sk-item {position: relative;z-index: 1;}#sk-container-id-3 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-3 div.sk-item::before, #sk-container-id-3 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-3 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-3 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-3 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-3 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-3 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-3 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-3 div.sk-label-container {text-align: center;}#sk-container-id-3 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-3 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-3\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>SVC(C=100, gamma=0.1)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" checked><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">SVC</label><div class=\"sk-toggleable__content\"><pre>SVC(C=100, gamma=0.1)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "SVC(C=100, gamma=0.1)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "svm_selected = SVC(**best_parameters)\n",
    "svm_selected.fit(X_train_std_selected, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "c2465ec0-fce1-4702-b778-39c03584c546",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training accuracy: 0.866067806432918\n",
      "Test accuracy: 0.8563894523326572\n"
     ]
    }
   ],
   "source": [
    "print('Training accuracy:', svm_selected.score(X_train_std_selected, y_train))\n",
    "print('Test accuracy:', svm_selected.score(X_test_std_selected, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5ad70a1-67c0-4061-b5d0-7892cf98fa2a",
   "metadata": {},
   "source": [
    "# Test on O2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "49e1423f-a0aa-4827-89d4-ac57302ce6bf",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# !wget https://zenodo.org/records/5649212/files/H1_O2.csv -P data/\n",
    "# !wget https://zenodo.org/records/5649212/files/L1_O2.csv -P data/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "488834b6-f424-4461-95ec-8ce654ccf3b1",
   "metadata": {
    "tags": []
   },
   "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>event_time</th>\n",
       "      <th>ifo</th>\n",
       "      <th>peak_time</th>\n",
       "      <th>peak_time_ns</th>\n",
       "      <th>start_time</th>\n",
       "      <th>start_time_ns</th>\n",
       "      <th>duration</th>\n",
       "      <th>peak_frequency</th>\n",
       "      <th>central_freq</th>\n",
       "      <th>bandwidth</th>\n",
       "      <th>snr</th>\n",
       "      <th>amplitude</th>\n",
       "      <th>ml_label</th>\n",
       "      <th>ml_confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.165768e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1165768097</td>\n",
       "      <td>644531011</td>\n",
       "      <td>1165768097</td>\n",
       "      <td>570312023</td>\n",
       "      <td>0.08594</td>\n",
       "      <td>1096.148560</td>\n",
       "      <td>1083.121948</td>\n",
       "      <td>57.869942</td>\n",
       "      <td>7.53510</td>\n",
       "      <td>1.367110e-22</td>\n",
       "      <td>1080Lines</td>\n",
       "      <td>0.995903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.164905e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1164905171</td>\n",
       "      <td>410156011</td>\n",
       "      <td>1164905171</td>\n",
       "      <td>367187023</td>\n",
       "      <td>0.06250</td>\n",
       "      <td>1075.608765</td>\n",
       "      <td>1145.550537</td>\n",
       "      <td>268.463440</td>\n",
       "      <td>8.44360</td>\n",
       "      <td>1.611410e-22</td>\n",
       "      <td>1080Lines</td>\n",
       "      <td>0.999650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.168366e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1168365843</td>\n",
       "      <td>900389909</td>\n",
       "      <td>1168365843</td>\n",
       "      <td>527343034</td>\n",
       "      <td>0.40234</td>\n",
       "      <td>1082.743164</td>\n",
       "      <td>1065.330444</td>\n",
       "      <td>487.726471</td>\n",
       "      <td>9.64410</td>\n",
       "      <td>2.413800e-22</td>\n",
       "      <td>1080Lines</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.165462e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1165461901</td>\n",
       "      <td>597656011</td>\n",
       "      <td>1165461901</td>\n",
       "      <td>570312023</td>\n",
       "      <td>0.16406</td>\n",
       "      <td>1096.148560</td>\n",
       "      <td>1093.113159</td>\n",
       "      <td>77.852058</td>\n",
       "      <td>7.80988</td>\n",
       "      <td>1.788780e-22</td>\n",
       "      <td>1080Lines</td>\n",
       "      <td>0.996695</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.166205e+09</td>\n",
       "      <td>H1</td>\n",
       "      <td>1166204777</td>\n",
       "      <td>933593034</td>\n",
       "      <td>1166204777</td>\n",
       "      <td>735351085</td>\n",
       "      <td>0.45215</td>\n",
       "      <td>1096.148560</td>\n",
       "      <td>4480.900879</td>\n",
       "      <td>6939.164062</td>\n",
       "      <td>10.02408</td>\n",
       "      <td>2.387170e-22</td>\n",
       "      <td>1080Lines</td>\n",
       "      <td>0.999990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80001</th>\n",
       "      <td>1.168467e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1168467437</td>\n",
       "      <td>875000000</td>\n",
       "      <td>1168467437</td>\n",
       "      <td>0</td>\n",
       "      <td>1.50000</td>\n",
       "      <td>15.622150</td>\n",
       "      <td>3085.205811</td>\n",
       "      <td>6147.033203</td>\n",
       "      <td>7.52634</td>\n",
       "      <td>6.186550e-21</td>\n",
       "      <td>Low_Frequency_Lines</td>\n",
       "      <td>0.993277</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80002</th>\n",
       "      <td>1.173242e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1173242444</td>\n",
       "      <td>187500000</td>\n",
       "      <td>1173242444</td>\n",
       "      <td>0</td>\n",
       "      <td>1.25000</td>\n",
       "      <td>68.606911</td>\n",
       "      <td>65.339828</td>\n",
       "      <td>68.975723</td>\n",
       "      <td>7.94020</td>\n",
       "      <td>6.658610e-23</td>\n",
       "      <td>Power_Line</td>\n",
       "      <td>0.979834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80003</th>\n",
       "      <td>1.170262e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1170261586</td>\n",
       "      <td>336425065</td>\n",
       "      <td>1170261586</td>\n",
       "      <td>328125000</td>\n",
       "      <td>0.01562</td>\n",
       "      <td>211.476593</td>\n",
       "      <td>458.770447</td>\n",
       "      <td>787.527283</td>\n",
       "      <td>9.08770</td>\n",
       "      <td>4.665710e-23</td>\n",
       "      <td>Blip</td>\n",
       "      <td>0.998494</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80004</th>\n",
       "      <td>1.168568e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1168567671</td>\n",
       "      <td>375000000</td>\n",
       "      <td>1168567670</td>\n",
       "      <td>500000000</td>\n",
       "      <td>2.03027</td>\n",
       "      <td>35.723122</td>\n",
       "      <td>3816.616943</td>\n",
       "      <td>7588.744141</td>\n",
       "      <td>13.42574</td>\n",
       "      <td>1.623810e-22</td>\n",
       "      <td>Scattered_Light</td>\n",
       "      <td>0.999996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80005</th>\n",
       "      <td>1.168860e+09</td>\n",
       "      <td>L1</td>\n",
       "      <td>1168859818</td>\n",
       "      <td>250000000</td>\n",
       "      <td>1168859817</td>\n",
       "      <td>0</td>\n",
       "      <td>2.50000</td>\n",
       "      <td>14.005720</td>\n",
       "      <td>2342.684814</td>\n",
       "      <td>4658.571777</td>\n",
       "      <td>7.87944</td>\n",
       "      <td>8.054930e-20</td>\n",
       "      <td>Low_Frequency_Lines</td>\n",
       "      <td>0.993542</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>80006 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         event_time ifo   peak_time  peak_time_ns  start_time  start_time_ns  \\\n",
       "0      1.165768e+09  H1  1165768097     644531011  1165768097      570312023   \n",
       "1      1.164905e+09  H1  1164905171     410156011  1164905171      367187023   \n",
       "2      1.168366e+09  H1  1168365843     900389909  1168365843      527343034   \n",
       "3      1.165462e+09  H1  1165461901     597656011  1165461901      570312023   \n",
       "4      1.166205e+09  H1  1166204777     933593034  1166204777      735351085   \n",
       "...             ...  ..         ...           ...         ...            ...   \n",
       "80001  1.168467e+09  L1  1168467437     875000000  1168467437              0   \n",
       "80002  1.173242e+09  L1  1173242444     187500000  1173242444              0   \n",
       "80003  1.170262e+09  L1  1170261586     336425065  1170261586      328125000   \n",
       "80004  1.168568e+09  L1  1168567671     375000000  1168567670      500000000   \n",
       "80005  1.168860e+09  L1  1168859818     250000000  1168859817              0   \n",
       "\n",
       "       duration  peak_frequency  central_freq    bandwidth       snr  \\\n",
       "0       0.08594     1096.148560   1083.121948    57.869942   7.53510   \n",
       "1       0.06250     1075.608765   1145.550537   268.463440   8.44360   \n",
       "2       0.40234     1082.743164   1065.330444   487.726471   9.64410   \n",
       "3       0.16406     1096.148560   1093.113159    77.852058   7.80988   \n",
       "4       0.45215     1096.148560   4480.900879  6939.164062  10.02408   \n",
       "...         ...             ...           ...          ...       ...   \n",
       "80001   1.50000       15.622150   3085.205811  6147.033203   7.52634   \n",
       "80002   1.25000       68.606911     65.339828    68.975723   7.94020   \n",
       "80003   0.01562      211.476593    458.770447   787.527283   9.08770   \n",
       "80004   2.03027       35.723122   3816.616943  7588.744141  13.42574   \n",
       "80005   2.50000       14.005720   2342.684814  4658.571777   7.87944   \n",
       "\n",
       "          amplitude             ml_label  ml_confidence  \n",
       "0      1.367110e-22            1080Lines       0.995903  \n",
       "1      1.611410e-22            1080Lines       0.999650  \n",
       "2      2.413800e-22            1080Lines       1.000000  \n",
       "3      1.788780e-22            1080Lines       0.996695  \n",
       "4      2.387170e-22            1080Lines       0.999990  \n",
       "...             ...                  ...            ...  \n",
       "80001  6.186550e-21  Low_Frequency_Lines       0.993277  \n",
       "80002  6.658610e-23           Power_Line       0.979834  \n",
       "80003  4.665710e-23                 Blip       0.998494  \n",
       "80004  1.623810e-22      Scattered_Light       0.999996  \n",
       "80005  8.054930e-20  Low_Frequency_Lines       0.993542  \n",
       "\n",
       "[80006 rows x 14 columns]"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_O2 = pd.concat([\n",
    "    pd.read_csv('data/H1_O2.csv'),\n",
    "    pd.read_csv('data/L1_O2.csv')\n",
    "])\n",
    "\n",
    "cols = ['event_time', 'ifo', 'peak_time', 'peak_time_ns', 'start_time', 'start_time_ns',\n",
    "        'duration', 'peak_frequency', 'central_freq', 'bandwidth', 'snr', \n",
    "        'amplitude', 'ml_label', 'ml_confidence']\n",
    "df_O2 = df_O2[cols]\n",
    "df_O2 = df_O2[df_O2['ml_confidence'] >= 0.95]\n",
    "df_O2 = df_O2.reset_index().drop('index', axis=1)\n",
    "df_O2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "2103eb17-3344-43c4-be85-996cde18b147",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'1080Lines', 'No_Glitch', 'None_of_the_Above'}"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set(pd.unique(df_O2.ml_label)) - set(target_labels)  # 除了我们想去掉的，没有额外的新glitch类型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "6f2f7d73-ea71-48f1-9f39-e9f02e957f00",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(50810, 14)"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_labels_O2 = [label for label in pd.unique(df_O2.ml_label) if label not in ['None_of_the_Above', '1080Lines', 'No_Glitch']]\n",
    "df_O2 = pd.concat([df_O2[df_O2.ml_label == label]\n",
    "                         for label in target_labels_O2])\n",
    "df_O2.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "fdca335d-02f0-4018-a326-2a7acc2c8400",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ml_label\n",
       "Low_Frequency_Lines    16714\n",
       "Blip                   11122\n",
       "Scattered_Light        10132\n",
       "Koi_Fish                3369\n",
       "Low_Frequency_Burst     2249\n",
       "Whistle                 1551\n",
       "Extremely_Loud          1537\n",
       "Power_Line              1433\n",
       "Scratchy                1094\n",
       "Repeating_Blips          969\n",
       "Tomte                    188\n",
       "1400Ripples              169\n",
       "Violin_Mode              159\n",
       "Light_Modulation          54\n",
       "Wandering_Line            25\n",
       "Helix                     20\n",
       "Paired_Doves              14\n",
       "Air_Compressor            10\n",
       "Chirp                      1\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 171,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_O2.ml_label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "id": "cb200fc5-3fa4-4dc2-9cdc-5051b26bc10e",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['peak_frequency', 'amplitude', 'duration', 'snr',\n",
       "       'start_time_fullgps', 'peak_time_fullgps', 'central_freq',\n",
       "       'bandwidth'], dtype='<U22')"
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 只要那 8 个特征即可\n",
    "np.array(feat_labels)[indices[:8]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "id": "7a3c4a2d-21bc-4ad8-910a-331434a0b29d",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 35.9 s, sys: 992 ms, total: 36.9 s\n",
      "Wall time: 35.6 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# 创建 ID\n",
    "df_O2.loc[:, 'id'] = df_O2.event_time.map(lambda x: from_gps(x).strftime('g%y%m%d_%H%M%S+')+str(int(from_gps(x).microsecond/1000)).zfill(3))\n",
    "# 对时间特征处理，取出关键信息\n",
    "df_O2.loc[:, 'peak_time_fullgps'] = df_O2.peak_time + 1e-9*df_O2.peak_time_ns\n",
    "df_O2.loc[:, 'start_time_fullgps'] = df_O2.start_time + 1e-9*df_O2.start_time_ns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "id": "3ffc8fac-2dc7-4c96-ac9d-9b93d327b6b6",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# 对标签类别进行编码\n",
    "# Note: 要用上 O1 数据里 fit 出来的 class_le\n",
    "df_O2.loc[:, 'ml_label'] = class_le.fit_transform(df_O2['ml_label'].values)\n",
    "df_O2.loc[:, 'label'] = df_O2.ml_label.astype('int')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "id": "a9b08211-a9c8-473e-b30e-88ea9cc0306f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 50810 entries, 7 to 65930\n",
      "Data columns (total 12 columns):\n",
      " #   Column              Non-Null Count  Dtype  \n",
      "---  ------              --------------  -----  \n",
      " 0   duration            50810 non-null  float64\n",
      " 1   peak_frequency      50810 non-null  float64\n",
      " 2   central_freq        50810 non-null  float64\n",
      " 3   bandwidth           50810 non-null  float64\n",
      " 4   snr                 50810 non-null  float64\n",
      " 5   amplitude           50810 non-null  float64\n",
      " 6   id                  50810 non-null  object \n",
      " 7   peak_time_fullgps   50810 non-null  float64\n",
      " 8   start_time_fullgps  50810 non-null  float64\n",
      " 9   label               50810 non-null  int64  \n",
      " 10  ifo_H1              50810 non-null  float64\n",
      " 11  ifo_L1              50810 non-null  float64\n",
      "dtypes: float64(10), int64(1), object(1)\n",
      "memory usage: 5.0+ MB\n"
     ]
    }
   ],
   "source": [
    "# 去掉不需要的特征列；把 ifo 特征进行独热向量编码\n",
    "df_O2 = df_O2.drop(['event_time', 'peak_time', 'peak_time_ns', 'start_time', 'start_time_ns', 'ml_confidence', 'ml_label'], axis=1)\n",
    "df_O2 = pd.concat([df_O2, pd.get_dummies(df_O2[['ifo']], dtype='float')], axis=1).drop('ifo', axis=1)\n",
    "df_O2.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "9254ef3b-a7cf-4415-b43b-65c3fbd90898",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_O2, y_O2 = df_O2[[col for col in df.columns.tolist() if col in np.array(feat_labels)[indices[:8]]]], df_O2.label.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "2b10c953-cfc7-4d26-8353-a09eb566d841",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# 标准化\n",
    "stdsc = StandardScaler()\n",
    "X_std_O2 = stdsc.fit_transform(X_O2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1738de2-0495-4ff0-a362-1595edd21dc1",
   "metadata": {},
   "source": [
    "- 果然：”理想很美好，现实很骨感“"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "1001dae3-cb9d-44d0-9d54-a02354e7c871",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "O2 accuracy: 0.5638260185002952\n"
     ]
    }
   ],
   "source": [
    "# 随机森林\n",
    "print('O2 accuracy:', forest_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "6f02c88c-4f32-4649-a568-9a749a166503",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "O2 accuracy: 0.605609132060618\n"
     ]
    }
   ],
   "source": [
    "# SVM\n",
    "print('O2 accuracy:', svm_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "1e104bde-08c5-420f-af23-e61dd1596fcf",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((35567, 8), (35567,), (15243, 8), (15243,))"
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 切分训练集和测试集\n",
    "X_train_O2, X_test_O2, y_train_O2, y_test_O2 = \\\n",
    "    train_test_split(X_O2, y_O2, test_size=0.3, random_state=0)\n",
    "X_train_O2.shape, y_train_O2.shape, X_test_O2.shape, y_test_O2.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ffd28256-a7ce-4eed-8805-b33949b6cb88",
   "metadata": {},
   "source": [
    "- 可能是数据分布差异太大，稍微调整一下O1 和O2 的gps特征，然后分别在 O1 和 O2 上重新训练一波"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "949ace9c-2d92-45e5-9e6b-d0ae560848c0",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'amplitude',\n",
       " 'bandwidth',\n",
       " 'central_freq',\n",
       " 'duration',\n",
       " 'peak_frequency',\n",
       " 'peak_start_distance',\n",
       " 'snr'}"
      ]
     },
     "execution_count": 179,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "modified_columns = (set(np.array(feat_labels)[indices[:8]]) | set(['peak_start_distance'])) - set(['peak_time_fullgps', 'start_time_fullgps'])\n",
    "modified_columns\n",
    "# 7个特征来搞事情"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "c297e5e3-0882-4e84-8e35-9d79b4470c4b",
   "metadata": {
    "tags": []
   },
   "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>duration</th>\n",
       "      <th>peak_frequency</th>\n",
       "      <th>central_freq</th>\n",
       "      <th>bandwidth</th>\n",
       "      <th>snr</th>\n",
       "      <th>amplitude</th>\n",
       "      <th>peak_start_distance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3.12500</td>\n",
       "      <td>23.161011</td>\n",
       "      <td>3083.450439</td>\n",
       "      <td>6125.699219</td>\n",
       "      <td>12.31222</td>\n",
       "      <td>9.112720e-22</td>\n",
       "      <td>1.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1.12500</td>\n",
       "      <td>53.730080</td>\n",
       "      <td>42.504478</td>\n",
       "      <td>28.540449</td>\n",
       "      <td>8.83714</td>\n",
       "      <td>7.523970e-23</td>\n",
       "      <td>0.437500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1.75000</td>\n",
       "      <td>24.363270</td>\n",
       "      <td>3203.047363</td>\n",
       "      <td>6367.409668</td>\n",
       "      <td>11.79201</td>\n",
       "      <td>6.935220e-22</td>\n",
       "      <td>0.937500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1.12500</td>\n",
       "      <td>33.173901</td>\n",
       "      <td>43.625809</td>\n",
       "      <td>42.762009</td>\n",
       "      <td>9.65102</td>\n",
       "      <td>2.106870e-22</td>\n",
       "      <td>0.750000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>3.37500</td>\n",
       "      <td>50.289051</td>\n",
       "      <td>2961.414307</td>\n",
       "      <td>5901.543945</td>\n",
       "      <td>20.92308</td>\n",
       "      <td>2.076380e-22</td>\n",
       "      <td>1.187500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25807</th>\n",
       "      <td>0.06641</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>2087.095215</td>\n",
       "      <td>1224.458984</td>\n",
       "      <td>10.76469</td>\n",
       "      <td>4.844820e-22</td>\n",
       "      <td>0.021484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25821</th>\n",
       "      <td>0.04688</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>1756.158569</td>\n",
       "      <td>562.585754</td>\n",
       "      <td>10.50213</td>\n",
       "      <td>5.606110e-22</td>\n",
       "      <td>0.025391</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25823</th>\n",
       "      <td>0.03955</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>1765.836426</td>\n",
       "      <td>581.941406</td>\n",
       "      <td>9.30373</td>\n",
       "      <td>4.117320e-22</td>\n",
       "      <td>0.014160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25832</th>\n",
       "      <td>0.04688</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>1518.905151</td>\n",
       "      <td>81.143478</td>\n",
       "      <td>9.53593</td>\n",
       "      <td>4.382050e-22</td>\n",
       "      <td>0.017578</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25845</th>\n",
       "      <td>0.10547</td>\n",
       "      <td>1527.022095</td>\n",
       "      <td>1802.779419</td>\n",
       "      <td>1581.367554</td>\n",
       "      <td>12.79415</td>\n",
       "      <td>5.758980e-22</td>\n",
       "      <td>0.076172</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>24650 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       duration  peak_frequency  central_freq    bandwidth       snr  \\\n",
       "0       3.12500       23.161011   3083.450439  6125.699219  12.31222   \n",
       "5       1.12500       53.730080     42.504478    28.540449   8.83714   \n",
       "10      1.75000       24.363270   3203.047363  6367.409668  11.79201   \n",
       "11      1.12500       33.173901     43.625809    42.762009   9.65102   \n",
       "13      3.37500       50.289051   2961.414307  5901.543945  20.92308   \n",
       "...         ...             ...           ...          ...       ...   \n",
       "25807   0.06641     1527.022095   2087.095215  1224.458984  10.76469   \n",
       "25821   0.04688     1527.022095   1756.158569   562.585754  10.50213   \n",
       "25823   0.03955     1527.022095   1765.836426   581.941406   9.30373   \n",
       "25832   0.04688     1527.022095   1518.905151    81.143478   9.53593   \n",
       "25845   0.10547     1527.022095   1802.779419  1581.367554  12.79415   \n",
       "\n",
       "          amplitude  peak_start_distance  \n",
       "0      9.112720e-22             1.250000  \n",
       "5      7.523970e-23             0.437500  \n",
       "10     6.935220e-22             0.937500  \n",
       "11     2.106870e-22             0.750000  \n",
       "13     2.076380e-22             1.187500  \n",
       "...             ...                  ...  \n",
       "25807  4.844820e-22             0.021484  \n",
       "25821  5.606110e-22             0.025391  \n",
       "25823  4.117320e-22             0.014160  \n",
       "25832  4.382050e-22             0.017578  \n",
       "25845  5.758980e-22             0.076172  \n",
       "\n",
       "[24650 rows x 7 columns]"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "temp = df[[col for col in df.columns.tolist()\n",
    "              if col in np.array(feat_labels)[indices[:8]]]]\n",
    "df.loc[:,'peak_start_distance'] = temp.peak_time_fullgps - temp.start_time_fullgps\n",
    "\n",
    "df_modified = df[[col for col in df.columns.tolist()\n",
    "                  if col in modified_columns]]\n",
    "df_modified"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "id": "8ba91f7e-37c8-4fc3-b29e-6fffd25e99e5",
   "metadata": {
    "tags": []
   },
   "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>duration</th>\n",
       "      <th>peak_frequency</th>\n",
       "      <th>central_freq</th>\n",
       "      <th>bandwidth</th>\n",
       "      <th>snr</th>\n",
       "      <th>amplitude</th>\n",
       "      <th>peak_start_distance</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.25000</td>\n",
       "      <td>211.476593</td>\n",
       "      <td>365.147858</td>\n",
       "      <td>645.632141</td>\n",
       "      <td>27.556669</td>\n",
       "      <td>1.619030e-22</td>\n",
       "      <td>0.112793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.30859</td>\n",
       "      <td>170.653900</td>\n",
       "      <td>708.072815</td>\n",
       "      <td>1347.825195</td>\n",
       "      <td>20.646790</td>\n",
       "      <td>1.161990e-22</td>\n",
       "      <td>0.176758</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>6.00000</td>\n",
       "      <td>402.439423</td>\n",
       "      <td>3000.409668</td>\n",
       "      <td>5992.298828</td>\n",
       "      <td>9.500670</td>\n",
       "      <td>7.523580e-23</td>\n",
       "      <td>2.400634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.07520</td>\n",
       "      <td>211.476593</td>\n",
       "      <td>2683.394531</td>\n",
       "      <td>5205.674805</td>\n",
       "      <td>9.551630</td>\n",
       "      <td>5.396630e-23</td>\n",
       "      <td>0.008301</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.08594</td>\n",
       "      <td>111.128151</td>\n",
       "      <td>178.368515</td>\n",
       "      <td>226.723389</td>\n",
       "      <td>8.450540</td>\n",
       "      <td>5.424690e-23</td>\n",
       "      <td>0.073242</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55017</th>\n",
       "      <td>0.46289</td>\n",
       "      <td>48.249069</td>\n",
       "      <td>1213.019775</td>\n",
       "      <td>2357.718994</td>\n",
       "      <td>7.716150</td>\n",
       "      <td>7.590440e-23</td>\n",
       "      <td>0.330078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56670</th>\n",
       "      <td>0.50000</td>\n",
       "      <td>48.836180</td>\n",
       "      <td>49.305050</td>\n",
       "      <td>14.360080</td>\n",
       "      <td>8.428330</td>\n",
       "      <td>7.086710e-23</td>\n",
       "      <td>0.218750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60916</th>\n",
       "      <td>0.37500</td>\n",
       "      <td>47.974152</td>\n",
       "      <td>47.716240</td>\n",
       "      <td>11.182450</td>\n",
       "      <td>7.534050</td>\n",
       "      <td>6.698290e-23</td>\n",
       "      <td>0.156250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62350</th>\n",
       "      <td>0.50000</td>\n",
       "      <td>48.836182</td>\n",
       "      <td>3424.374756</td>\n",
       "      <td>6755.920410</td>\n",
       "      <td>7.964590</td>\n",
       "      <td>7.925360e-23</td>\n",
       "      <td>0.343750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>65930</th>\n",
       "      <td>0.10938</td>\n",
       "      <td>90.541160</td>\n",
       "      <td>144.194260</td>\n",
       "      <td>204.138490</td>\n",
       "      <td>8.202260</td>\n",
       "      <td>4.164060e-23</td>\n",
       "      <td>0.072265</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>50810 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       duration  peak_frequency  central_freq    bandwidth        snr  \\\n",
       "7       0.25000      211.476593    365.147858   645.632141  27.556669   \n",
       "9       0.30859      170.653900    708.072815  1347.825195  20.646790   \n",
       "10      6.00000      402.439423   3000.409668  5992.298828   9.500670   \n",
       "13      0.07520      211.476593   2683.394531  5205.674805   9.551630   \n",
       "15      0.08594      111.128151    178.368515   226.723389   8.450540   \n",
       "...         ...             ...           ...          ...        ...   \n",
       "55017   0.46289       48.249069   1213.019775  2357.718994   7.716150   \n",
       "56670   0.50000       48.836180     49.305050    14.360080   8.428330   \n",
       "60916   0.37500       47.974152     47.716240    11.182450   7.534050   \n",
       "62350   0.50000       48.836182   3424.374756  6755.920410   7.964590   \n",
       "65930   0.10938       90.541160    144.194260   204.138490   8.202260   \n",
       "\n",
       "          amplitude  peak_start_distance  \n",
       "7      1.619030e-22             0.112793  \n",
       "9      1.161990e-22             0.176758  \n",
       "10     7.523580e-23             2.400634  \n",
       "13     5.396630e-23             0.008301  \n",
       "15     5.424690e-23             0.073242  \n",
       "...             ...                  ...  \n",
       "55017  7.590440e-23             0.330078  \n",
       "56670  7.086710e-23             0.218750  \n",
       "60916  6.698290e-23             0.156250  \n",
       "62350  7.925360e-23             0.343750  \n",
       "65930  4.164060e-23             0.072265  \n",
       "\n",
       "[50810 rows x 7 columns]"
      ]
     },
     "execution_count": 181,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "temp = df_O2[[col for col in df.columns.tolist()\n",
    "              if col in np.array(feat_labels)[indices[:8]]]]\n",
    "df_O2.loc[:,'peak_start_distance'] = temp.peak_time_fullgps - temp.start_time_fullgps\n",
    "df_O2_modified = df_O2[[col for col in df_O2.columns.tolist()\n",
    "                        if col in modified_columns]]\n",
    "df_O2_modified"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "id": "00b109d8-1010-455e-af45-9c7ed8264d01",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_O1, y_O1 = df_modified.values, df.label.values\n",
    "X_O2, y_O2 = df_O2_modified.values, df_O2.label.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "id": "86b16797-ef3f-4e4e-9ca0-348ba0383fb1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# 标准化\n",
    "stdsc = StandardScaler()\n",
    "stdsc.fit(np.concatenate([X_O1, X_O2]))  # 大家一起标准化\n",
    "X_std_O1 = stdsc.transform(X_O1)\n",
    "X_std_O2 = stdsc.transform(X_O2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "id": "2d5e4e7d-e801-4d34-8954-09969cc224e8",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 52min 3s, sys: 46.2 s, total: 52min 49s\n",
      "Wall time: 1min 54s\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-9 {color: black;}#sk-container-id-9 pre{padding: 0;}#sk-container-id-9 div.sk-toggleable {background-color: white;}#sk-container-id-9 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-9 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-9 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-9 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-9 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-9 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-9 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-9 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-9 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-9 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-9 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-9 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-9 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-9 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-9 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-9 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-9 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-9 div.sk-item {position: relative;z-index: 1;}#sk-container-id-9 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-9 div.sk-item::before, #sk-container-id-9 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-9 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-9 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-9 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-9 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-9 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-9 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-9 div.sk-label-container {text-align: center;}#sk-container-id-9 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-9 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-9\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-9\" type=\"checkbox\" checked><label for=\"sk-estimator-id-9\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(n_estimators=10000, n_jobs=-1, random_state=0)"
      ]
     },
     "execution_count": 185,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "# 用这 8 个重要性强的特征再重新训练随机森林模型\n",
    "forest_O1_selected = RandomForestClassifier(n_estimators=10000,\n",
    "                                random_state=0,\n",
    "                                n_jobs=-1)\n",
    "forest_O1_selected.fit(X_std_O1, y_O1)\n",
    "\n",
    "forest_O2_selected = RandomForestClassifier(n_estimators=10000,\n",
    "                                random_state=0,\n",
    "                                n_jobs=-1)\n",
    "forest_O2_selected.fit(X_std_O2, y_O2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "id": "e80368c3-2c6b-4a36-b15f-a6559e89fb64",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test on O1 accuracy using O1 RF model: 1.0\n",
      "Test on O2 accuracy using O2 RF model: 1.0\n"
     ]
    }
   ],
   "source": [
    "# 随机森林\n",
    "print('Test on O1 accuracy using O1 RF model:', forest_O1_selected.score(X_std_O1, y_O1))\n",
    "print('Test on O2 accuracy using O2 RF model:', forest_O2_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "id": "a8a631c2-8b87-4d5a-823b-8112fa73eb0f",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test on O1 accuracy using O2 RF model: 0.7133468559837728\n",
      "Test on O2 accuracy using O1 RF model: 0.8317850816768353\n"
     ]
    }
   ],
   "source": [
    "# 随机森林\n",
    "print('Test on O1 accuracy using O2 RF model:', forest_O2_selected.score(X_std_O1, y_O1))\n",
    "print('Test on O2 accuracy using O1 RF model:', forest_O1_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "0cd1abf4-18f9-4661-b9a1-72914f021afc",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 40.2 s, sys: 598 ms, total: 40.8 s\n",
      "Wall time: 40.8 s\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-10 {color: black;}#sk-container-id-10 pre{padding: 0;}#sk-container-id-10 div.sk-toggleable {background-color: white;}#sk-container-id-10 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-10 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-10 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-10 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-10 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-10 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-10 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-10 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-10 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-10 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-10 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-10 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-10 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-10 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-10 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-10 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-10 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-10 div.sk-item {position: relative;z-index: 1;}#sk-container-id-10 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-10 div.sk-item::before, #sk-container-id-10 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-10 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-10 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-10 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-10 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-10 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-10 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-10 div.sk-label-container {text-align: center;}#sk-container-id-10 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-10 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-10\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>SVC(C=100, gamma=0.1)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-10\" type=\"checkbox\" checked><label for=\"sk-estimator-id-10\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">SVC</label><div class=\"sk-toggleable__content\"><pre>SVC(C=100, gamma=0.1)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "SVC(C=100, gamma=0.1)"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "# 用这 8 个重要性强的特征再重新训练SVM模型\n",
    "svm_O1_selected = SVC(**best_parameters)\n",
    "svm_O1_selected.fit(X_std_O1, y_O1)\n",
    "\n",
    "svm_O2_selected = SVC(**best_parameters)\n",
    "svm_O2_selected.fit(X_std_O2, y_O2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "id": "f81f813e-02d8-49cc-8c23-2b02728f10b9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test on O1 accuracy using O1 SVM model: 0.8489655172413794\n",
      "Test on O2 accuracy using O2 SVM model: 0.9009643770911238\n"
     ]
    }
   ],
   "source": [
    "# SVM\n",
    "print('Test on O1 accuracy using O1 SVM model:', svm_O1_selected.score(X_std_O1, y_O1))\n",
    "print('Test on O2 accuracy using O2 SVM model:', svm_O2_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "id": "ba58eb04-baff-487a-bca1-741c5d77b039",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test on O1 accuracy using O2 SVM model: 0.6979310344827586\n",
      "Test on O2 accuracy using O1 SVM model: 0.8032670734107459\n"
     ]
    }
   ],
   "source": [
    "# SVM\n",
    "print('Test on O1 accuracy using O2 SVM model:', svm_O2_selected.score(X_std_O1, y_O1))\n",
    "print('Test on O2 accuracy using O1 SVM model:', svm_O1_selected.score(X_std_O2, y_O2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "694b9d93-0041-4edb-888a-eaa4f744d4f7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "ICTP-AP",
   "language": "python",
   "name": "ictp-ap"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}