analyst & create data.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "667dd2a0-51db-4530-aa7d-ae1e85ac0c45",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torchaudio\n",
    "import glob\n",
    "from tqdm.auto import tqdm\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import os\n",
    "import json\n",
    "import numpy as np\n",
    "import ftfy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f024a784-08da-4f1b-9cb5-45a44956802e",
   "metadata": {},
   "outputs": [],
   "source": [
    "root = 'md_d_vlsp_2023/data'\n",
    "\n",
    "paths = {\n",
    "    'train': {\n",
    "        'phone_csv': root + '/metadata/train_phones.csv',\n",
    "        'word_csv': root + '/metadata/train.csv',\n",
    "        'root': root\n",
    "    },\n",
    "    'public_test': {\n",
    "        'phone_csv': root + '/VMD-VLSP23-public-test/meta_data/public_test_phones.csv',\n",
    "        'word_csv': root + '/VMD-VLSP23-public-test/meta_data/public_test.csv',\n",
    "        'root': root + '/VMD-VLSP23-public-test'\n",
    "    },\n",
    "    'private_test': {\n",
    "        'phone_csv': root + '/VMD-VLSP23-private-test/metadata/private_test_submission.csv',\n",
    "        'root': root + '/VMD-VLSP23-private-test'\n",
    "    }\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0e855163-3cfb-407e-a1c5-a2806130b1e9",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3384"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_audios = glob.glob(paths['train']['root'] + '/audio_data/train/*.wav')\n",
    "len(train_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5f2d66e5-3de6-448f-a9af-6907abf514cd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "718"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "public_test_audios = glob.glob(paths['public_test']['root'] + '/audio_data/public_test/*.wav')\n",
    "len(public_test_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "10e01a0f-05cc-4a89-bb67-a8c991f8feb1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1182"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "private_test_audios = glob.glob(paths['private_test']['root'] + '/audio_data/private_test/*.wav')\n",
    "len(private_test_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "7f97fee4-a444-4251-8f7a-8ed6f66c68d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c989d54205b64f0d9fa4aa7a0690693b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1182 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in tqdm(private_test_audios):\n",
    "    fix_path = ftfy.fix_text(i)\n",
    "    wav, sr = torchaudio.load(i)\n",
    "    torchaudio.save(fix_path, wav, sr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c4edb883-9483-4bda-865d-a7d6d89cae33",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_info(name, paths):\n",
    "    # training\n",
    "    lengths = []\n",
    "    for i in tqdm(paths):\n",
    "        wav, sr = torchaudio.load(i)\n",
    "        lengths.append(wav.size(1) / sr)\n",
    "\n",
    "    return {\n",
    "        'name': name,\n",
    "        'min': round(np.min(lengths), 2),\n",
    "        'max': round(np.max(lengths), 2),\n",
    "        'mean': np.mean(lengths),\n",
    "        'median': np.median(lengths),\n",
    "        'quantile_96': np.quantile(lengths, 0.96),\n",
    "        'total': round(np.sum(lengths) / 60 / 60, 2)\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "664cfef3-c794-45c1-a277-c7e7149dab55",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1c9456976dcc4708a0c4ef4261cc2647",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/3384 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'name': 'train',\n",
       " 'min': 0.54,\n",
       " 'max': 19.81,\n",
       " 'mean': 4.16186628250591,\n",
       " 'median': 3.3929375,\n",
       " 'quantile_96': 10.266742499999994,\n",
       " 'total': 3.91}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_info('train', train_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b64bf74b-08b9-4576-87e5-62ce5bf4f27a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "60b64c43da7440868fa230337d9d0d8b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/718 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'name': 'public_test',\n",
       " 'min': 1.2,\n",
       " 'max': 18.46,\n",
       " 'mean': 3.2566323119777154,\n",
       " 'median': 2.5915,\n",
       " 'quantile_96': 7.12761499999999,\n",
       " 'total': 0.65}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_info('public_test', public_test_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1c843e83-fbca-4e24-9dda-a102b1592ce4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7ec679d36ddc48bdad7ad43a01e18623",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/1182 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{'name': 'private_test',\n",
       " 'min': 1.11,\n",
       " 'max': 18.45,\n",
       " 'mean': 2.841986992385787,\n",
       " 'median': 2.24,\n",
       " 'quantile_96': 6.575875,\n",
       " 'total': 0.93}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "get_info('private_test', private_test_audios)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "7cbe3d1a-1a2d-46f6-8ad2-a0e7bc9de5a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "tonal_symbols = {\n",
    "    6: 'ạặậẹệọộợịụỵ',\n",
    "    5: 'ãẵẫẽễõỗỡĩũỹ',\n",
    "    4: 'ảẳẩẻểỏổởỉủỷ',\n",
    "    3: 'áắấéếóốớíúý',\n",
    "    2: 'àằầèềòồờìùỳ',\n",
    "}\n",
    "\n",
    "def get_tonal(word):\n",
    "    for k, symbols in tonal_symbols.items():\n",
    "        for v in symbols:\n",
    "            if v in word:\n",
    "                return k\n",
    "    return 1\n",
    "\n",
    "def get_list_tonal(text):\n",
    "    return [get_tonal(w) for w in text.split()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "926ec734-a515-47b9-857d-e197075767d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_data(phone_csv, word_csv, root, have_text=False):\n",
    "    df = pd.read_csv(phone_csv)\n",
    "    word_df = pd.read_csv(word_csv)\n",
    "\n",
    "    sub_df = df[['id', 'path', 'canonical', 'transcript']]\n",
    "    sub_df['text'] = word_df['transcript']\n",
    "    \n",
    "    sub_df['tonal'] = sub_df['text'].map(get_list_tonal)\n",
    "\n",
    "    if have_text:\n",
    "        shuffled_df = shuffled_df.drop('text', axis=1)\n",
    "\n",
    "    sub_df['path'] = sub_df.path.map(lambda x: root + '/' + x)\n",
    "\n",
    "    sub_data = list(sub_df.T.to_dict().values())\n",
    "    return sub_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "72e2ce4b-e4b7-472f-886f-15976c999981",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data = create_data(**paths['train'])\n",
    "public_test_data = create_data(**paths['public_test'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "113e847e-cb2a-4d2b-869c-d7a579ebc706",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(717, 718)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_public_test_data = []\n",
    "for i in public_test_data:\n",
    "    if 'bia-dat' not in i['path']:\n",
    "        new_public_test_data.append(i)\n",
    "len(new_public_test_data), len(public_test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "048212b3-eaed-4c29-815d-770d954c9dd1",
   "metadata": {},
   "outputs": [],
   "source": [
    "public_test_data = new_public_test_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "b4889207-f9a5-4156-874d-7ae8f954e414",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'id': 'bia-dat_1618738038209',\n",
       " 'path': '/home/stud_vantuan/projects/md_d_vlsp_2023/data/audio_data/train/bia-dat_1618738038209.wav',\n",
       " 'canonical': 'ɓ iə-5 $ ɗ a-0 tz',\n",
       " 'transcript': 'ɓ iə-4 $ ɗ a-0 tz',\n",
       " 'text': 'bía đăt',\n",
       " 'tonal': [3, 1]}"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a3e83287-0c8b-4fa7-a96e-2f647617b4b4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'id': 'vao-nui_1618754903173',\n",
       "  'path': '/home/stud_vantuan/projects/md_d_vlsp_2023/data/VMD-VLSP23-private-test/audio_data/private_test/vao-nui_1618754903173.wav',\n",
       "  'canonical': 'v aː-1 uz $ n w i-4',\n",
       "  'transcript': 'v aː-1 uz $ n w i-3'},\n",
       " {'id': 'vao-nui_1618754909868',\n",
       "  'path': '/home/stud_vantuan/projects/md_d_vlsp_2023/data/VMD-VLSP23-private-test/audio_data/private_test/vao-nui_1618754909868.wav',\n",
       "  'canonical': 'v aː-1 uz $ n w i-4',\n",
       "  'transcript': 'v aː-1 uz $ n w i-4'}]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(paths['private_test']['phone_csv'])\n",
    "df['path'] = df['path'].map(lambda x: ftfy.fix_text(paths['private_test']['root'] + '/' + x))\n",
    "private_test_data = list(df[['id', 'path', 'canonical', 'transcript']].T.to_dict().values())\n",
    "private_test_data[:2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "366b7291-26ff-4a05-8bb6-6dc80a992c05",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3384, 717, 880)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_data), len(public_test_data), len(private_test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "1314aff7-7eab-4324-8403-3282fec6cabf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'id': 'THA_Nu_6_S00042_201',\n",
       " 'path': '/home/stud_vantuan/projects/md_d_vlsp_2023/data/audio_data/train/THA_Nu_6_S00042_201.wav',\n",
       " 'canonical': 'aː-0 $ m aː-4 $ ɗ aː-2 $ k ɔ-4 $ ɓ ɛ-4 $ l e-0 $ k ɔ-4 $ h aː-1 $ l aː-1 $ k w aː-1 $ k w i-4',\n",
       " 'transcript': 'aː-0 $ m aː-4 $ ɗ aː-2 $ k ɔ-4 $ ɓ ɛ-4 $ l e-0 $ k ɔ-4 $ h aː-1 $ l aː-1 $ k w aː-1 $ k w i-4',\n",
       " 'text': 'à má đã có bé lê có hà là quà quý',\n",
       " 'tonal': [2, 3, 5, 3, 3, 1, 3, 2, 2, 2, 3]}"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "b22ca81c-702c-4f19-a31b-0c50f70cb0f2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'id': 'CPT_Nu_6_S0004_98',\n",
       " 'path': '/home/stud_vantuan/projects/md_d_vlsp_2023/data/VMD-VLSP23-public-test/audio_data/public_test/CPT_Nu_6_S0004_98.wav',\n",
       " 'canonical': 'l aː-1 $ v uə-0 $ k uə-3 $ ɓ iə-3 nz',\n",
       " 'transcript': 'l aː-1 $ v uə-0 $ k uə-3 $ ɓ iə-3 nz',\n",
       " 'text': 'là vua của biển',\n",
       " 'tonal': [2, 1, 4, 4]}"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "public_test_data[600]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "6e267e21-31df-4fd1-8d20-f600b9582f72",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "120"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(set([j for i in [' '.join(k['canonical'].split(' $ ')).split(' ') for k in private_test_data] for j in i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "a396a194-18d3-4905-b200-d103310861e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_cnt = [(i['canonical'].count('$') > 1) for i in train_data]\n",
    "public_test_cnt = [(i['canonical'].count('$') > 1) for i in public_test_data]\n",
    "private_test_cnt = [(i['canonical'].count('$') > 1) for i in private_test_data]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "6b7ccf81-f3d4-4b98-ba2a-52da438699be",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "91.52"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "round(sum(train_cnt) / len(train_cnt) * 100, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "4daf609d-f601-4f7f-9de9-31a394c0dd89",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "65.69"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "round(sum(public_test_cnt) / len(public_test_cnt) * 100, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "61a92184-bd6c-4b65-a52e-fc3b0c0406e1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "74.09"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "round(sum(private_test_cnt) / len(private_test_cnt) * 100, 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "d413e3ec-c250-41f2-a006-1ff132bac625",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "save_path = '/home/stud_vantuan/projects/md_d_vlsp_2023/data/splitted_data'\n",
    "\n",
    "with open(os.path.join(save_path, 'train.json'), 'w', encoding='utf-8') as f:\n",
    "    json.dump(train_data, f, ensure_ascii=False)\n",
    "\n",
    "with open(os.path.join(save_path, 'public_test.json'), 'w', encoding='utf-8') as f:\n",
    "    json.dump(public_test_data, f, ensure_ascii=False)\n",
    "\n",
    "with open(os.path.join(save_path, 'private_test.json'), 'w', encoding='utf-8') as f:\n",
    "    json.dump(private_test_data, f, ensure_ascii=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "ed4be8f0-d888-471d-913e-7f415f869e0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def cal_diff(i):\n",
    "    return len(set(i['transcript']) ^ set(i['canonical']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "6d5d733f-25b3-4217-a97d-a2573d955879",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_diff = [cal_diff(i) for i in train_data]\n",
    "public_test_diff = [cal_diff(i) for i in public_test_data]\n",
    "private_test_diff = [cal_diff(i) for i in private_test_data]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "3dbe68b3-2d62-4745-9c65-d996fee7469f",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "010ec521-5f9b-41e0-a06a-f9fd32d5c882",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>diff</th>\n",
       "      <th>kind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>train</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4976</th>\n",
       "      <td>3</td>\n",
       "      <td>private_test</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4977</th>\n",
       "      <td>0</td>\n",
       "      <td>private_test</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4978</th>\n",
       "      <td>0</td>\n",
       "      <td>private_test</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4979</th>\n",
       "      <td>5</td>\n",
       "      <td>private_test</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4980</th>\n",
       "      <td>3</td>\n",
       "      <td>private_test</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4981 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      diff          kind\n",
       "0        0         train\n",
       "1        0         train\n",
       "2        0         train\n",
       "3        0         train\n",
       "4        0         train\n",
       "...    ...           ...\n",
       "4976     3  private_test\n",
       "4977     0  private_test\n",
       "4978     0  private_test\n",
       "4979     5  private_test\n",
       "4980     3  private_test\n",
       "\n",
       "[4981 rows x 2 columns]"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.DataFrame({\n",
    "    'diff': train_diff + public_test_diff + private_test_diff,\n",
    "    'kind': len(train_diff) * ['train'] + len(public_test_diff) * ['public_test'] + len(private_test_diff) * ['private_test']\n",
    "})\n",
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "3f28cb05-790f-4a1b-8108-3026a7dca89b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create(kind):\n",
    "    a = data[data['kind'] == kind][['diff']].value_counts()\n",
    "    y = list(a)\n",
    "    x = list(a.reset_index()['diff'].values)\n",
    "    return x, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "42f4d375-6027-4154-aa93-0f99417d611f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x350 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(6, 3.5), nrows=3, sharex=True)\n",
    "\n",
    "def draw(ax, kind):\n",
    "    x, y = create(kind)\n",
    "    # sns.barplot(data=pd.DataFrame({'x': x, 'y': y}), x='x', y='y', ax=ax)\n",
    "    ax.bar(x, y, color='cyan')\n",
    "    \n",
    "    for rect in ax.patches:\n",
    "        # Get X and Y placement of label from rect.\n",
    "        y_value = rect.get_height()\n",
    "        x_value = rect.get_x() + rect.get_width() / 2\n",
    "    \n",
    "        # Number of points between bar and label. Change to your liking.\n",
    "        space = 3\n",
    "        # Vertical alignment for positive values\n",
    "        va = 'bottom'\n",
    "    \n",
    "        # If value of bar is negative: Place label below bar\n",
    "        if y_value < 0:\n",
    "            # Invert space to place label below\n",
    "            space *= -1\n",
    "            # Vertically align label at top\n",
    "            va = 'top'\n",
    "    \n",
    "        # Use Y value as label and format number with one decimal place\n",
    "        label = \"{:.0f}\".format(y_value)\n",
    "    \n",
    "        # Create annotation\n",
    "        ax.annotate(\n",
    "            label,                      # Use `label` as label\n",
    "            (x_value, y_value),         # Place label at end of the bar\n",
    "            xytext=(0, space),          # Vertically shift label by `space`\n",
    "            textcoords=\"offset points\", # Interpret `xytext` as offset in points\n",
    "            ha='center',\n",
    "            va=va,\n",
    "            color='black')                      # Vertically align label differently for\n",
    "    \n",
    "    ax.spines['top'].set_visible(False)\n",
    "    ax.spines['right'].set_visible(False)\n",
    "    ax.spines['bottom'].set_visible(False)\n",
    "    ax.spines['left'].set_visible(True)\n",
    "\n",
    "    ax.set_ylabel(kind)\n",
    "    ax.set_xlabel('')\n",
    "\n",
    "draw(ax[0], 'train')\n",
    "draw(ax[1], 'public_test')\n",
    "draw(ax[2], 'private_test')\n",
    "\n",
    "fig.tight_layout()\n",
    "plt.savefig('miss_distribution.pdf')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "a32ccc67-bc9e-4b50-873f-ab8a96fa405b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/stud_vantuan/anaconda3/envs/noise/lib/python3.9/site-packages/seaborn/_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "/home/stud_vantuan/anaconda3/envs/noise/lib/python3.9/site-packages/seaborn/_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "/home/stud_vantuan/anaconda3/envs/noise/lib/python3.9/site-packages/seaborn/_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "/home/stud_vantuan/anaconda3/envs/noise/lib/python3.9/site-packages/seaborn/_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "/home/stud_vantuan/anaconda3/envs/noise/lib/python3.9/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
      "  with pd.option_context('mode.use_inf_as_na', True):\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "sns.histplot(data, x='diff', hue='kind', bins=50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b5975fee-c83a-46d4-8998-57f5ce94fa39",
   "metadata": {},
   "source": [
    "# Old"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "1ecd912e-3a13-4933-a4ff-048bf1ffd1bf",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# np.random.seed(100)\n",
    "# shuffled_df = sub_df.sample(frac=1)\n",
    "# shuffled_df['tonal'] = shuffled_df['text'].map(get_list_tonal)\n",
    "# shuffled_df = shuffled_df.drop('text', axis=1)\n",
    "# shuffled_df['path'] = shuffled_df.path.map(lambda x: root_audio + '/' + x)\n",
    "# shuffled_data = list(shuffled_df.T.to_dict().values())\n",
    "# train_ratio = round(0.95 * df.shape[0])\n",
    "# train_data = shuffled_data[:train_ratio]\n",
    "# test_data = shuffled_data[train_ratio:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "43009484-1a92-41e9-b92c-ac3186b190ba",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "13cd4c3e-8fe8-45af-887b-cf8c384f2260",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}