newsqa_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import os, sys\n",
    "from transformers import BertTokenizerFast\n",
    "tokenizer = BertTokenizerFast.from_pretrained(\"bert-base-uncased\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds = json.load(open(\"data/newsqa/combined-newsqa-data-v1.json\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "ds = ds['data']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = [_ for _ in ds if _['type'] == 'train']\n",
    "dev = [_ for _ in ds if _['type'] == 'dev']\n",
    "test = [_ for _ in ds if _['type'] == 'test']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['text', 'type', 'questions', 'storyId'])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds[0].keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9\n",
      "dict_keys(['isQuestionBad', 'consensus', 'validatedAnswers', 'answers', 'q', 'isAnswerAbsent'])\n",
      "train\n"
     ]
    }
   ],
   "source": [
    "print(len(ds[0]['questions']))\n",
    "print( ds[0]['questions'][0].keys() )\n",
    "print(ds[0]['type'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12744\n"
     ]
    }
   ],
   "source": [
    "print(len(ds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'s': 294, 'e': 297}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[0]['questions'][0]['consensus']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11469 638 637\n"
     ]
    }
   ],
   "source": [
    "print(len(train), len(dev), len(test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# qa items\n",
    "def get_consus_stage_1_questions(ds):\n",
    "    consensus_qa_items_stage_1 = {'id':[], 'context':[], 'question':[], 'answers':[]}\n",
    "    consensus_after_validate = {'id':[], 'context':[], 'question':[], 'answers':[]} # _  for _ in validate_questions if len(_['validatedAnswers']) <= 2] # 'consensus' in _]\n",
    "    no_consensus_after_validate = {'id':[], 'context':[], 'question':[], 'answers':[]} #_  for _ in validate_questions if len(_['validatedAnswers']) > 2]     \n",
    "    \n",
    "    for item in ds:\n",
    "        text = item['text']\n",
    "        type = item['type']\n",
    "        questions = item['questions']\n",
    "        storyId = item['storyId']\n",
    "        for qidx, question in enumerate(questions):\n",
    "            qid = '#'.join([storyId, str(qidx)])\n",
    "            if 'validatedAnswers' not in question:\n",
    "                # stage 1\n",
    "                cons = question['consensus']\n",
    "                if 's' in cons:\n",
    "                    # consensused on answer\n",
    "                    answers = {\n",
    "                        'answer_start': [cons['s']],\n",
    "                        'text': [ text[cons['s']: cons['e'] ] ],\n",
    "                    }\n",
    "                else:\n",
    "                    answers = {\n",
    "                        'answer_start': [],\n",
    "                        'text': [],\n",
    "                    }\n",
    "                consensus_qa_items_stage_1['id'].append(qid)\n",
    "                consensus_qa_items_stage_1['context'].append(text)\n",
    "                consensus_qa_items_stage_1['question'].append( question['q'])\n",
    "                consensus_qa_items_stage_1['answers'].append(answers)\n",
    "            else:\n",
    "                # stage 1\n",
    "                cons = question['consensus']\n",
    "                if 's' in cons:\n",
    "                    # consensused on answer\n",
    "                    answers = {\n",
    "                        'answer_start': [cons['s']],\n",
    "                        'text': [ text[cons['s']: cons['e'] ] ],\n",
    "                    }\n",
    "                else:\n",
    "                    answers = {\n",
    "                        'answer_start': [],\n",
    "                        'text': [],\n",
    "                    }\n",
    "                if len(question['validatedAnswers']) <= 2:                    \n",
    "                    # consensus_after_validate.append(\n",
    "                    #     {'id': qid, 'context': text, 'question': question['q'], 'answers': answers, \n",
    "                    #     'annotations':question['answers'], 'validatedAnswers': question['validatedAnswers']}\n",
    "                    # )\n",
    "                    consensus_after_validate['id'].append(qid)\n",
    "                    consensus_after_validate['context'].append(text)\n",
    "                    consensus_after_validate['question'].append( question['q'])\n",
    "                    consensus_after_validate['answers'].append(answers)\n",
    "                else:\n",
    "                    # no_consensus_after_validate.append(\n",
    "                    #     {'id': qid, 'context': text, 'question': question['q'], 'answers': answers,}\n",
    "                    #     #'annotations':question['answers'], 'validatedAnswers': question['validatedAnswers']}\n",
    "                    # )\n",
    "                    no_consensus_after_validate['id'].append(qid)\n",
    "                    no_consensus_after_validate['context'].append(text)\n",
    "                    no_consensus_after_validate['question'].append( question['q'])\n",
    "                    no_consensus_after_validate['answers'].append(answers)                    \n",
    "    return consensus_qa_items_stage_1, consensus_after_validate, no_consensus_after_validate\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_consensus, train_consensus_after_validate, train_no_consensus = get_consus_stage_1_questions(train)\n",
    "dev_consensus, dev_consensus_after_validate, dev_no_consensus = get_consus_stage_1_questions(dev)\n",
    "test_consensus, test_consensus_after_validate, test_no_consensus = get_consus_stage_1_questions(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "questions = []\n",
    "for d in ds:\n",
    "    for _ in d['questions']:\n",
    "        _['text'] = d['text']\n",
    "        questions.append(_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "qa_items = [_ for d in train for _ in d['questions']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'0.0': 76338, '0.33333333333299997': 14154, '0.2': 232, '0.25': 1481, '0.5': 6168, '1.0': 5528, '0.666666666667': 3540, '0.4': 92, '0.6': 26, '0.166666666667': 42, '0.75': 60, '0.125': 4, '0.14285714285700002': 2, '0.8': 4, '0.714285714286': 1, '0.8571428571430001': 1, '0.181818181818': 1}\n"
     ]
    }
   ],
   "source": [
    "answer_absents = np.array( [_['isAnswerAbsent'] for _ in qa_items] )\n",
    "a = set(answer_absents)\n",
    "ansp = list(a)\n",
    "b = {str(_):0 for _ in a}\n",
    "for _ in qa_items:\n",
    "    b[str(_['isAnswerAbsent'])] = b[str(_['isAnswerAbsent'])] + 1\n",
    "print(b)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "cons = [_['consensus'] for _ in qa_items]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'noAnswer': True}"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cons[9356]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'q': 'wAs he deported',\n",
       " 'isQuestionBad': 0.2,\n",
       " 'consensus': {'noAnswer': True},\n",
       " 'answers': [{'sourcerAnswers': [{'noAnswer': True}]},\n",
       "  {'sourcerAnswers': [{'noAnswer': True}]},\n",
       "  {'sourcerAnswers': [{'s': 1623, 'e': 1650}]},\n",
       "  {'sourcerAnswers': [{'s': 776, 'e': 871}]},\n",
       "  {'sourcerAnswers': [{'noAnswer': True}]}],\n",
       " 'isAnswerAbsent': 0.6}"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "qa_items[1152]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  1152,   9356,  15503,  16308,  30919,  35748,  39666,  41812,\n",
       "        46751,  50068,  53192,  61999,  69653,  74374,  75998,  77771,\n",
       "        89394,  91791,  94228,  94326,  97475, 101776, 104188, 105328,\n",
       "       105672, 107313])"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.where(answer_absents == 0.6)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "consensus_stage_1_questions = [_ for _ in questions if 'validatedAnswers' not in _]\n",
    "validate_questions = [_ for _ in questions if 'validatedAnswers' in _]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "68003 51630\n"
     ]
    }
   ],
   "source": [
    "print(len(consensus_stage_1_questions), len(validate_questions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "consensus_after_validate = [_  for _ in validate_questions if len(_['validatedAnswers']) <= 2] # 'consensus' in _]\n",
    "no_consensus_after_validate = [_  for _ in validate_questions if len(_['validatedAnswers']) > 2] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "45240 6390\n"
     ]
    }
   ],
   "source": [
    "print(len(consensus_after_validate), len(no_consensus_after_validate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'isQuestionBad': 0.0, 'consensus': {'noAnswer': True}, 'validatedAnswers': [{'count': 1, 'noAnswer': True}, {'count': 1, 's': 1833, 'e': 1945}, {'count': 1, 's': 1810, 'e': 1945}], 'answers': [{'sourcerAnswers': [{'noAnswer': True}]}, {'sourcerAnswers': [{'s': 1810, 'e': 1945}]}, {'sourcerAnswers': [{'s': 1810, 'e': 1816}, {'s': 1819, 'e': 1826}, {'s': 1833, 'e': 1945}]}], 'q': 'what is the African National Congress ?', 'isAnswerAbsent': 0.33333333333299997, 'text': 'Johannesburg (CNN) -- Miffed by a visa delay that led the Dalai Lama to cancel a trip to South Africa, Archbishop Desmond Tutu lashed out at his government Tuesday, saying it had acted worse than apartheid regimes and had forgotten all that the nation stood for.\\n\\n\\n\\n\\n\\n\"When we used to apply for passports under the apartheid government, we never knew until the last moment what their decision was,\" Tutu said at a news conference. \"Our government is worse than the apartheid government because at least you were expecting it from the apartheid government.\\n\\n\\n\\n\\n\\n\"I have to say that I can\\'t believe this. I really can\\'t believe this,\" Tutu said. \"You have to wake me up and tell me this is actually happening here.\"\\n\\n\\n\\n\\n\\nThe Dalai Lama scrapped his planned trip to South Africa this week after the nation failed to issue him a visa in time, his spokesman said.\\n\\n\\n\\n\\n\\nVisa applications for him and his entourage were submitted to the South African High Commission in New Delhi, India, at the end of August, and original passports were submitted on September 20, more than two weeks ago, a statement on his website said.\\n\\n\\n\\n\\n\\nHowever, South Africa\\'s foreign affairs office said it did not refuse a visa.\\n\\n\\n\\n\\n\\n\"South Africa will not comment on the decision, because it is not our decision, it is his decision,\" according to spokesman Clayson Monyela, who said the visa application was still under consideration.\\n\\n\\n\\n\\n\\nThe Dalai Lama had been invited to the country to receive the Mahatma Gandhi International Award for Peace and Reconciliation and to speak at a number of events, including a lecture in honor of Tutu\\'s 80th birthday. Tutu and the Dalai Lama are recipients of the Nobel Peace Prize.\\n\\n\\n\\n\\n\\nTutu said he would pray for the defeat of South Africa\\'s government, led by the African National Congress (ANC), which is rooted in the fight against the system of apartheid, or legal racial separation, that was present in South Africa until 1994.\\n\\n\\n\\n\\n\\n\"You are disgraceful,\" Tutu said about the government. \"You are behaving in a way that is totally at variance with the things for which we stood.\"\\n\\n\\n\\n\\n\\nThe ANC plans to call on government officials to explain to South Africans why the visa process was delayed, spokesman Jackson Mtembu said. He said everyone was in the dark about this matter.\\n\\n\\n\\n\\n\\nBut he also suggested that Tutu calm down. A comparison to apartheid regimes, he said, was unfair.\\n\\n\\n\\n\\n\\nThis is not the first time the Dalai Lama has not been able to visit South Africa. In 2009, South Africa refused the Tibetan spiritual leader a visa to attend an international peace conference, saying it was not in the country\\'s interest for him to attend.\\n\\n\\n\\n\\n\\nIn refusing the 2009 application, South Africa said that if the Dalai Lama attended the conference, the focus would shift away from the 2010 World Cup, the global soccer championship it was hosting.\\n\\n\\n\\n\\n\\n\"We cannot allow focus to shift to China and Tibet,\" presidential spokesman Thabo Masebe said, adding that South Africa had gained much from its trading relationship with China.\\n\\n\\n\\n\\n\\nThe Dalai Lama fled Tibet in 1959 after a failed uprising against Chinese rule, and China pressures governments around the world to deny him any legitimacy.\\n\\n\\n\\n\\n\\nSpeculation surfaced Tuesday that this year\\'s visit was also affected by South Africa\\'s relationship with China.\\n\\n\\n\\n\\n\\nSouth African Vice President Kgalema Motlanthe visited Beijing last week and met with Chinese President Hu Jintao to discuss bolstering bilateral ties.\\n\\n\\n\\n\\n\\nMotlanthe said South Africa was ready to boost the strategic partnership between the two countries to a new stage, according to the official Chinese news agency Xinhua.\\n\\n\\n\\n\\n\\nBut Monyela said the application had nothing to do with China.\\n\\n\\n\\n\\n\\n\"We are a sovereign nation which takes decisions in our domestic interest,\" Monyela said.\\n\\n\\n\\n\\n\\nThe Dalai Lama posted a message on Twitter last week that said: \"Even if the Chinese leave nothing but ashes, Tibet will rise from these ashes as a free country even if it takes a long time to do so.\"\\n\\n\\n\\n\\n\\nKim Norgaard, CNN\\'s Johannesburg bureau chief, contributed to this report.'}\n"
     ]
    }
   ],
   "source": [
    "print(no_consensus_after_validate[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "size = 0.7\n",
    "vals = [len(consensus_stage_1_questions), len(consensus_after_validate), len(no_consensus_after_validate)]\n",
    "\n",
    "cmap = plt.get_cmap(\"tab20c\")\n",
    "outer_colors = cmap(np.arange(3)*4)\n",
    "\n",
    "def func(pct, allvals):\n",
    "    absolute = int(round(pct/100.*np.sum(allvals)))\n",
    "    return \"{:.1f}%\\n({:d} )\".format(pct, absolute)\n",
    "\n",
    "wedges, texts, autotexts = ax.pie(vals, radius=1, colors=outer_colors,\n",
    "       wedgeprops=dict(width=size, edgecolor='w'),\n",
    "       autopct=lambda pct: func(pct, vals),\n",
    "                textprops=dict(color=\"w\"))\n",
    "# ax.legend(wedges, [f\"{_}\" for _ in range(1, 4)],\n",
    "#           title=\"Number of Answers\",\n",
    "#           loc=\"center left\",\n",
    "#           bbox_to_anchor=(-0.3, 0, 0, 1))\n",
    "\n",
    "plt.setp(autotexts, size=8, weight=\"bold\")\n",
    "plt.tight_layout()\n",
    "# ax.set(aspect=\"equal\", title='Number of different answers')\n",
    "plt.savefig('newsqa_analysis/newsqa_number_of_items_with_answers.pdf', dpi=600, bbox_inches = 'tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_answers(item):\n",
    "    print(item['q'])\n",
    "    print(len(item['text'].split()))\n",
    "    print('----sourcer------')\n",
    "    for ans in item['answers']:\n",
    "        print(ans)\n",
    "        if 's' in ans['sourcerAnswers'][0]:\n",
    "            for _ans in ans['sourcerAnswers']:\n",
    "                print(item['text'][_ans['s']:_ans['e']])\n",
    "    if 'validatedAnswers' not in item:\n",
    "        return        \n",
    "    print('----validate------')\n",
    "    for ans in item['validatedAnswers']:\n",
    "        print(ans)\n",
    "        if 's' in ans:\n",
    "            print(item['text'][ans['s']:ans['e']])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "According to Joan Rivers who was \"totally devoted\"?\n",
      "635\n",
      "----sourcer------\n",
      "{'sourcerAnswers': [{'s': 1623, 'e': 1639}, {'s': 1642, 'e': 1654}]}\n",
      "totally devoted \n",
      "each other. \n",
      "{'sourcerAnswers': [{'s': 1769, 'e': 1809}, {'s': 353, 'e': 364}, {'s': 390, 'e': 402}]}\n",
      "He doted on what she said, she doted on \n",
      "Richardson \n",
      "Liam Neeson \n",
      "{'sourcerAnswers': [{'s': 1642, 'e': 1654}]}\n",
      "each other. \n",
      "----validate------\n",
      "{'count': 2, 'noAnswer': True}\n"
     ]
    }
   ],
   "source": [
    "print_answers(consensus_after_validate[300])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "unswerable_only_after_validation = [_  for _ in consensus_after_validate if 's' not in _['consensus']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6012"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(unswerable_only_after_validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.050253692543027424"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "6012/len(questions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_after_validation = [_  for _ in consensus_after_validate if 's' in _['consensus']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Is the rating similar to the 100 day mark?\n",
      "369\n",
      "----sourcer------\n",
      "{'sourcerAnswers': [{'noAnswer': True}]}\n",
      "{'sourcerAnswers': [{'s': 1953, 'e': 1961}, {'s': 1968, 'e': 2004}, {'s': 2007, 'e': 2030}]}\n",
      "current \n",
      "certainly does not indicate that he \n",
      "out of the woods yet.\"\n",
      "\n",
      "{'sourcerAnswers': [{'s': 620, 'e': 646}, {'s': 652, 'e': 698}]}\n",
      "Obama getting fairly high \n",
      "on most of the issues he has handled so far,\" \n",
      "----validate------\n",
      "{'count': 2, 'noAnswer': True}\n"
     ]
    }
   ],
   "source": [
    "print_answers(unswerable_only_after_validation[300])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "When was the 6.3 magnitude quake reported?\n",
      "281\n",
      "----sourcer------\n",
      "{'sourcerAnswers': [{'s': 137, 'e': 145}]}\n",
      "Monday, \n",
      "{'sourcerAnswers': [{'s': 1099, 'e': 1114}]}\n",
      "1 p.m. Monday.\n",
      "\n",
      "{'sourcerAnswers': [{'s': 1177, 'e': 1182}]}\n",
      "2:20 \n",
      "----validate------\n",
      "{'count': 2, 's': 1099, 'e': 1114}\n",
      "1 p.m. Monday.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print_answers(answer_after_validation[300])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "stage1_badquestions = [_ for _ in consensus_stage_1_questions if 'isQuestionBad' in _ and _['isQuestionBad'] >= 0.5]\n",
    "stage1_good_questions_no_answer = [_ for _ in consensus_stage_1_questions if 'noAnswer' in _['consensus'] and _['consensus']['noAnswer']  and not ('isQuestionBad' in _ and _['isQuestionBad'] != 0)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# bad questions: 11377; # good questions with no answer: 10530\n"
     ]
    }
   ],
   "source": [
    "print(f\"# bad questions: {len(stage1_badquestions)}; # good questions with no answer: {len(stage1_good_questions_no_answer)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'q': 'What could be more militant than a commando?',\n",
       " 'consensus': {'noAnswer': True},\n",
       " 'answers': [{'sourcerAnswers': [{'noAnswer': True}]},\n",
       "  {'sourcerAnswers': [{'noAnswer': True}]}],\n",
       " 'isAnswerAbsent': 1.0,\n",
       " 'text': '(CNN)  -- One Australian soldier, three civilians and Taliban militants were killed early Friday during heavy fighting in southern Afghanistan, according to information from Australian and NATO officials.\\n\\nFour Australian troops have now died in the conflict in Afghanistan.\\n\\nThe incident occurred in Tarin Kowt in Uruzgan province, where Taliban militants killed an Australian commando, the Australian Defence Ministry said.\\n\\nThe 26-year-old commando -- Pvt. Luke Worsley of Sydney -- served with the Special Operations Task Group. This is the fourth Australian troop to die in the Afghan conflict.\\n\\n\"The action in which Private Worsley died only concluded in the last few hours and was characterized by heavy, close quarter fighting. The SOTG was conducting an operation to clear an identified Taliban bomb making facility in Uruzgan province, when the soldier was hit by small arms fire,\" Chief of the Defence Force Air Chief Marshal Angus Houston said.\\n\\nNATO\\'s International Security Assistance Force said \"a significant number of Taliban insurgents were killed or captured as part of the operation. Taliban insurgents initiated the firefight which lasted several hours.\"\\n\\nGen. Carlos Branco, ISAF spokesman, said it is not known how the civilians, two women and a child, died.\\n\\n\"However, we do know that the insurgents fired upon ISAF soldiers from the compound in which the Afghan civilians (two women and one child) were found after the fight. ISAF makes all effort to prevent losses of innocent civilian lives.\"  E-mail to a friend'}"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stage1_good_questions_no_answer[500]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import re\n",
    "import numpy as numpy\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import pandas\n",
    "import numpy as np\n",
    "import tqdm\n",
    "from nltk import wordpunct_tokenize\n",
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "newsqa = json.load(open(\"data/newsqa/combined-newsqa-data-v1.json\"))['data']\n",
    "    \n",
    "dataset = {'train':[], 'dev':[], 'test':[]}\n",
    "\n",
    "for item in newsqa:\n",
    "    dataset[item['type']].append(item)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "662"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(wordpunct_tokenize(dataset['test'][0]['text']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 11469/11469 [00:03<00:00, 3469.74it/s]\n",
      "100%|██████████| 638/638 [00:00<00:00, 3482.06it/s]\n",
      "100%|██████████| 637/637 [00:00<00:00, 3386.01it/s]\n"
     ]
    }
   ],
   "source": [
    "context_word_cnt = {}\n",
    "cons_answer_num_cnt = {}\n",
    "ncons_answer_num_cnt = {}\n",
    "answer_word_cnt = {}\n",
    "\n",
    "for dataset_subset in dataset.keys():\n",
    "    context_word_cnt[dataset_subset] = []\n",
    "    ncons_answer_num_cnt[dataset_subset] = []\n",
    "    cons_answer_num_cnt[dataset_subset] = []\n",
    "    answer_word_cnt[dataset_subset] = []\n",
    "\n",
    "    for i, item in enumerate(tqdm.tqdm(dataset[dataset_subset])):\n",
    "        src = item['text']\n",
    "        storyid = item['storyId']\n",
    "        tokenized_src = wordpunct_tokenize(src)\n",
    "        context_word_cnt[dataset_subset].append(len(tokenized_src))\n",
    "        for j, question in enumerate(item['questions']):\n",
    "            q_ans = []\n",
    "            if 'validatedAnswers' in question and len(question['validatedAnswers']) <= 2:                  \n",
    "                consensus = question['consensus']\n",
    "                consensus_ans_text = None\n",
    "                all_ans_text = []\n",
    "                ans_idx = []\n",
    "                if 's' not in consensus or 'e' not in consensus:\n",
    "                    no_ans = True\n",
    "                else:\n",
    "                    consensus_ans_text = src[consensus['s']:consensus['e']].strip()\n",
    "                # aggregation after validation.            \n",
    "                for answer in question['answers']:\n",
    "                    for ans in answer['sourcerAnswers']:\n",
    "                        q_ans.append(ans)\n",
    "                        if 's' in ans and 'e' in ans:\n",
    "                            ans_t = wordpunct_tokenize(src[ans['s']:ans['e']])\n",
    "                            answer_word_cnt[dataset_subset].append(len(ans_t))\n",
    "                ncons_answer_num_cnt[dataset_subset].append(len(q_ans))\n",
    "            elif 'validatedAnswers' not in question:\n",
    "                for answer in question['answers']:\n",
    "                    for ans in answer['sourcerAnswers']:\n",
    "                        q_ans.append(ans)\n",
    "                        if 's' in ans and 'e' in ans:\n",
    "                            ans_t = wordpunct_tokenize(src[ans['s']:ans['e']])\n",
    "                            answer_word_cnt[dataset_subset].append(len(ans_t))\n",
    "                cons_answer_num_cnt[dataset_subset].append(len(q_ans))\n",
    "    \n",
    "\n",
    "                \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "rotation = 0\n",
    "xtick_size=16\n",
    "xlabel_size=16\n",
    "ylabel_size=16\n",
    "legend_size=15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 648x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "labels = ['Primary-C', 'Primary-NC', 'Test-C', 'Test-NC']\n",
    "widths = [0.55] * len(labels)\n",
    "colors = ['seagreen'] * len(labels)\n",
    "hatch_patterns = ['/'] * len(labels)\n",
    "fig, axs = plt.subplots(figsize=(9, 6))\n",
    "bplot = axs.boxplot(\n",
    "    [cons_answer_num_cnt['train'], ncons_answer_num_cnt['train'], cons_answer_num_cnt['test'], ncons_answer_num_cnt['test']],\n",
    "    patch_artist=True,\n",
    "    labels=labels,\n",
    "    boxprops={'alpha':0.7},\n",
    "    flierprops={'marker':'o', 'markerfacecolor':'darkblue', 'color':'black'},\n",
    "    meanprops={'marker':'D', 'markerfacecolor':'gold'},\n",
    "    medianprops={'linestyle':'--', 'color':'red'},\n",
    "    showmeans=True,\n",
    "    widths=widths,\n",
    "    showfliers=True #False\n",
    ")\n",
    "for patch, color, hatch in zip(bplot['boxes'], colors, hatch_patterns):\n",
    "    patch.set_facecolor(color)\n",
    "    patch.set_hatch(hatch)\n",
    "axs.set_ylim(0, 15)\n",
    "# control fliers\n",
    "fliers = bplot['fliers']\n",
    "for fly in fliers:\n",
    "    fdata = fly.get_data()    \n",
    "    fly.set_data([fdata[0][0]],fdata[1][:8])\n",
    "\n",
    "plt.xticks(rotation=rotation, fontsize=xtick_size)\n",
    "plt.yticks(fontsize=xtick_size)\n",
    "# Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "axs.set_ylabel('#answer annotations', fontsize=18)\n",
    "plt.savefig('number_candidates.pdf', dpi=600, bbox_inches='tight', pad_inches=0)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_subset = 'test' #  'train' # \n",
    "model_name = \"bert-base-uncased\" #   'facebook/bart-base' # \"bert-large-uncased\" # \"roberta-base\" # \n",
    "device_no=4\n",
    "qa_device = 5\n",
    "model_name_in_path = model_name.replace('/', '-')\n",
    "all_question_candidate_answers_with_null_candidate = torch.load(f\"{dataset_subset}_all_question_candidate_answers_qa_no_noanswer_add_null_{model_name_in_path}.pt\")  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_correct_ness = [float(_[1])  for d in all_question_candidate_answers_with_null_candidate for _ in d[3]]\n",
    "gt_answer_correct_ness = [float(_[1]) for d in all_question_candidate_answers_with_null_candidate for i, _ in enumerate(d[3]) if i in d[4]]\n",
    "wrong_answer_correct_ness = [float(_[1]) for d in all_question_candidate_answers_with_null_candidate for i, _ in enumerate(d[3]) if i not in d[4]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_confidence = [float(_) for d in all_question_candidate_answers_with_null_candidate for _ in d[-3]]\n",
    "gt_answer_confidence = [float(_) for d in all_question_candidate_answers_with_null_candidate for i, _ in enumerate(d[-3]) if i in d[4]]\n",
    "wrong_answer_confidence = [float(_) for d in all_question_candidate_answers_with_null_candidate for i, _ in enumerate(d[-3]) if i not in d[4]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Spearmans correlation: 0.325\n"
     ]
    }
   ],
   "source": [
    "from scipy.stats import spearmanr\n",
    "corr, _ = spearmanr(answer_confidence, answer_correct_ness)\n",
    "print('Spearmans correlation: %.3f' % corr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([3581.,  447.,  248.,  178.,  176.,  150.,  115.,   90.,  107.,\n",
       "          82.,   64.,  100.,   94.,   74.,   62.,   58.,   51.,   55.,\n",
       "          51.,   58.,   77.,   54.,   46.,   53.,   58.,   53.,   58.,\n",
       "          59.,   62.,   61.,   64.,   60.,   78.,   56.,   57.,   73.,\n",
       "          57.,  101.,   67.,   95.,   80.,   73.,   54.,   92.,   68.,\n",
       "         113.,   85.,  114.,  120.,  199.]),\n",
       " array([2.52803467e-08, 1.99707219e-02, 3.99414186e-02, 5.99121153e-02,\n",
       "        7.98828119e-02, 9.98535086e-02, 1.19824205e-01, 1.39794902e-01,\n",
       "        1.59765599e-01, 1.79736295e-01, 1.99706992e-01, 2.19677689e-01,\n",
       "        2.39648385e-01, 2.59619082e-01, 2.79589779e-01, 2.99560475e-01,\n",
       "        3.19531172e-01, 3.39501868e-01, 3.59472565e-01, 3.79443262e-01,\n",
       "        3.99413958e-01, 4.19384655e-01, 4.39355352e-01, 4.59326048e-01,\n",
       "        4.79296745e-01, 4.99267442e-01, 5.19238138e-01, 5.39208835e-01,\n",
       "        5.59179532e-01, 5.79150228e-01, 5.99120925e-01, 6.19091622e-01,\n",
       "        6.39062318e-01, 6.59033015e-01, 6.79003712e-01, 6.98974408e-01,\n",
       "        7.18945105e-01, 7.38915802e-01, 7.58886498e-01, 7.78857195e-01,\n",
       "        7.98827892e-01, 8.18798588e-01, 8.38769285e-01, 8.58739982e-01,\n",
       "        8.78710678e-01, 8.98681375e-01, 9.18652072e-01, 9.38622768e-01,\n",
       "        9.58593465e-01, 9.78564162e-01, 9.98534858e-01]),\n",
       " <BarContainer object of 50 artists>)"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(answer_correct_ness, bins=50, alpha=0.3)\n",
    "plt.hist(answer_confidence, bins=50, alpha=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(wrong_answer_correct_ness, alpha=0.3, bins=50, density=True, cumulative=True, color='#fe7f2d', label='Wrong Answer')\n",
    "plt.hist(gt_answer_correct_ness, alpha=0.6, bins=50, density=True, cumulative=True, color='#6a994e', label='Correct Answer')\n",
    "plt.xlabel('Answer Correctness', fontsize=xlabel_size)\n",
    "plt.grid()\n",
    "plt.legend(fontsize=legend_size)\n",
    "plt.xticks(fontsize=xtick_size)\n",
    "plt.yticks(fontsize=xtick_size)\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.02])\n",
    "plt.savefig('answer_correctness_distribution.pdf', dpi=600, bbox_inches='tight', pad_inches=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "plt.hist(wrong_answer_confidence, alpha=0.3, bins=50, density=True, cumulative=True, color='#fe7f2d', label='Wrong Answer')\n",
    "plt.hist(gt_answer_confidence, alpha=0.6, bins=50, density=True, cumulative=True, color='#6a994e', label='Correct Answer')\n",
    "plt.xlabel('Answer Confidence', fontsize=xlabel_size)\n",
    "plt.grid()\n",
    "plt.legend(loc=4,fontsize=legend_size)\n",
    "plt.xticks(fontsize=xtick_size)\n",
    "plt.yticks(fontsize=xtick_size)\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.02])\n",
    "plt.savefig('answer_confidence_distribution.pdf', dpi=600, bbox_inches='tight', pad_inches=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f037fbbb760>"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(gt_answer_correct_ness, gt_answer_confidence, alpha=0.5, c='#52b69a')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[200.,  19.,  12.,   6.,  21.,  17.,  13.,  26.,  16.,  22.],\n",
       "        [ 30.,   1.,   1.,   3.,   0.,   1.,   2.,   5.,   1.,   2.],\n",
       "        [ 23.,   4.,   2.,   3.,   0.,   4.,   1.,   0.,   2.,   4.],\n",
       "        [ 12.,   4.,   0.,   1.,   1.,   1.,   0.,   3.,   1.,   2.],\n",
       "        [ 16.,   5.,   2.,   2.,   1.,   0.,   2.,   2.,   0.,   1.],\n",
       "        [ 10.,   3.,   8.,   7.,   2.,   0.,   1.,   0.,   0.,   0.],\n",
       "        [ 17.,   3.,   3.,   1.,   1.,   2.,   0.,   1.,   2.,   3.],\n",
       "        [ 19.,   3.,   3.,   5.,   1.,   5.,   2.,   2.,   6.,   8.],\n",
       "        [ 66.,  14.,   4.,   7.,   6.,   6.,   5.,   5.,  10.,  12.],\n",
       "        [715., 133., 145., 101., 103., 124., 157., 204., 186., 362.]]),\n",
       " array([0.00137275, 0.1010504 , 0.20072805, 0.30040571, 0.40008336,\n",
       "        0.49976101, 0.59943866, 0.69911632, 0.79879397, 0.89847162,\n",
       "        0.99814928]),\n",
       " array([2.37164016e-07, 9.98536993e-02, 1.99707161e-01, 2.99560623e-01,\n",
       "        3.99414086e-01, 4.99267548e-01, 5.99121010e-01, 6.98974472e-01,\n",
       "        7.98827934e-01, 8.98681396e-01, 9.98534858e-01]),\n",
       " <matplotlib.collections.QuadMesh at 0x7f037f312730>)"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist2d(gt_answer_correct_ness, gt_answer_confidence)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_confidence = [float(_) for _ in answer_confidence]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([4630.,  544.,  394.,  273.,  288.,  293.,  315.,  393.,  367.,\n",
       "         631.]),\n",
       " array([2.52803467e-08, 9.98535086e-02, 1.99706992e-01, 2.99560475e-01,\n",
       "        3.99413958e-01, 4.99267442e-01, 5.99120925e-01, 6.98974408e-01,\n",
       "        7.98827892e-01, 8.98681375e-01, 9.98534858e-01]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(answer_confidence)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import svm, datasets\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import label_binarize\n",
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.metrics import roc_auc_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "answer_labels = np.array([i in d[4] for d in all_question_candidate_answers_with_null_candidate for i in range(len(d[-3]))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0 0.700492991348696\n",
      "0.05 0.7338948989116774\n",
      "0.1 0.7385438666478934\n",
      "0.15000000000000002 0.7414086007206813\n",
      "0.2 0.7432490887188873\n",
      "0.25 0.7445759870256514\n",
      "0.30000000000000004 0.7453586588653702\n",
      "0.35000000000000003 0.7456791065012541\n",
      "0.4 0.7459246569984825\n",
      "0.45 0.7466313235608538\n",
      "0.5 0.7467194205868702\n",
      "0.55 0.7470422474617107\n",
      "0.6000000000000001 0.7473707128069544\n",
      "0.65 0.7466897941730156\n",
      "0.7000000000000001 0.7457458555612918\n",
      "0.75 0.7440193754791071\n",
      "0.8 0.7420466278687754\n",
      "0.8500000000000001 0.7386772995833398\n",
      "0.9 0.7345307097823877\n",
      "0.9500000000000001 0.7290826436799591\n"
     ]
    }
   ],
   "source": [
    "for _x in np.arange(0.0, 1.0, 0.05):\n",
    "    answer_scores_q = _x*np.array(answer_confidence) + (1-_x)*np.array(answer_correct_ness)\n",
    "    fpr_q, tpr_q, _ = roc_curve(answer_labels, answer_scores_q)\n",
    "    roc_auc_q = auc(fpr_q, tpr_q)\n",
    "    print(_x, roc_auc_q)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "answer_confidence = np.array(answer_confidence)\n",
    "fpr, tpr, _ = roc_curve(answer_labels, answer_confidence)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "answer_correct_ness = np.array(answer_correct_ness)\n",
    "fpr_c, tpr_c, _ = roc_curve(answer_labels, answer_correct_ness)\n",
    "roc_auc_c = auc(fpr_c, tpr_c)\n",
    "\n",
    "answer_scores_s = (answer_confidence+answer_correct_ness)/2\n",
    "fpr_s, tpr_s, _ = roc_curve(answer_labels, answer_scores_s)\n",
    "roc_auc_s = auc(fpr_s, tpr_s)\n",
    "\n",
    "answer_scores_q = np.sqrt(answer_confidence*answer_correct_ness)\n",
    "fpr_q, tpr_q, _ = roc_curve(answer_labels, answer_scores_q)\n",
    "roc_auc_q = auc(fpr_q, tpr_q)\n",
    "\n",
    "\n",
    "# Compute micro-average ROC curve and ROC area\n",
    "# fpr[\"micro\"], tpr[\"micro\"], _ = roc_curve(answer_labels, answer_confidence)\n",
    "# roc_auc[\"micro\"] = auc(fpr[\"micro\"], tpr[\"micro\"])\n",
    "\n",
    "plt.figure()\n",
    "lw = 2\n",
    "plt.plot(\n",
    "    fpr_c,\n",
    "    tpr_c,\n",
    "    color=\"#6a994e\",\n",
    "    lw=lw,\n",
    "    ls='--',\n",
    "    alpha=0.8,\n",
    "    label=\"AC (area = %0.2f)\" % roc_auc_c,\n",
    ")\n",
    "plt.plot(\n",
    "    fpr,\n",
    "    tpr,\n",
    "    color=\"#fe7f2d\",\n",
    "    lw=lw,\n",
    "    ls=':',\n",
    "    alpha=0.8,\n",
    "    label=\"AF (area = %0.2f)\" % roc_auc,\n",
    ")\n",
    "plt.plot(\n",
    "    fpr_s,\n",
    "    tpr_s,\n",
    "    color=\"#6a4c93\",\n",
    "    lw=lw,\n",
    "    ls='-',\n",
    "    alpha=0.6,\n",
    "    label=\"AC+AF (area = %0.2f)\" % roc_auc_s,\n",
    ")\n",
    "# plt.plot(\n",
    "#     fpr_q,\n",
    "#     tpr_q,\n",
    "#     color=\"#fe7f2d\",\n",
    "#     lw=lw,\n",
    "#     ls=':',\n",
    "#     alpha=0.8,\n",
    "#     label=\"AC*AF (area = %0.2f)\" % roc_auc_q,\n",
    "# )\n",
    "plt.plot([0, 1], [0, 1], color=\"navy\", lw=lw, linestyle=\"--\")\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.02])\n",
    "plt.xlabel(\"False Positive Rate\", fontsize=xlabel_size)\n",
    "plt.ylabel(\"True Positive Rate\", fontsize=ylabel_size)\n",
    "# plt.title(\"Receiver operating characteristic example\")\n",
    "plt.legend(loc=\"lower right\", fontsize=legend_size)\n",
    "plt.xticks(fontsize=xtick_size)\n",
    "plt.yticks(fontsize=xtick_size)\n",
    "plt.tight_layout()\n",
    "plt.savefig('roc_curve.pdf', dpi=600, bbox_inches='tight', pad_inches=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.8 ('qgenv')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "1e34c16e4df8892777ca66302c3f06a8d376cf9d4f56d86312236daa28e8d5ab"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}