clustering.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/yiyi/Documents/computerlinguistik-master/ss2019/semantik2/project/cdl\n"
     ]
    }
   ],
   "source": [
    "!pwd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_arg_types_binary= '/home/yiyi/Documents/computerlinguistik-master/ss2019/semantik2/project/data/pred_arg_types_binary.json'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import numpy as np\n",
    "from collections import defaultdict, Counter\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_all_argument_pairs(pred_arg_types_binary):\n",
    "    with open(pred_arg_types_binary) as fp:\n",
    "        pred_arg_types_binary_dict= json.load(fp)\n",
    "        \n",
    "    pred_arg_types_dict={}    # for each pred, {(loc, pers):[(mc,ls),(us, luis)...]}\n",
    "    \n",
    "    pred_args_dict=defaultdict(list) # {'star':[(mc,ls),(us,luis)...]}\n",
    "    pred_arg_types_distribution={}\n",
    "    \n",
    "    arg_all=[]\n",
    "    type_all=[]\n",
    "    arg_pairs_all=[]\n",
    "    type_pairs_all=[]\n",
    "    for pred, args_dict in pred_arg_types_binary_dict.items():\n",
    "        type_arg_dict=defaultdict(list)\n",
    "        arg_pairs=[]\n",
    "        type_pairs=[]\n",
    "        for line  in args_dict:\n",
    "            arg_pair=[]\n",
    "            types=[]\n",
    "            \n",
    "            for index, words_type in sorted(line.items()):\n",
    "                word, type= words_type\n",
    "                types.append(type)\n",
    "                type_all.append(type)\n",
    "                arg_pair.append(word.lower())\n",
    "                arg_all.append(word.lower())\n",
    "            if len(arg_pair)==2:\n",
    "                arg_pairs.append(tuple(arg_pair))\n",
    "                type_pairs.append(tuple(types))\n",
    "                \n",
    "            if len(arg_pair)==3:\n",
    "                arg_pairs.append(tuple((arg_pair[0], arg_pair[1])))\n",
    "                arg_pairs.append(tuple((arg_pair[0], arg_pair[2])))\n",
    "                type_pairs.append(tuple((types[0], types[1])))\n",
    "                type_pairs.append(tuple((types[0], types[2])))\n",
    "            \n",
    "            if len(arg_pair)==4:\n",
    "                arg_pairs.append(tuple((arg_pair[0], arg_pair[1])))\n",
    "                arg_pairs.append(tuple((arg_pair[0], arg_pair[2])))\n",
    "                arg_pairs.append(tuple((arg_pair[0], arg_pair[3])))\n",
    "                type_pairs.append(tuple((types[0], types[1])))\n",
    "                type_pairs.append(tuple((types[0], types[2])))\n",
    "                type_pairs.append(tuple((types[0], types[3])))\n",
    "                \n",
    "        pred_args_dict[pred].extend(arg_pairs)\n",
    "        \n",
    "        arg_pairs_all.extend(arg_pairs)\n",
    "        type_pairs_all.extend(type_pairs)\n",
    "        \n",
    "        first_types=[type_pair[0] for type_pair in type_pairs]\n",
    "        second_types=[type_pair[1] for type_pair in type_pairs]\n",
    "        \n",
    "        first_types_counter= Counter(first_types)\n",
    "        second_types_counter = Counter(second_types)\n",
    "        pred_arg_types_distribution[pred]={\n",
    "            'arg0_type':{type_:freq/len(first_types) for type_, freq in first_types_counter.items() },\n",
    "            'arg1_type':{type_:freq/len(second_types) for type_, freq in second_types_counter.items() }\n",
    "        }\n",
    "        \n",
    "        \n",
    "        for idx, type_pair in enumerate(type_pairs):\n",
    "            type_arg_dict[type_pair].append(arg_pairs[idx])\n",
    "        \n",
    "        pred_arg_types_dict[pred]=type_arg_dict\n",
    "    \n",
    "    pred_type_arg_freq={}\n",
    "    for pred, type_arg_dict in pred_arg_types_dict.items():\n",
    "        types_arg_counter={}\n",
    "        for type_, args in type_arg_dict.items():\n",
    "            types_arg_counter[type_]=Counter(args)\n",
    "        pred_type_arg_freq[pred]= types_arg_counter\n",
    "    return arg_pairs_all, type_pairs_all, arg_all, type_all,  pred_type_arg_freq, pred_arg_types_distribution, pred_args_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14859, 15442, 16, 1418, 50)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "arg_pairs, type_pairs,arg_all, type_all, pred_arg_types_dict, pred_arg_types_distribution, pred_args_dict = extract_all_argument_pairs(pred_arg_types_binary)\n",
    "unique_arg_pairs=list(set(arg_pairs))\n",
    "len(unique_arg_pairs),len(arg_pairs), len(set(type_pairs)), len(pred_arg_types_dict), len(pred_args_dict['star'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "pred_arg_types_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "arg_types_dict=defaultdict(list)\n",
    "for arg, type_ in zip(arg_all, type_all):\n",
    "    arg_types_dict[arg].append(type_)\n",
    "    \n",
    "# arg_type_dict={ arg: type_ for arg, type_ in zip(arg_all, type_all)}\n",
    "# arg_type_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "arg_types_prob_dict={}\n",
    "for arg, types in arg_types_dict.items():\n",
    "    type_dict=Counter(types)\n",
    "    total = len(types)\n",
    "    prob_dict={key:float(\"{0:.2f}\".format(value/total)) for key, value in type_dict.items()}\n",
    "    arg_types_prob_dict[arg]= prob_dict\n",
    "    \n",
    "arg_types_prob_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "15780"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(arg_types_prob_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# pred_arg_types_distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "14859"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "args2id= {pair:idx for idx, pair in enumerate(unique_arg_pairs)}\n",
    "id2args={idx:pair for idx, pair in enumerate(unique_arg_pairs)}\n",
    "len(args2id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1418"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predicates=pred_arg_types_dict.keys()\n",
    "len(predicates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "ls=[0 for i in range(len(args2id))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "14859"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(ls)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# arg_type_dict={arg:type}\n",
    "# vectors with augmentation\n",
    "vector_dict={}\n",
    "for pred, type_pair_dict in pred_arg_types_dict.items():\n",
    "    type_vector_dict={}\n",
    "    print(pred)\n",
    "    ## if applied \n",
    "    arg_type_distribution_dict=pred_arg_types_distribution[pred]\n",
    "    print(type_pair_dict)\n",
    "\n",
    "    for type_pair, arg_counter in type_pair_dict.items():\n",
    "        vector_list=[0 for i in range(len(args2id))]\n",
    "        type0, type1 = type_pair\n",
    "        # initialization with the predictions\n",
    "        for idx, args in id2args.items():\n",
    "            arg0, arg1 = args\n",
    "            prob_dict0, prob_dict1= arg_types_prob_dict[arg0], arg_types_prob_dict[arg1]\n",
    "            joint_distribution = prob_dict0.get(type0,0)* prob_dict1.get(type1,0)\n",
    "            vector_list[idx] = joint_distribution\n",
    "            # print(prob_dict0, prob_dict1)\n",
    "            # type1, type2 = arg_type_dict[arg0], arg_type_dict[arg1]\n",
    "            \n",
    "#             type1_prob = arg_type_distribution_dict['arg0_type'].get(type0,0)\n",
    "#             type2_prob = arg_type_distribution_dict['arg1_type'].get(type1,0)\n",
    "#             vector_list[idx]=  type1_prob* type2_prob+ joint_distribution\n",
    "\n",
    "        \n",
    "        # which has the arg_pair in typed_predicate\n",
    "        for arg_pair, freq in arg_counter.items():            \n",
    "            vector_list[args2id[arg_pair]] += freq \n",
    "            \n",
    "        # type_vector_dict[type_pair]=np.asarray(vector_list)\n",
    "        pred_type_pair=pred+'-'+type_pair[0]+'-'+type_pair[1]\n",
    "        vector_dict[pred_type_pair]= np.asarray(vector_list)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4560"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(vector_dict.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "vector_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# without augmentation\n",
    "vector_pred={}\n",
    "for pred, arg_pair_list in pred_args_dict.items():\n",
    "    vector_list=[0 for i in range(len(args2id))]\n",
    "    for arg_pair in arg_pair_list:\n",
    "        vector_list[args2id[arg_pair]]+=1\n",
    "    vector_pred[pred]=np.asarray(vector_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "keys_pred= []\n",
    "matrix_pred= []\n",
    "for key, vector in vector_pred.items():\n",
    "    keys_pred.append(key)\n",
    "    matrix_pred.append(vector)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1418, 14859)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_pred = np.vstack(matrix_pred)\n",
    "features_pred.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1418,)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label2id_pred={label:idx for idx, label in enumerate(keys_pred)}\n",
    "id2label_pred={idx:label for idx, label in enumerate(keys_pred)}\n",
    "labels_pred= np.array(keys_pred)\n",
    "labels_pred.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f20a4c66198>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from scipy import sparse\n",
    "\n",
    "kmeans_pred= KMeans(n_clusters=4)\n",
    "kmeans_pred.fit(features_pred)\n",
    "y_kmeans_pred= kmeans_pred.predict(features_pred)\n",
    "plt.scatter(features_pred[:, 0], features_pred[:, 1],  c=y_kmeans_pred, s=50, cmap='viridis')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# augmented data\n",
    "keys= []\n",
    "matrix= []\n",
    "for key, vector in vector_dict.items():\n",
    "    keys.append(key)\n",
    "    matrix.append(vector)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4560, 14859)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features=np.vstack(matrix)\n",
    "features.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "label2id={label:idx for idx, label in enumerate(keys)}\n",
    "id2label={idx:label for idx, label in enumerate(keys)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4560,)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels=np.asarray(keys)\n",
    "labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=df.rename(index= {0:\"write_{arg0:PER,arg1:Book}\",1:\"suggest_{arg0:ORG, arg1:LOC}\",2:\"suggest_{arg0:PER, arg1:PER}\"})\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans\n",
    "from scipy import sparse\n",
    "\n",
    "kmeans= KMeans(n_clusters=4)\n",
    "kmeans.fit(features)\n",
    "y_kmeans= kmeans.predict(features)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f208ef1ec18>"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(features[:, 0], features[:, 1],  c=y_kmeans, s=50, cmap='viridis')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: ('harrison', 'john lennon'),\n",
       " 1: ('german', 'languages of the european union'),\n",
       " 2: ('j. m. coetzee', 'waiting for the barbarians'),\n",
       " 3: ('finland', 'finland'),\n",
       " 4: ('hypercard', 'violawww'),\n",
       " 5: ('owendo', 'booué'),\n",
       " 6: ('shimon peres', 'kadima'),\n",
       " 7: ('john galt', 'atlas shrugged'),\n",
       " 8: ('cpr', 'budd rail diesel cars'),\n",
       " 9: ('brazil', 'renault'),\n",
       " 10: ('jehosaphat', 'asa'),\n",
       " 11: ('turku', 'finnish language'),\n",
       " 12: ('soyuz rocket', 'french guiana'),\n",
       " 13: ('euromonitor international', 'artistry'),\n",
       " 14: ('dc', 'the spirit'),\n",
       " 15: ('roger bacon', 'hebrew'),\n",
       " 16: ('artemis', 'apollo'),\n",
       " 17: ('audi', 'new york yankees'),\n",
       " 18: ('european union', 'hungary'),\n",
       " 19: ('kraków', 'roman catholic church'),\n",
       " 20: ('dayton', 'united states'),\n",
       " 21: ('european economic community', 'eu'),\n",
       " 22: ('f-14', 'u.s. navy'),\n",
       " 23: ('eighth avenue', 'broadway'),\n",
       " 24: ('mamet', 'israel'),\n",
       " 25: ('hurst', 'argentina'),\n",
       " 26: ('davis', 'garfield'),\n",
       " 27: ('palácio do planalto', 'brasília'),\n",
       " 28: ('nhs scotland', 'aberdeen'),\n",
       " 29: ('party', 'stalin'),\n",
       " 30: ('iowa', 'mississippi river'),\n",
       " 31: ('holy see', 'committee of ministers'),\n",
       " 32: ('malcolm pasley', 'bodleian library'),\n",
       " 33: ('ponce de león', 'cortés'),\n",
       " 34: ('italy', 'hannibal'),\n",
       " 35: ('ecuadorian jews', 'israel'),\n",
       " 36: ('selznick', 'goldwyn'),\n",
       " 37: ('galois', 'republican'),\n",
       " 38: ('eisenhower', 'arbenz'),\n",
       " 39: ('john sebastian', 'god only knows'),\n",
       " 40: ('arthur jaffe', 'cmi'),\n",
       " 41: ('cameroon range', 'mount cameroon'),\n",
       " 42: ('dlr', 'river thames'),\n",
       " 43: ('alcmene', 'iphicles'),\n",
       " 44: ('mircea the elder', 'wallachia'),\n",
       " 45: ('elba', 'napoleon'),\n",
       " 46: ('mcfarlane', 'tehran'),\n",
       " 47: ('chile', '1959 fiba world championship'),\n",
       " 48: ('indians', 'cc sabathia'),\n",
       " 49: ('wallach', '2007 emmy'),\n",
       " 50: ('8th army', 'tunisia campaign'),\n",
       " 51: ('eastern orthodox churches', 'septuagint'),\n",
       " 52: ('calif.', 'caltech'),\n",
       " 53: ('brown', 'united kingdom'),\n",
       " 54: ('j.p. morgan', 'u.s. steel'),\n",
       " 55: ('fulham', 'the football league'),\n",
       " 56: ('capra', 'academy awards'),\n",
       " 57: ('east german', 'gdr'),\n",
       " 58: ('columbia records', 'bob dylan'),\n",
       " 59: ('georgia', 'new jersey'),\n",
       " 60: ('nomar garciaparra', 'chicago cubs'),\n",
       " 61: ('boccaccio', 'petrarch'),\n",
       " 62: ('andrew', 'halych'),\n",
       " 63: ('gaius verres', 'cicero'),\n",
       " 64: ('yahweh', 'david'),\n",
       " 65: ('kingdom of ireland', 'united kingdom of great britain and ireland'),\n",
       " 66: ('kulmerland', 'chełmno'),\n",
       " 67: ('arthur cayley', 'de morgan'),\n",
       " 68: ('sinatra', 'dorsey'),\n",
       " 69: ('crick', 'nature'),\n",
       " 70: ('london', 'greater london'),\n",
       " 71: ('cuban missile crisis', 'u.s.'),\n",
       " 72: ('marlins', 'yusmeiro petit'),\n",
       " 73: ('byzantine empire', 'constantinople'),\n",
       " 74: ('cadillac', 'northern michigan'),\n",
       " 75: ('university of new brunswick', 'north america'),\n",
       " 76: ('truffaut', 'jules and jim'),\n",
       " 77: ('rommel', 'friedrich paulus'),\n",
       " 78: ('adelaide', 'port adelaide power'),\n",
       " 79: ('bosnia herzegovina',\n",
       "  'united nations transitional authority in cambodia'),\n",
       " 80: ('henry', 'matilda'),\n",
       " 81: ('sugar & spice', 'bring it on'),\n",
       " 82: ('great northern war', 'sorau'),\n",
       " 83: ('huns', 'athanaric'),\n",
       " 84: ('history of spain', 'europe'),\n",
       " 85: ('roosevelt', 'massachusetts'),\n",
       " 86: ('henry', 'pembrokeshire'),\n",
       " 87: ('inner temple', 'city of london'),\n",
       " 88: ('connecticut', 'america'),\n",
       " 89: ('irish free state', 'dáil'),\n",
       " 90: ('republican', 'barack obama'),\n",
       " 91: ('turkey river', 'bow'),\n",
       " 92: ('jewish', 'artaxerxes'),\n",
       " 93: ('azad kashmir', 'a devastating earthquake'),\n",
       " 94: ('anna komnene', 'pisans'),\n",
       " 95: ('aim-54 phoenix', 'usn'),\n",
       " 96: ('peter ustinov', 'death on the nile'),\n",
       " 97: ('fenian cycle', 'munster'),\n",
       " 98: ('formosus', 'arnulf'),\n",
       " 99: ('rfc', 'rfc'),\n",
       " 100: ('chaucer', 'philippa ( de ) roet'),\n",
       " 101: ('nicolas cage', 'the cotton club'),\n",
       " 102: ('epirus', 'achilles'),\n",
       " 103: ('hudson', 'adirondacks'),\n",
       " 104: ('east germans', 'west'),\n",
       " 105: ('dick grayson', 'nightwing'),\n",
       " 106: ('claudius ptolemaeus', 'isle of wight'),\n",
       " 107: ('limerick', \"st. patrick 's day\"),\n",
       " 108: ('farnborough', 'royal air force'),\n",
       " 109: ('u.s.', 'c.a.r'),\n",
       " 110: ('john xxiii', 'second vatican council'),\n",
       " 111: ('electron', 'dragon 32'),\n",
       " 112: ('soviet', 'american'),\n",
       " 113: ('thomery', 'fontainebleau'),\n",
       " 114: ('english kings', 'northumbria'),\n",
       " 115: ('haman', 'jews'),\n",
       " 116: ('clapton', \"donald ' duck ' dunn\"),\n",
       " 117: ('edward gibbon', 'the foundation trilogy'),\n",
       " 118: ('constantius', 'gallus'),\n",
       " 119: ('catholics', 'spanish'),\n",
       " 120: ('zeus', 'maia'),\n",
       " 121: ('amsterdam', 'montessori'),\n",
       " 122: ('spanish new mexico', 'snake river plain'),\n",
       " 123: ('thyestes', 'atreus'),\n",
       " 124: ('state of arkansas', 'union'),\n",
       " 125: ('himmler', 'gotthard heinrici'),\n",
       " 126: ('maximian', 'carausius'),\n",
       " 127: ('frances fisher', 'jill taylor'),\n",
       " 128: ('rohrabacher', 'abramoff'),\n",
       " 129: ('f-14a', 'us navy'),\n",
       " 130: ('spain', 'european union'),\n",
       " 131: ('elizabeth ii', 'terminal 1'),\n",
       " 132: ('greece', 'european union'),\n",
       " 133: ('alexander mackendrick', 'sweet smell of success'),\n",
       " 134: ('conservative', 'clement attlee'),\n",
       " 135: ('farrell', 'houston'),\n",
       " 136: ('lorenzo', 'botticelli'),\n",
       " 137: ('aberfoyle', 'glasgow'),\n",
       " 138: ('heraklion', 'crete'),\n",
       " 139: ('florida', 'florida'),\n",
       " 140: ('medieval latin', 'shqipëri'),\n",
       " 141: ('edgar', 'dunstan'),\n",
       " 142: ('the godfather', 'don vito corleone'),\n",
       " 143: ('aclu', 'nixon'),\n",
       " 144: ('haryana', 'national capital region'),\n",
       " 145: ('iran', 'unesco'),\n",
       " 146: ('john locke', 'thomas jefferson'),\n",
       " 147: ('major league baseball', 'hank aaron award'),\n",
       " 148: ('lerner', 'world war ii'),\n",
       " 149: ('botswana', 'southern africa'),\n",
       " 150: ('channel 4', 'er'),\n",
       " 151: ('sagan', 'american space program'),\n",
       " 152: ('the machinist', 'memento'),\n",
       " 153: ('seneca river', 'lake ontario'),\n",
       " 154: ('fra angelico', 'gentile da fabriano'),\n",
       " 155: ('aim', 'oscar'),\n",
       " 156: (\"qur'an\", 'benjamin'),\n",
       " 157: ('london', 'æthelberht'),\n",
       " 158: ('flanders', 'brussels'),\n",
       " 159: ('balderus', 'høtherus'),\n",
       " 160: ('chelsea', 'tottenham hotspur'),\n",
       " 161: ('islands in the detroit river', 'grosse ile'),\n",
       " 162: ('biblical', 'bible'),\n",
       " 163: ('reinhard heydrich', 'sa'),\n",
       " 164: ('church of england', 'charles i'),\n",
       " 165: ('council of europe', 'european nations'),\n",
       " 166: ('bedford park', 'jerome park reservoir'),\n",
       " 167: ('british aerospace', 'harrier ii'),\n",
       " 168: ('coolidge', 'plymouth notch'),\n",
       " 169: ('air tanzania', 'dar es salaam'),\n",
       " 170: ('a10', 'paris'),\n",
       " 171: ('leary', 'american'),\n",
       " 172: ('river lynher', 'england'),\n",
       " 173: ('gates', 'lakeside school'),\n",
       " 174: ('ghana', 'gulf of guinea'),\n",
       " 175: ('chigwell', 'greater london urban area'),\n",
       " 176: ('bas', 'bulgaria'),\n",
       " 177: ('brownlow medal', 'james hird'),\n",
       " 178: ('kerensky', 'lvov'),\n",
       " 179: ('frederick', 'germany'),\n",
       " 180: ('clapton', 'albion'),\n",
       " 181: ('west berlin', 'east'),\n",
       " 182: ('babylonian', 'old persian'),\n",
       " 183: ('saab', 'asja'),\n",
       " 184: ('alexander', 'elagabalus'),\n",
       " 185: ('arabic', 'sahara desert'),\n",
       " 186: ('sullan', 'pompeii'),\n",
       " 187: ('candide', 'pangloss'),\n",
       " 188: ('huck', 'indian territory'),\n",
       " 189: ('balochistan', 'pakistan'),\n",
       " 190: ('yevgeny zamyatin', 'orwell'),\n",
       " 191: ('kaltenbrunner', 'austria'),\n",
       " 192: ('sooke', 'capital regional district'),\n",
       " 193: ('zeus', 'hera'),\n",
       " 194: ('quran', 'arabic'),\n",
       " 195: ('american airlines', 'aa'),\n",
       " 196: ('northsound radio', 'northsound two'),\n",
       " 197: ('jones', 'jimmy johnson'),\n",
       " 198: ('banks', 'wembley'),\n",
       " 199: ('welch', 'furst'),\n",
       " 200: ('clinton', 'continental army'),\n",
       " 201: ('preston gomez', 'bill virdon'),\n",
       " 202: (\"vasari 's elevated corridor\", 'palazzo pitti'),\n",
       " 203: ('u.s. news & world report', 'university of chicago medical center'),\n",
       " 204: ('joseph white', 'latin'),\n",
       " 205: ('detroit', 'pittsburgh pirates'),\n",
       " 206: ('karimata strait', 'borneo'),\n",
       " 207: ('serrarius', 'spinoza'),\n",
       " 208: ('tomb raider', 'eidos'),\n",
       " 209: ('young', 'boston'),\n",
       " 210: ('octavian', 'octavia minor'),\n",
       " 211: ('santa maria sopra minerva', 'pantheon'),\n",
       " 212: ('robert palmer', 'ken olsen'),\n",
       " 213: ('afonso', 'castile'),\n",
       " 214: ('forbes', 'american'),\n",
       " 215: ('madero', 'coahuila'),\n",
       " 216: ('nuremberg nazi party rally', 'hitler'),\n",
       " 217: ('mandelbrot', 'centre national de la recherche scientifique'),\n",
       " 218: ('chile', 'almirante latorre'),\n",
       " 219: ('apple', 'dell'),\n",
       " 220: ('ansip', 'reform party'),\n",
       " 221: ('adelaide', 'sanfl'),\n",
       " 222: ('artemis', 'artimus'),\n",
       " 223: ('port of chicago', 'mississippi river'),\n",
       " 224: ('warhol superstar', 'warhol'),\n",
       " 225: ('afyonkarahisar', 'afyon province'),\n",
       " 226: ('fsf', 'gcc'),\n",
       " 227: ('laozi', 'shiji'),\n",
       " 228: ('strafford', 'charles'),\n",
       " 229: ('germany', 'denmark'),\n",
       " 230: ('globalfoundries', 'amd'),\n",
       " 231: ('nobel prize in literature', 'solzhenitsyn'),\n",
       " 232: ('carl perkins', 'jefferson'),\n",
       " 233: ('canadian', 'us'),\n",
       " 234: ('jan wils', 'olympic stadium'),\n",
       " 235: ('achilles', 'briseis'),\n",
       " 236: ('great gate of kiev', 'hartmann'),\n",
       " 237: ('persian language', 'dari'),\n",
       " 238: ('lycus', 'maeander river'),\n",
       " 239: ('columbia', 'rotc'),\n",
       " 240: ('miles austin', 'kansas city'),\n",
       " 241: ('eadulf', 'northumbria'),\n",
       " 242: ('dimona', \"israel 's solar transformation\"),\n",
       " 243: ('tim burton', 'joel schumacher'),\n",
       " 244: ('romans', 'ancyra'),\n",
       " 245: ('rick saada', 'castle of the winds'),\n",
       " 246: ('lee', 'toronto'),\n",
       " 247: ('us', \"scarlett o'hara\"),\n",
       " 248: ('charlemagne', 'charles'),\n",
       " 249: ('andré rené roussimoff', 'french'),\n",
       " 250: ('americans', 'u.s.'),\n",
       " 251: ('my many colored days', 'geisel'),\n",
       " 252: ('persians', 'greece'),\n",
       " 253: ('robert mueller municipal airport', 'austin'),\n",
       " 254: ('alice guy blaché', 'la fée aux choux'),\n",
       " 255: ('albania', 'mediterranean sea'),\n",
       " 256: ('chievo', 'portuguese'),\n",
       " 257: ('cicero', 'rhodes'),\n",
       " 258: ('germany', 'czech republic'),\n",
       " 259: ('edinburgh crossrail', 'newcraighall'),\n",
       " 260: ('gotham city', 'warner brothers'),\n",
       " 261: ('fred lane', 'indianapolis colts'),\n",
       " 262: ('internazionale', 'massimo moratti'),\n",
       " 263: ('idaho', 'mackay'),\n",
       " 264: ('first punic war', 'italian peninsula'),\n",
       " 265: ('auchinleck', 'ruweisat ridge'),\n",
       " 266: ('nintendo', 'game boy'),\n",
       " 267: ('hawke', 'melbourne'),\n",
       " 268: ('austria', 'war against napoleon'),\n",
       " 269: ('porfirio díaz', 'madero'),\n",
       " 270: ('finns', 'russia'),\n",
       " 271: ('chrysler building', '40 wall street'),\n",
       " 272: ('leibniz', 'spinoza'),\n",
       " 273: ('apollo 14', 'shepard'),\n",
       " 274: ('terry venables', 'mark hughes'),\n",
       " 275: ('catherine', 'mary i'),\n",
       " 276: ('doubleday', 'lincoln'),\n",
       " 277: ('a28 road', 'isle of thanet'),\n",
       " 278: ('bamberg', 'germany'),\n",
       " 279: ('des moines', 'ames'),\n",
       " 280: ('capp', 'newsweek'),\n",
       " 281: ('carme', 'jupiter'),\n",
       " 282: ('daly', 'tony award'),\n",
       " 283: ('china', 'u.s.'),\n",
       " 284: ('f-15', 'u.s. air force'),\n",
       " 285: ('mcgovern', 'hubert humphrey'),\n",
       " 286: ('hugo awards', 'taklamakan'),\n",
       " 287: ('natasha stott despoja', 'meg lees'),\n",
       " 288: ('mca', 'umno'),\n",
       " 289: ('nobel prize in chemistry', 'edwin mcmillan'),\n",
       " 290: ('gnome project', 'gimp'),\n",
       " 291: ('bosnia', 'bosnia and herzegovina'),\n",
       " 292: ('spanish civil war', 'homage to catalonia'),\n",
       " 293: ('gdr', 'comecon'),\n",
       " 294: ('pausanias', 'asia minor'),\n",
       " 295: ('jake delhomme', 'carolina panthers'),\n",
       " 296: ('dimona', 'southern district of israel'),\n",
       " 297: ('agamemnon', 'briseis'),\n",
       " 298: ('plato', 'las vegas'),\n",
       " 299: ('eugene', 'marlborough'),\n",
       " 300: ('sts-119', 'integrated truss structure'),\n",
       " 301: ('roberts', 'eisenhower'),\n",
       " 302: ('gallatin', 'jackson'),\n",
       " 303: ('ahab', 'israel'),\n",
       " 304: ('manchester united', 'rio ferdinand'),\n",
       " 305: ('disney xd', 'digimon data squad'),\n",
       " 306: ('ottomans', 'hungarian army'),\n",
       " 307: ('aol', 'america online'),\n",
       " 308: ('geriadur brezhoneg an here', 'breton'),\n",
       " 309: ('belarus', 'eurovision song contest'),\n",
       " 310: ('german', 'germany'),\n",
       " 311: ('arena motorsport', 'british touring car championship'),\n",
       " 312: ('united methodist church', 'council of bishops'),\n",
       " 313: ('npd group', 'united states'),\n",
       " 314: ('frederick iii', 'wilhelm ii'),\n",
       " 315: ('ronald reagan', 'sdi'),\n",
       " 316: ('croydon flyover', 'taberner house'),\n",
       " 317: ('georges prosper remi', 'belgian'),\n",
       " 318: ('pacific', 'british columbia'),\n",
       " 319: ('naomi', 'bethlehem'),\n",
       " 320: ('bill ward', 'english'),\n",
       " 321: ('romania', 'danube'),\n",
       " 322: ('new york airways', 'bv 107-ii'),\n",
       " 323: ('in africa', 'tanzania'),\n",
       " 324: ('friedrich schiller', 'to joy'),\n",
       " 325: ('carleton', 'arnold'),\n",
       " 326: ('surat huseynov', 'soviet azerbaijan'),\n",
       " 327: ('clapton', \"knockin ' on heaven 's door\"),\n",
       " 328: ('delaware department of transportation', 'deldot'),\n",
       " 329: ('armed forces of bosnia and herzegovina', 'bosnia and herzegovina'),\n",
       " 330: ('capp', 'tv guide'),\n",
       " 331: ('red sox fans', 'red sox nation'),\n",
       " 332: ('charles', 'fox'),\n",
       " 333: ('doris day', 'american'),\n",
       " 334: ('italian', 'constantine'),\n",
       " 335: ('silvestre de sacy', 'french'),\n",
       " 336: ('vienna conservatory',\n",
       "  'imperial academy of music and the performing arts'),\n",
       " 337: ('french', 'egypt'),\n",
       " 338: ('romans', 'roman empire'),\n",
       " 339: ('arcturus', 'boötes'),\n",
       " 340: ('hampshire', 'great britain'),\n",
       " 341: ('cardiff', 'pierhead building'),\n",
       " 342: ('candide', 'americas'),\n",
       " 343: ('european council', 'josé manuel durão barroso'),\n",
       " 344: ('h & k', 'bae systems'),\n",
       " 345: ('roman', 'agrippina'),\n",
       " 346: ('simon', 'progress and poverty'),\n",
       " 347: ('goldoni', 'italian'),\n",
       " 348: ('tupper', 'liberal'),\n",
       " 349: ('freddy taylor', 'nagasaki'),\n",
       " 350: ('constantius', 'constantine'),\n",
       " 351: ('a prairie home companion', 'new zealand'),\n",
       " 352: ('adelaide', 'cluny'),\n",
       " 353: ('galenic', 'de humani corporis fabrica'),\n",
       " 354: ('american league', 'los angeles'),\n",
       " 355: ('rose', 'bbc one'),\n",
       " 356: ('tamil nadu', 'west bengal'),\n",
       " 357: ('professional football', 'american football league'),\n",
       " 358: ('boniface', 'saxon'),\n",
       " 359: ('diomedes', 'ares'),\n",
       " 360: ('peru', 'tacna'),\n",
       " 361: ('grímur jónsson thorkelin', 'latin'),\n",
       " 362: ('makuria', 'nobadia'),\n",
       " 363: ('alexander haig', 'b-1'),\n",
       " 364: ('springfield', 'missouri'),\n",
       " 365: ('hirokazu tanaka', 'nintendo'),\n",
       " 366: ('network ten', 'adelaide'),\n",
       " 367: ('godzilla vs. biollante', 'toho'),\n",
       " 368: ('sharon', 'yitzhak rabin'),\n",
       " 369: ('domesday book', 'city'),\n",
       " 370: ('galilee', 'druze'),\n",
       " 371: ('windows server 2008', 'windows server 2003'),\n",
       " 372: ('adobe', 'opentype'),\n",
       " 373: ('gypsy', 'arthur laurents'),\n",
       " 374: ('puget sound', 'strait of juan de fuca'),\n",
       " 375: ('andreas palaiologos', 'ferdinand and isabella'),\n",
       " 376: ('the davison', 'detroit'),\n",
       " 377: ('albert einstein', 'kant'),\n",
       " 378: ('carson city', 'america'),\n",
       " 379: ('napoleon', 'paoli'),\n",
       " 380: ('allah', 'lut'),\n",
       " 381: ('chaldean aramaic', 'babylonia'),\n",
       " 382: ('hadar', 'centaurus'),\n",
       " 383: ('kelly', 'casey'),\n",
       " 384: ('finland', 'united states'),\n",
       " 385: ('france', 'clipperton'),\n",
       " 386: ('sinatra', 'las vegas'),\n",
       " 387: ('hellas verona', 'fiorentina'),\n",
       " 388: ('carlos salinas de gortari', 'mexico'),\n",
       " 389: ('göring', 'berlin lichterfelde'),\n",
       " 390: ('lincoln', 'our american cousin'),\n",
       " 391: ('lancastrians', 'wars of the roses'),\n",
       " 392: ('samuel pepys', 'harwich'),\n",
       " 393: ('david koepp', 'snake eyes'),\n",
       " 394: ('fellini', 'disneyland'),\n",
       " 395: ('anatolia', 'europe'),\n",
       " 396: ('james boswell', 'scottish enlightenment'),\n",
       " 397: ('nansen', 'ga.'),\n",
       " 398: ('franklin', 'london'),\n",
       " 399: ('jane marple', 'sleeping murder'),\n",
       " 400: ('channel tunnel', 'france'),\n",
       " 401: ('beretta cx4 storm', 'beretta px4 storm'),\n",
       " 402: ('schwarzenegger', 'california'),\n",
       " 403: ('syria', 'roman'),\n",
       " 404: ('digby', 'alcester'),\n",
       " 405: ('fryderyk chopin', 'warsaw lyceum'),\n",
       " 406: ('smith', 'church of christ'),\n",
       " 407: ('church of ireland', 'disraeli'),\n",
       " 408: ('fisher', 'lion class'),\n",
       " 409: ('new york city site', 'westbeth artists community'),\n",
       " 410: ('e-3', 'boeing e-767'),\n",
       " 411: ('bill gates', 'microsoft'),\n",
       " 412: ('easter road stadium', 'leith'),\n",
       " 413: ('love', 'hole'),\n",
       " 414: ('cologne', 'river rhine'),\n",
       " 415: ('vasco da gama', 'india'),\n",
       " 416: ('alberto núñez feijóo', 'galicia'),\n",
       " 417: ('harwich', 'esbjerg'),\n",
       " 418: ('art modell', 'baltimore'),\n",
       " 419: ('eric cheney', 'british'),\n",
       " 420: ('hugh marston hefner', 'american'),\n",
       " 421: ('andreas aagesen', 'danish'),\n",
       " 422: ('disney', 'devo 2.0'),\n",
       " 423: ('patassé', 'libyan'),\n",
       " 424: ('republicans', 'hoover'),\n",
       " 425: ('vespasian', 'rome'),\n",
       " 426: ('delaware', 'i-95'),\n",
       " 427: ('buffy', 'tv series of the same name'),\n",
       " 428: ('sagan', 'venus'),\n",
       " 429: ('gypsy rose lee', 'ethel merman'),\n",
       " 430: ('cleveland', 'democrats'),\n",
       " 431: ('rommel', 'normandy'),\n",
       " 432: ('multitronic', 'elvis presley'),\n",
       " 433: ('judaism', 'messiah'),\n",
       " 434: ('silla', 'goguryeo'),\n",
       " 435: ('ammonites', 'moabites'),\n",
       " 436: ('isaac albéniz', 'felipe pedrell'),\n",
       " 437: ('heinkel he 111', 'ju 52'),\n",
       " 438: ('bank of england', 'city'),\n",
       " 439: ('bosch', 'puerto rico'),\n",
       " 440: ('zooppa', 'mini cooper'),\n",
       " 441: ('alfred the great', 'wessex'),\n",
       " 442: ('vulgate', 'hebrew'),\n",
       " 443: ('canso causeway', 'mainland nova scotia'),\n",
       " 444: ('dreyfuss', 'american league'),\n",
       " 445: ('china', 'washington'),\n",
       " 446: ('norwegian', \"a doll 's house\"),\n",
       " 447: ('nazi germany', 'independent state of croatia'),\n",
       " 448: ('ibm', 'uts'),\n",
       " 449: ('galen', 'rome'),\n",
       " 450: ('eugene police department', 'eugene'),\n",
       " 451: ('united states department of veterans affairs',\n",
       "  \"harry s. truman memorial veterans ' hospital\"),\n",
       " 452: ('uda', 'imperial'),\n",
       " 453: ('lopez portillo', 'cortés'),\n",
       " 454: ('heidegger', 'nietzsche'),\n",
       " 455: ('angles', 'baltic sea'),\n",
       " 456: ('andré', 'jake \" the snake \" roberts'),\n",
       " 457: ('audi', 'auto union'),\n",
       " 458: ('cadillac', 'china'),\n",
       " 459: ('greenpeace', 'stuart oil shale project'),\n",
       " 460: ('bates', 'america the beautiful'),\n",
       " 461: ('altona', 'elbe'),\n",
       " 462: ('richard thompson', 'the washington post'),\n",
       " 463: ('crystal', 'rabbit test'),\n",
       " 464: ('ivan reitman', 'twins'),\n",
       " 465: ('cput', 'cape town'),\n",
       " 466: ('bowie', 'pontius pilate'),\n",
       " 467: ('robot novels', 'elijah baley'),\n",
       " 468: ('berlin', 'icc'),\n",
       " 469: ('american league', 'national league'),\n",
       " 470: ('finland', 'vantaa'),\n",
       " 471: ('ottoman empire', 'byzantine empire'),\n",
       " 472: ('abbasids', 'baghdad'),\n",
       " 473: ('cleveland', 'harrison'),\n",
       " 474: ('commodore', 'c64'),\n",
       " 475: ('ambrose', 'catechetical school of alexandria'),\n",
       " 476: ('joint stars', 'operation joint endeavor'),\n",
       " 477: ('frankish', 'frisians'),\n",
       " 478: ('haile selassie', 'oau'),\n",
       " 479: ('union mundial pro interlingua', 'interlingua'),\n",
       " 480: ('malcolm pasley', 'kafka'),\n",
       " 481: ('arnulf', 'italy'),\n",
       " 482: ('burj dubai', 'cn tower'),\n",
       " 483: ('skinner', 'chomsky'),\n",
       " 484: ('mathers', 'farr'),\n",
       " 485: ('ioc', 'fide'),\n",
       " 486: ('the rolling stones', 'a bigger bang tour'),\n",
       " 487: ('napoleon', 'jena'),\n",
       " 488: ('the cw', 'alexandria'),\n",
       " 489: ('henry', 'brittany'),\n",
       " 490: ('social democrats', 'nationalists'),\n",
       " 491: ('germania', 'otho'),\n",
       " 492: ('southeast', 'são paulo'),\n",
       " 493: ('devon and cornwall constabulary', 'exeter'),\n",
       " 494: ('nicolas louis de lacaille', 'fornax'),\n",
       " 495: ('crick', 'franklin'),\n",
       " 496: ('tucker', 'american'),\n",
       " 497: ('khorasan', 'balkh'),\n",
       " 498: ('william chambers', 'g. b. cipriani'),\n",
       " 499: ('flores', 'east nusa tenggara'),\n",
       " 500: ('audi 80', 'audi a4'),\n",
       " 501: ('ussr', 'mohammad najibullah'),\n",
       " 502: ('love for sale', 'belfast'),\n",
       " 503: ('khalid', 'damascus'),\n",
       " 504: ('boeing', 'yc-14'),\n",
       " 505: ('dionysus', 'agave'),\n",
       " 506: ('thompson',\n",
       "  'witness against the beast : william blake and the moral law'),\n",
       " 507: ('john ashcroft', 'earth liberation front'),\n",
       " 508: ('univac i', 'u.s. census bureau'),\n",
       " 509: ('gotham city', 'new york city'),\n",
       " 510: ('italian', 'eu'),\n",
       " 511: ('stilicho', 'western roman empire'),\n",
       " 512: ('richard aldington', 'world war i'),\n",
       " 513: ('elijah', 'israel'),\n",
       " 514: ('roman catholic church', 'timothy'),\n",
       " 515: ('lothair ii', 'staufen'),\n",
       " 516: ('ariane 5', 'esa'),\n",
       " 517: ('maimonides', 'mishneh torah'),\n",
       " 518: ('iec', 'iec'),\n",
       " 519: ('fox television stations', 'wfxt'),\n",
       " 520: ('ala.', 'doonesbury'),\n",
       " 521: ('rookwood', 'john wright'),\n",
       " 522: ('wallachian', 'ottomans'),\n",
       " 523: ('the late show', 'tonight'),\n",
       " 524: ('satellaview', 'radical dreamers : nusumenai hōseki'),\n",
       " 525: ('israel', 'sinai'),\n",
       " 526: ('haiti', 'haitian creole'),\n",
       " 527: ('zappa', 'zubin mehta'),\n",
       " 528: ('illinois territory', 'kaskaskia'),\n",
       " 529: ('ajax', 'psv eindhoven'),\n",
       " 530: ('tim caldwell', 'u.s.'),\n",
       " 531: ('gaza strip', 'hamas'),\n",
       " 532: ('johnny evers', 'chalmers award'),\n",
       " 533: ('bob boone', 'phillies'),\n",
       " 534: ('condé nast', 'cape breton island'),\n",
       " 535: ('drake', 'thomas doughty'),\n",
       " 536: ('soviet', 'kabul'),\n",
       " 537: ('poe', 'maryland'),\n",
       " 538: ('second coming of christ', 'christian eschatology'),\n",
       " 539: ('royal hobart regatta', 'tasmania'),\n",
       " 540: ('thomas aquinas', 'albertus'),\n",
       " 541: ('chetniks', 'montenegrins'),\n",
       " 542: ('doctor', 'susan foreman'),\n",
       " 543: ('azerbaijan national academy of sciences', 'azerbaijan'),\n",
       " 544: ('carthaginians', 'romans'),\n",
       " 545: ('imperial household agency', 'kōkō'),\n",
       " 546: ('apple iii', 'macintosh'),\n",
       " 547: ('commission', 'european union'),\n",
       " 548: ('helsinki', 'vantaa'),\n",
       " 549: ('apollo 15', 'apollo 12'),\n",
       " 550: ('todd williams', 'miguel cabrera'),\n",
       " 551: ('ray lewis', 'defensive player of the year'),\n",
       " 552: ('dao de jing', 'chinese classic text'),\n",
       " 553: ('scottish water', 'dundee'),\n",
       " 554: ('stalin', 'yalta'),\n",
       " 555: ('menger', 'vienna'),\n",
       " 556: ('furst', 'burton'),\n",
       " 557: ('articolo 31', 'italian'),\n",
       " 558: ('hampshire', 'atlantic'),\n",
       " 559: ('edwin austin abbey', 'american'),\n",
       " 560: ('elgar', 'brahms'),\n",
       " 561: ('imperial japanese navy', 'world war i'),\n",
       " 562: ('hathor', 'ra'),\n",
       " 563: ('pascal', 'evangelista torricelli'),\n",
       " 564: ('boston red stockings', 'nl central'),\n",
       " 565: ('ralph wilson stadium', 'buffalo'),\n",
       " 566: ('deng', 'china'),\n",
       " 567: ('heathrow', 'a30 road'),\n",
       " 568: ('nfl', 'nfc'),\n",
       " 569: ('czech republic', 'europe'),\n",
       " 570: ('bush', 'kosovo'),\n",
       " 571: ('ealdred', 'gloucester abbey'),\n",
       " 572: ('aaron', 'baseball hall of fame'),\n",
       " 573: ('astros', 'ken forsch'),\n",
       " 574: ('the apples', 'the apples in stereo'),\n",
       " 575: ('barry gray', 'ufo'),\n",
       " 576: ('burnette', 'bekka & billy'),\n",
       " 577: ('augustus', 'parthia'),\n",
       " 578: ('fsm', 'united states'),\n",
       " 579: ('adalbert', 'prague'),\n",
       " 580: ('hopper', 'easy rider'),\n",
       " 581: ('chancellorsville', 'northern virginia campaign'),\n",
       " 582: ('new york city landmarks preservation commission', 'hunter'),\n",
       " 583: ('j. michael straczynski', 'american'),\n",
       " 584: ('greece', 'europe'),\n",
       " 585: ('a song to remember', 'best actor'),\n",
       " 586: ('ælle', 'mons badonicus'),\n",
       " 587: ('charles garnier', 'paris opéra'),\n",
       " 588: ('cape town', 'cape dutch'),\n",
       " 589: ('ball state', 'indiana'),\n",
       " 590: ('us', 'm1128 stryker mgs'),\n",
       " 591: ('eleanor', 'ivanhoe'),\n",
       " 592: ('bordeaux', 'opéra national de bordeaux'),\n",
       " 593: ('fia formula two championship', 'f2'),\n",
       " 594: ('federal reserve', 'u.s.'),\n",
       " 595: ('french polynesians', 'nicolas sarkozy'),\n",
       " 596: ('honorius', 'constantius iii'),\n",
       " 597: ('apollo guidance computer', 'command module'),\n",
       " 598: ('kubrick', 'british academy of film and television arts'),\n",
       " 599: ('gurgenidze', 'parliament of georgia'),\n",
       " 600: ('french', 'scots'),\n",
       " 601: ('caledonian sleeper', 'london'),\n",
       " 602: ('schumacher', 'batman'),\n",
       " 603: ('cassander', 'alexander iv'),\n",
       " 604: ('yamunotri', 'char dham'),\n",
       " 605: ('best picture',\n",
       "  'academia mexicana de artes y ciencias cinematográficas'),\n",
       " 606: ('egypt', 'sinai peninsula'),\n",
       " 607: ('honda', '2006 hungarian grand prix'),\n",
       " 608: ('hubble', '1373 cincinnati'),\n",
       " 609: ('arsenal', 'first division'),\n",
       " 610: ('aberdeen', 'senior british open'),\n",
       " 611: ('cambodia', 'east asia summit'),\n",
       " 612: ('indian subcontinent', 'england'),\n",
       " 613: ('lucius', 'lucius verus'),\n",
       " 614: ('republic of cameroon', 'africa'),\n",
       " 615: ('germany', 'mackensen class'),\n",
       " 616: ('egypt', 'ottoman empire'),\n",
       " 617: ('elizabeth', 'france'),\n",
       " 618: ('toffler', 'heidi toffler'),\n",
       " 619: ('mugabe', 'harare'),\n",
       " 620: ('reagan', 'bethesda naval hospital'),\n",
       " 621: ('morris', 'emmy'),\n",
       " 622: ('ti', 'sr-10'),\n",
       " 623: ('chesterton', 'clarence darrow'),\n",
       " 624: ('hatfield', 'de havilland'),\n",
       " 625: ('mujib', 'bangladesh'),\n",
       " 626: ('tacitus', 'flavians'),\n",
       " 627: ('saturday morning', 'the curiosity shop'),\n",
       " 628: ('oddie', 'alan'),\n",
       " 629: ('hamburg', 'hamburg'),\n",
       " 630: ('little wars', 'floor games'),\n",
       " 631: ('milk river', 'saskatchewan'),\n",
       " 632: ('dsi', 'dsi xl'),\n",
       " 633: ('jackson', 'tennessee'),\n",
       " 634: ('cardiff city stadium', 'cardiff blues'),\n",
       " 635: ('felix frankfurter', 'supreme court of the united states'),\n",
       " 636: ('catalan parliament', 'penedès'),\n",
       " 637: ('tapti', 'arabian sea'),\n",
       " 638: ('finger lakes', 'iroquois'),\n",
       " 639: ('oaxaca', 'chiapas'),\n",
       " 640: ('soviet', 'petrozavodsk'),\n",
       " 641: ('heydrich', 'sa'),\n",
       " 642: ('rachel', 'joey'),\n",
       " 643: ('fbi', 'mccoy'),\n",
       " 644: ('wilfrid laurier', 'national policy'),\n",
       " 645: ('zogu', 'italy'),\n",
       " 646: ('kannauj', 'northern india'),\n",
       " 647: ('new york city landmarks preservation commission', 'south village'),\n",
       " 648: ('bobby heenan', 'hogan'),\n",
       " 649: ('aarhus', 'swansea'),\n",
       " 650: ('british airways', 'england'),\n",
       " 651: ('douglas adams', 'hhgg'),\n",
       " 652: ('tool time', 'home improvement'),\n",
       " 653: ('aberfoyle', 'loch lomond and the trossachs national park'),\n",
       " 654: ('central america', 'central american common market'),\n",
       " 655: ('john collins', 'cnd'),\n",
       " 656: ('fury in the slaughterhouse', 'hanover'),\n",
       " 657: ('dar es salaam', 'temeke'),\n",
       " 658: ('kournikova', 'leipzig'),\n",
       " 659: ('tecumseh', 'u.s.'),\n",
       " 660: ('german democratic republic', 'gdr'),\n",
       " 661: ('amazon', 'brazilian'),\n",
       " 662: ('australia', 'england'),\n",
       " 663: ('joseph h. lewis', 'the big combo'),\n",
       " 664: ('stalin', 'poland'),\n",
       " 665: ('sennacherib', 'jerusalem'),\n",
       " 666: ('phillip', 'gosford'),\n",
       " 667: ('kroeber', 'sigmund freud'),\n",
       " 668: ('beinn nibheis', 'british isles'),\n",
       " 669: ('ttpi', 'republic of palau'),\n",
       " 670: ('matisse', 'picasso'),\n",
       " 671: ('thompson', 'picture of che'),\n",
       " 672: ('kitase', 'final fantasy v'),\n",
       " 673: ('dc', 'the new teen titans'),\n",
       " 674: ('world cup', 'paraguay'),\n",
       " 675: ('philip the arab', 'rome'),\n",
       " 676: ('tim schafer', 'day of the tentacle'),\n",
       " 677: ('apple', 'apple iie'),\n",
       " 678: ('mumbai', 'indian ocean'),\n",
       " 679: ('uefa cup', 'chievo'),\n",
       " 680: ('ford gt', 'gt40'),\n",
       " 681: ('adobe systems incorporated', 'adobe systems incorporated'),\n",
       " 682: ('sts-93', 'coleman'),\n",
       " 683: ('diego garcia', 'maldives'),\n",
       " 684: ('der blaue reiter', 'marc'),\n",
       " 685: ('ryde', 'shanklin'),\n",
       " 686: ('disciples', 'christian church'),\n",
       " 687: ('charles lyell', 'principles of geology'),\n",
       " 688: ('claudia roth', 'angelika beer'),\n",
       " 689: ('robot series', 'foundation series'),\n",
       " 690: ('liechtenstein', 'german'),\n",
       " 691: ('august derleth', 'lovecraft'),\n",
       " 692: (\"the view 's\", 'uk'),\n",
       " 693: ('alaska', 'u.s'),\n",
       " 694: ('eureka springs', 'arkansas'),\n",
       " 695: ('frankfurt international airport', 'germany'),\n",
       " 696: ('colombia', '1994 fifa world cup'),\n",
       " 697: ('b-25', 'u.s.'),\n",
       " 698: ('f-14', 'a-6 intruder'),\n",
       " 699: ('land speed record', 'hüsker dü'),\n",
       " 700: ('iommi', 'black sabbath'),\n",
       " 701: ('amsterdam airport schiphol', 'amsterdam central station'),\n",
       " 702: ('bev bevan', 'terry chimes'),\n",
       " 703: ('lug lámfada', 'balor'),\n",
       " 704: ('balard', 'montpellier'),\n",
       " 705: ('otto preminger', \"lady windermere 's fan\"),\n",
       " 706: ('modell', 'ted marchibroda'),\n",
       " 707: ('cunimund', 'alboin'),\n",
       " 708: ('u.s.', 'nicaraguan contras'),\n",
       " 709: ('trasa kwiatkowskiego', 'a1 motorway'),\n",
       " 710: ('dessau', 'gropius'),\n",
       " 711: ('german navy', 'operation enduring freedom'),\n",
       " 712: ('haryana', 'northern india'),\n",
       " 713: ('syria', 'scud'),\n",
       " 714: ('lombards', 'cividale del friuli'),\n",
       " 715: ('sforza', 'savona'),\n",
       " 716: ('aquitanian', 'goths'),\n",
       " 717: ('commodore', 'vic-20'),\n",
       " 718: ('ainu', 'ainu language'),\n",
       " 719: ('darwin city council', 'city of darwin'),\n",
       " 720: ('fraser river', 'columbia'),\n",
       " 721: ('sinatra', 'francis a. & edward k.'),\n",
       " 722: ('new guinea', 'banda'),\n",
       " 723: ('gary brooker', 'a whiter shade of pale'),\n",
       " 724: ('labour', 'neville chamberlain'),\n",
       " 725: ('ethiopian church', 'ethiopia'),\n",
       " 726: ('confederate states of america', 'confederate congress'),\n",
       " 727: ('auburn road', 'buffalo'),\n",
       " 728: ('hungarian football federation', 'hungarian national football team'),\n",
       " 729: ('macintosh', 'apple ii'),\n",
       " 730: ('carrie underwood', 'islands in the stream'),\n",
       " 731: ('harvard', 'peirce'),\n",
       " 732: ('electronic frontier foundation', 'des'),\n",
       " 733: ('wumpus', 'trade wars'),\n",
       " 734: ('u.s.', 'israel'),\n",
       " 735: ('canary islands', 'spain'),\n",
       " 736: ('innocent x', 'bernini'),\n",
       " 737: ('maradona', 'buenos aires'),\n",
       " 738: ('mercury', 'castor'),\n",
       " 739: ('beja people', 'arabic'),\n",
       " 740: ('james princep', 'girnar'),\n",
       " 741: ('emperor simeon i the great', 'bulgaria'),\n",
       " 742: ('china', 'deng xiaoping'),\n",
       " 743: ('gabon', 'lambaréné'),\n",
       " 744: ('microsoft', 'windows 95'),\n",
       " 745: ('brunei', 'u.s.'),\n",
       " 746: ('u.k.', 'world war ii'),\n",
       " 747: ('dante bichette', 'cincinnati reds'),\n",
       " 748: ('baghdad', 'persian gulf'),\n",
       " 749: ('panthers', 'nfl'),\n",
       " 750: ('european union', 'africa'),\n",
       " 751: ('vespucci', 'southern cross'),\n",
       " 752: ('the mutants', 'doctor who'),\n",
       " 753: ('commodore', 'ford falcon'),\n",
       " 754: ('napoleon i', 'hamburg'),\n",
       " 755: ('ibm', 'microsoft'),\n",
       " 756: ('alexandria', 'scotland'),\n",
       " 757: ('eastern league', 'national league'),\n",
       " 758: ('bodmin college', 'bodmin moor'),\n",
       " 759: ('columbia', 'rutgers'),\n",
       " 760: ('anders sunesen', 'elbe river'),\n",
       " 761: ('palestine', 'ottomans'),\n",
       " 762: ('arthur', 'caledfwlch'),\n",
       " 763: ('accrington', 'accy'),\n",
       " 764: ('stuart highway', 'palmerston'),\n",
       " 765: ('sicily', 'messana'),\n",
       " 766: ('gestapo', 'poland'),\n",
       " 767: ('constantius', 'shapur ii'),\n",
       " 768: ('the daleks', 'dalek'),\n",
       " 769: ('rudolf caracciola', 'mercedes ssk'),\n",
       " 770: ('alvaro colom', 'pp'),\n",
       " 771: ('robert mugabe', 'zimbabwe'),\n",
       " 772: ('new south wales', 'australia'),\n",
       " 773: ('hatfield', 'welwyn hatfield'),\n",
       " 774: ('antony', 'caesar'),\n",
       " 775: ('bush', 'ronald reagan'),\n",
       " 776: ('india', 'british'),\n",
       " 777: ('al taliaferro', 'donald duck'),\n",
       " 778: ('honda', 'hero honda'),\n",
       " 779: ('heracles', 'argonauts'),\n",
       " 780: ('chatti', 'saturninus'),\n",
       " 781: ('gcc', 'the cathedral and the bazaar'),\n",
       " 782: ('north end', 'whitgift centre'),\n",
       " 783: ('gnome press', 'foundation trilogy'),\n",
       " 784: ('italy', 'roman catholic'),\n",
       " 785: ('southern illinois miners', 'frontier league'),\n",
       " 786: ('flambards', 'english'),\n",
       " 787: ('grimoald', 'frankish'),\n",
       " 788: ('algiers', 'france'),\n",
       " 789: ('broadway', '23rd street'),\n",
       " 790: ('egypt', 'gaza strip'),\n",
       " 791: ('bainimarama', 'jona senilagakali'),\n",
       " 792: ('west exe technology college', 'devon'),\n",
       " 793: ('captain kentucky', 'lancelot pertwillaby'),\n",
       " 794: ('bosnia and hercegovina', 'eu'),\n",
       " 795: ('eric & us', 'blair'),\n",
       " 796: ('speer', 'hitler'),\n",
       " 797: ('lisvane', 'wales'),\n",
       " 798: ('spears', 'sean preston federline'),\n",
       " 799: ('atreus', 'thyestes'),\n",
       " 800: ('jews', 'limoges'),\n",
       " 801: ('cambridge', 'u.s.'),\n",
       " 802: ('freyr', 'gerðr'),\n",
       " 803: ('frank inn', 'green acres'),\n",
       " 804: ('artemisia gentileschi', 'caravaggio'),\n",
       " 805: ('rosa', 'disney'),\n",
       " 806: ('demetrios vikelas', 'switzerland'),\n",
       " 807: ('atlantic ocean', 'gironde estuary'),\n",
       " 808: ('darwin', 'hidden valley raceway'),\n",
       " 809: ('israeli government', 'druze'),\n",
       " 810: ('elizabeth i', 'ireland'),\n",
       " 811: ('salvador allende', 'kgb'),\n",
       " 812: (\"mu'awiyah\", 'ali'),\n",
       " 813: ('darwin', 'galton'),\n",
       " 814: ('borg', 'guillermo vilas'),\n",
       " 815: ('steinbrenner', 'new york yankees'),\n",
       " 816: ('ark', 'shiloh'),\n",
       " 817: ('ney', 'wellington'),\n",
       " 818: ('vespucci', 'christopher columbus'),\n",
       " 819: ('esperanto', 'george soros'),\n",
       " 820: ('faroese language', 'faroese'),\n",
       " 821: ('alan rickman', 'emmy'),\n",
       " 822: ('hamburg freezers', 'hamburg'),\n",
       " 823: ('green river', 'colo.'),\n",
       " 824: ('babylonians', 'assyria'),\n",
       " 825: ('danish', 'kingdom of iceland'),\n",
       " 826: ('washington', 'washington'),\n",
       " 827: ('university of limerick', 'national institute for higher education'),\n",
       " 828: ('california', 'democratic'),\n",
       " 829: ('golden dawn', 'aleister crowley'),\n",
       " 830: ('gorillaz', 'coachella music festival'),\n",
       " 831: ('solid snake', 'snake plissken'),\n",
       " 832: ('city of san francisco', 'marin county'),\n",
       " 833: ('forbes', 'dayton'),\n",
       " 834: ('adelaide', 'government of south australia'),\n",
       " 835: ('tippi hedren', 'rod taylor'),\n",
       " 836: ('apple ii series', 'apple'),\n",
       " 837: ('saint paul', 'new testament'),\n",
       " 838: ('cabral', 'santa maria'),\n",
       " 839: ('nightwish', 'finland'),\n",
       " 840: ('numbers', 'iwork'),\n",
       " 841: ('russi taylor', 'louie'),\n",
       " 842: ('jim thome', 'phillies'),\n",
       " 843: ('persian', 'burzoe'),\n",
       " 844: ('thetis', 'achilles'),\n",
       " 845: ('belle and sebastian', 'nick drake'),\n",
       " 846: ('n.y.', 'southwestern connecticut'),\n",
       " 847: ('archimedes', 'syracuse'),\n",
       " 848: ('air canada', 'halifax'),\n",
       " 849: ('history of the russian federation', 'soviet union'),\n",
       " 850: ('jerry narron', 'pete mackanin'),\n",
       " 851: ('china', 'yangtze river'),\n",
       " 852: ('heraklion', 'greece'),\n",
       " 853: ('fifa', 'maradona'),\n",
       " 854: ('atlanta', 'kasim reed'),\n",
       " 855: ('rafael perestrello', 'china'),\n",
       " 856: ('russia', 'bangladesh'),\n",
       " 857: ('cliff bastin', 'english football'),\n",
       " 858: ('claudius', 'nero'),\n",
       " 859: ('bill murto', 'compaq'),\n",
       " 860: ('condor games', 'diablo'),\n",
       " 861: ('red army', 'stalingrad'),\n",
       " 862: ('jackson', 'mississippi'),\n",
       " 863: ('claudius', 'laertes'),\n",
       " 864: ('william bradford shockley', 'bell laboratories'),\n",
       " 865: ('final crisis', 'darkseid'),\n",
       " 866: ('hebrides', 'scotland'),\n",
       " 867: ('gardiner', 'salisbury'),\n",
       " 868: ('neil stubenhaus', 'berklee college'),\n",
       " 869: ('soviet il-28', 'russia'),\n",
       " 870: ('macbook', 'sudden motion sensor'),\n",
       " 871: ('houston texans', 'afc'),\n",
       " 872: ('hilter', 'germany'),\n",
       " 873: ('batman', 'clark kent'),\n",
       " 874: ('amtrak', 'union station'),\n",
       " 875: ('gilbert plains railway station', 'via rail'),\n",
       " 876: ('south east england', 'greater london'),\n",
       " 877: ('concord', 'united states'),\n",
       " 878: ('william s. harley', 'university of wisconsin -- madison'),\n",
       " 879: ('dumnonii', 'dumnonos'),\n",
       " 880: ('u.k.', 'europe'),\n",
       " 881: ('clinton', 'clinton bush haiti fund'),\n",
       " 882: ('pedro', 'sicily'),\n",
       " 883: ('random house',\n",
       "  \"hell 's angels : the strange and terrible saga of the outlaw motorcycle gangs\"),\n",
       " 884: ('international olympic committee', 'chicago'),\n",
       " 885: ('marlins', 'detroit tigers'),\n",
       " 886: ('beverly lynn burns', 'aa'),\n",
       " 887: ('justin broadrick', 'godflesh'),\n",
       " 888: ('teach', \"queen anne 's revenge\"),\n",
       " 889: ('luke', \"jesus '\"),\n",
       " 890: ('cleveland', 'democratic party'),\n",
       " 891: ('cronenberg', 'the fly'),\n",
       " 892: ('tomcat', 'soviet'),\n",
       " 893: ('rodrigue beaubois', 'guadeloupe'),\n",
       " 894: ('united states', 'iraq'),\n",
       " 895: ('matt le tissier', 'england'),\n",
       " 896: ('barack obama', 'republican'),\n",
       " 897: ('masato kato', 'chrono trigger'),\n",
       " 898: ('coolidge', 'new england'),\n",
       " 899: ('john philip kemble', 'drury lane'),\n",
       " 900: ('chicago metropolitan area', 'fortune 500'),\n",
       " 901: ('raaf', 'p-40'),\n",
       " 902: ('bronx', 'lehman college'),\n",
       " 903: ('university of oxford', 'challenger expedition'),\n",
       " 904: ('constantius', 'galla placidia'),\n",
       " 905: ('chrono trigger', 'final fantasy vii'),\n",
       " 906: ('rich harden', 'oakland athletics'),\n",
       " 907: ('aaron', 'braves'),\n",
       " 908: ('fiorentina', 'serie b'),\n",
       " 909: ('athenians', 'macedonia'),\n",
       " 910: ('ray knight', 'mets'),\n",
       " 911: ('charlton', 'premier league'),\n",
       " 912: ('medea', 'pelias'),\n",
       " 913: ('chicago', 'chicago democratic machine'),\n",
       " 914: ('moses', 'joshua'),\n",
       " 915: ('england', 'roman catholic stuart'),\n",
       " 916: ('transparency international', 'finland'),\n",
       " 917: ('brazil', 'colombia'),\n",
       " 918: ('delphi', 'pythian games'),\n",
       " 919: ('bronx', 'manhattan'),\n",
       " 920: ('dlr', 'greenwich'),\n",
       " 921: ('trent reznor', 'the hearts filthy lesson'),\n",
       " 922: ('france', 'port st. louis'),\n",
       " 923: ('bush', 'west texas'),\n",
       " 924: ('disney', 'the walt disney company'),\n",
       " 925: ('ma and pa kettle', 'american'),\n",
       " 926: ('davy', 'faraday'),\n",
       " 927: ('portugal', 'león'),\n",
       " 928: ('gialloblu', 'verona'),\n",
       " 929: ('danny blind', 'koeman'),\n",
       " 930: ('canada', 'us'),\n",
       " 931: ('bobby charlton', 'david herd'),\n",
       " 932: ('amal', 'katyusha'),\n",
       " 933: ('angelo', 'inter'),\n",
       " 934: ('semele', 'zeus'),\n",
       " 935: ('cpr', 'powder river basin'),\n",
       " 936: ('sebastián vizcaíno', 'new spain'),\n",
       " 937: ('hbc', 'northern california'),\n",
       " 938: ('michael wisher', 'death to the daleks'),\n",
       " 939: ('canadian', 'hal 9000'),\n",
       " 940: ('bindusara', 'ashoka'),\n",
       " 941: ('cypriot national guard', 'republic of cyprus'),\n",
       " 942: ('tiberius', 'carthage'),\n",
       " 943: ('b-29', 'paul tibbets'),\n",
       " 944: ('east germany', 'egon krenz'),\n",
       " 945: ('vkhutemas', 'german'),\n",
       " 946: ('genesis', 'el'),\n",
       " 947: ('eurotunnel', 'eurostar'),\n",
       " 948: ('guangzhou', 'panyu'),\n",
       " 949: ('nxai pan national park', 'central district'),\n",
       " 950: ('galileo', \"cosimo de ' medici\"),\n",
       " 951: ('church of england', 'king charles the martyr'),\n",
       " 952: ('italy', 'ottoman empire'),\n",
       " 953: ('paul weiss', 'peirce'),\n",
       " 954: ('university of tasmania', 'sandy bay'),\n",
       " 955: ('chinese nationalist party', 'qing'),\n",
       " 956: ('guantanamo', 'cuba'),\n",
       " 957: ('sighthill', 'old town'),\n",
       " 958: ('houston oilers', 'enron field'),\n",
       " 959: ('black sea', 'russia'),\n",
       " 960: ('hurricane janet', 'corozal'),\n",
       " 961: ('soviet', 'west berlin'),\n",
       " 962: ('roman catholic howard', 'catherine'),\n",
       " 963: ('peter underwood', 'ghosts of borley'),\n",
       " 964: ('beckham', '2010 fifa world cup'),\n",
       " 965: ('manaus', 'amazon'),\n",
       " 966: ('earnhardt', 'rcr'),\n",
       " 967: ('republicans', 'blaine'),\n",
       " 968: ('berlin', 'capital of europe'),\n",
       " 969: ('ukraine', 'europe'),\n",
       " 970: ('rtd', 'denver metropolitan area'),\n",
       " 971: ('russia', 'east slavs'),\n",
       " 972: ('j. p. morgan', 'max shreck'),\n",
       " 973: ('david', 'goliath'),\n",
       " 974: ('napoleon i', 'eugene'),\n",
       " 975: ('babbage', 'frederick winslow taylor'),\n",
       " 976: ('hamas', 'gaza'),\n",
       " 977: ('assur', 'asia minor'),\n",
       " 978: ('montgomery', 'american'),\n",
       " 979: ('poindexter', 'frank carlucci'),\n",
       " 980: (\"l'ouverture\", 'french republic'),\n",
       " 981: ('mátyás szűrös', 'hungary'),\n",
       " 982: ('irish military', 'nato'),\n",
       " 983: ('tiwi', 'northern territory'),\n",
       " 984: ('lions', 'west coast eagles'),\n",
       " 985: ('artemis', 'brauron'),\n",
       " 986: ('wings', 'nhl'),\n",
       " 987: ('hannibal', 'carthage'),\n",
       " 988: ('arne næss', 'spinoza'),\n",
       " 989: ('chopin', 'french'),\n",
       " 990: ('diderot', 'rosenkranz'),\n",
       " 991: ('blizzard', 'world of warcraft'),\n",
       " 992: ('karen gillan', 'amy pond'),\n",
       " 993: ('huxley', 'lawrence'),\n",
       " 994: ('munich', 'jewish'),\n",
       " 995: ('westchester county', 'bronx'),\n",
       " 996: ('george gamow', 'rna tie club'),\n",
       " 997: ('narses', 'lombards'),\n",
       " 998: ('hannah storm', 'costas'),\n",
       " 999: ('evers', 'fayette'),\n",
       " ...}"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "id2args"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "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>(boc tower, hong kong)</th>\n",
       "      <th>(chaplin, chaplin)</th>\n",
       "      <th>(walter gropius, bauhaus)</th>\n",
       "      <th>(hawaii state supreme court, ali ʻ iōlani hale)</th>\n",
       "      <th>(thurgot, saint andrews)</th>\n",
       "      <th>(santiniketan, indian)</th>\n",
       "      <th>(goldwater, reagan)</th>\n",
       "      <th>(dostum, rabbani)</th>\n",
       "      <th>(coloman, béla)</th>\n",
       "      <th>(rockies, oakland athletics)</th>\n",
       "      <th>...</th>\n",
       "      <th>(bali strait, java)</th>\n",
       "      <th>(brown, sarah macaulay)</th>\n",
       "      <th>(statenvertaling, dutch)</th>\n",
       "      <th>(sagan, aaas 's)</th>\n",
       "      <th>(spain, coimbra)</th>\n",
       "      <th>(france, cascade mountains)</th>\n",
       "      <th>(love, brian wilson)</th>\n",
       "      <th>(lenin, reds)</th>\n",
       "      <th>(kelly, daisy bell)</th>\n",
       "      <th>(mclean, lisa kindred)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>star-PER-MISC</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>star-PER-PER</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.38</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>star-ORG-PER</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.62</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>star-MISC-MISC</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>star-MISC-PER</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.22</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 14859 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                (boc tower, hong kong)  (chaplin, chaplin)  \\\n",
       "star-PER-MISC                      0.0                 0.0   \n",
       "star-PER-PER                       0.0                 1.0   \n",
       "star-ORG-PER                       0.0                 0.0   \n",
       "star-MISC-MISC                     0.0                 0.0   \n",
       "star-MISC-PER                      0.0                 0.0   \n",
       "\n",
       "                (walter gropius, bauhaus)  \\\n",
       "star-PER-MISC                         0.0   \n",
       "star-PER-PER                          0.0   \n",
       "star-ORG-PER                          0.0   \n",
       "star-MISC-MISC                        0.0   \n",
       "star-MISC-PER                         0.0   \n",
       "\n",
       "                (hawaii state supreme court, ali ʻ iōlani hale)  \\\n",
       "star-PER-MISC                                               0.0   \n",
       "star-PER-PER                                                0.0   \n",
       "star-ORG-PER                                                0.0   \n",
       "star-MISC-MISC                                              0.0   \n",
       "star-MISC-PER                                               0.0   \n",
       "\n",
       "                (thurgot, saint andrews)  (santiniketan, indian)  \\\n",
       "star-PER-MISC                        0.0                     0.0   \n",
       "star-PER-PER                         0.0                     0.0   \n",
       "star-ORG-PER                         0.0                     0.0   \n",
       "star-MISC-MISC                       0.0                     0.0   \n",
       "star-MISC-PER                        0.0                     0.0   \n",
       "\n",
       "                (goldwater, reagan)  (dostum, rabbani)  (coloman, béla)  \\\n",
       "star-PER-MISC                   0.0                0.0              0.0   \n",
       "star-PER-PER                    1.0                1.0              1.0   \n",
       "star-ORG-PER                    0.0                0.0              0.0   \n",
       "star-MISC-MISC                  0.0                0.0              0.0   \n",
       "star-MISC-PER                   0.0                0.0              0.0   \n",
       "\n",
       "                (rockies, oakland athletics)  ...  (bali strait, java)  \\\n",
       "star-PER-MISC                            0.0  ...                  0.0   \n",
       "star-PER-PER                             0.0  ...                  0.0   \n",
       "star-ORG-PER                             0.0  ...                  0.0   \n",
       "star-MISC-MISC                           0.0  ...                  0.0   \n",
       "star-MISC-PER                            0.0  ...                  0.0   \n",
       "\n",
       "                (brown, sarah macaulay)  (statenvertaling, dutch)  \\\n",
       "star-PER-MISC                      0.00                       0.0   \n",
       "star-PER-PER                       0.38                       0.0   \n",
       "star-ORG-PER                       0.62                       0.0   \n",
       "star-MISC-MISC                     0.00                       1.0   \n",
       "star-MISC-PER                      0.00                       0.0   \n",
       "\n",
       "                (sagan, aaas 's)  (spain, coimbra)  \\\n",
       "star-PER-MISC                0.0               0.0   \n",
       "star-PER-PER                 0.0               0.0   \n",
       "star-ORG-PER                 0.0               0.0   \n",
       "star-MISC-MISC               0.0               0.0   \n",
       "star-MISC-PER                0.0               0.0   \n",
       "\n",
       "                (france, cascade mountains)  (love, brian wilson)  \\\n",
       "star-PER-MISC                           0.0                  0.00   \n",
       "star-PER-PER                            0.0                  0.78   \n",
       "star-ORG-PER                            0.0                  0.00   \n",
       "star-MISC-MISC                          0.0                  0.00   \n",
       "star-MISC-PER                           0.0                  0.22   \n",
       "\n",
       "                (lenin, reds)  (kelly, daisy bell)  (mclean, lisa kindred)  \n",
       "star-PER-MISC             0.0                  1.0                     0.0  \n",
       "star-PER-PER              0.0                  0.0                     1.0  \n",
       "star-ORG-PER              0.0                  0.0                     0.0  \n",
       "star-MISC-MISC            0.0                  0.0                     0.0  \n",
       "star-MISC-PER             0.0                  0.0                     0.0  \n",
       "\n",
       "[5 rows x 14859 columns]"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_df = pd.DataFrame(features)\n",
    "features_df = features_df.rename(columns= id2args)\n",
    "features_df = features_df.rename(index= id2label)\n",
    "features_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "features_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Augmented data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X = StandardScaler().fit_transform(features)\n",
    "\n",
    "\n",
    "pca = PCA(n_components=16)\n",
    "x_7d = pca.fit_transform(X)\n",
    "\n",
    "pca4 = PCA(n_components=4)\n",
    "x_4d = pca.fit_transform(X)\n",
    "\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 3)\n",
    "\n",
    "#Compute cluster centers and predict cluster indices\n",
    "X_clustered = kmeans.fit_predict(x_7d)\n",
    "\n",
    "#Define our own color map\n",
    "LABEL_COLOR_MAP = {0:'r', 1: 'g', 2: 'b'}\n",
    "label_color = [LABEL_COLOR_MAP[l] for l in X_clustered]\n",
    "\n",
    "# Plot the scatter digram\n",
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_7d[:,0],x_7d[:,3], c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X = StandardScaler().fit_transform(features_df)\n",
    "\n",
    "pca4 = PCA(n_components=16)\n",
    "x_4d = pca.fit_transform(X)\n",
    "\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 4)\n",
    "\n",
    "X_clustered= kmeans.fit(x_4d)\n",
    "cluster_map = pd.DataFrame()\n",
    "cluster_map['data_index']=features_df.index.values\n",
    "cluster_map['cluster']=X_clustered.labels_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "X = StandardScaler().fit_transform(features_df)\n",
    "\n",
    "pca4 = PCA(n_components=16)\n",
    "x_4d = pca.fit_transform(X)\n",
    "\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 4)\n",
    "\n",
    "#Define our own color map\n",
    "X_clustered = kmeans.fit_predict(x_4d)\n",
    "LABEL_COLOR_MAP = {0:'r', 1: 'g', 2: 'b', 3:'orange'}\n",
    "label_color = [LABEL_COLOR_MAP[l] for l in X_clustered]\n",
    "\n",
    "# Plot the scatter digram\n",
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,2],c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,1],c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,3],c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,4],c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,5],c= label_color, alpha=0.5, cmap='viridis') \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4560, 16)"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_4d.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                 data_index  cluster\n",
      "16               be-LOC-LOC        0\n",
      "35      established-LOC-LOC        0\n",
      "48            carry-LOC-LOC        0\n",
      "70             give-LOC-LOC        0\n",
      "74             hold-LOC-LOC        0\n",
      "93             have-LOC-LOC        0\n",
      "106         consist-LOC-LOC        0\n",
      "121            play-LOC-LOC        0\n",
      "129          remain-LOC-LOC        0\n",
      "145            lose-LOC-LOC        0\n",
      "161            beat-LOC-LOC        0\n",
      "165           clash-LOC-LOC        0\n",
      "173       establish-LOC-LOC        0\n",
      "185           leave-LOC-LOC        0\n",
      "196     incorporate-LOC-LOC        0\n",
      "213            lead-LOC-LOC        0\n",
      "233           write-LOC-LOC        0\n",
      "244           serve-LOC-LOC        0\n",
      "300          follow-LOC-LOC        0\n",
      "315           offer-LOC-LOC        0\n",
      "322        separate-LOC-LOC        0\n",
      "331          manage-LOC-LOC        0\n",
      "335           cause-LOC-LOC        0\n",
      "341            link-LOC-LOC        0\n",
      "348          become-LOC-LOC        0\n",
      "365         provide-LOC-LOC        0\n",
      "377        continue-LOC-LOC        0\n",
      "390            form-LOC-LOC        0\n",
      "410         feature-LOC-LOC        0\n",
      "417      contribute-LOC-LOC        0\n",
      "...                     ...      ...\n",
      "4173     categorize-LOC-LOC        0\n",
      "4174           rent-LOC-LOC        0\n",
      "4177           rest-LOC-LOC        0\n",
      "4196         double-LOC-LOC        0\n",
      "4204         centre-LOC-LOC        0\n",
      "4209         depart-LOC-LOC        0\n",
      "4211        resolve-LOC-LOC        0\n",
      "4217       resupply-LOC-LOC        0\n",
      "4257   transmigrate-LOC-LOC        0\n",
      "4284            cut-LOC-LOC        0\n",
      "4303         resign-LOC-LOC        0\n",
      "4312      catalogue-LOC-LOC        0\n",
      "4318           line-LOC-LOC        0\n",
      "4320            war-LOC-LOC        0\n",
      "4330      reinforce-LOC-LOC        0\n",
      "4331       encircle-LOC-LOC        0\n",
      "4351         switch-LOC-LOC        0\n",
      "4393  transplanting-LOC-LOC        0\n",
      "4395           flow-LOC-LOC        0\n",
      "4406     endangered-LOC-LOC        0\n",
      "4420        bracket-LOC-LOC        0\n",
      "4425      protected-LOC-LOC        0\n",
      "4441        sheathe-LOC-LOC        0\n",
      "4467         blight-LOC-LOC        0\n",
      "4481          store-LOC-LOC        0\n",
      "4491   rehabilitate-LOC-LOC        0\n",
      "4493       inundate-LOC-LOC        0\n",
      "4497            peg-LOC-LOC        0\n",
      "4514     anglicised-LOC-LOC        0\n",
      "4530     trailblaze-LOC-LOC        0\n",
      "\n",
      "[425 rows x 2 columns]\n",
      "                 data_index  cluster\n",
      "2              star-ORG-PER        1\n",
      "3            star-MISC-MISC        1\n",
      "4             star-MISC-PER        1\n",
      "5             star-LOC-MISC        1\n",
      "6              star-PER-LOC        1\n",
      "7              star-PER-ORG        1\n",
      "8                be-ORG-ORG        1\n",
      "9               be-LOC-MISC        1\n",
      "11               be-PER-LOC        1\n",
      "12               be-ORG-LOC        1\n",
      "14               be-PER-ORG        1\n",
      "15             be-MISC-MISC        1\n",
      "17              be-MISC-PER        1\n",
      "18               be-LOC-ORG        1\n",
      "19               be-LOC-PER        1\n",
      "20              be-MISC-LOC        1\n",
      "21              be-ORG-MISC        1\n",
      "22               be-ORG-PER        1\n",
      "23              be-MISC-ORG        1\n",
      "24          appoint-ORG-PER        1\n",
      "25          appoint-PER-ORG        1\n",
      "27          appoint-PER-LOC        1\n",
      "28         appoint-MISC-PER        1\n",
      "30      established-ORG-ORG        1\n",
      "31      established-PER-LOC        1\n",
      "32     established-MISC-LOC        1\n",
      "33     established-ORG-MISC        1\n",
      "34      established-LOC-ORG        1\n",
      "36    established-MISC-MISC        1\n",
      "37     established-MISC-ORG        1\n",
      "...                     ...      ...\n",
      "4517         reside-ORG-LOC        1\n",
      "4518    promulgate-LOC-MISC        1\n",
      "4519         notify-PER-ORG        1\n",
      "4521         purify-PER-LOC        1\n",
      "4523    prescribed-LOC-MISC        1\n",
      "4527    orchestrate-PER-LOC        1\n",
      "4528       plotting-PER-LOC        1\n",
      "4529       christen-ORG-LOC        1\n",
      "4531          judge-PER-LOC        1\n",
      "4533          land-ORG-MISC        1\n",
      "4534     corrupted-MISC-LOC        1\n",
      "4536       postpone-LOC-PER        1\n",
      "4537   substitute-MISC-MISC        1\n",
      "4538        slate-MISC-MISC        1\n",
      "4540      hitchhike-PER-LOC        1\n",
      "4543    proposition-PER-ORG        1\n",
      "4544    patterned-MISC-MISC        1\n",
      "4545        elevate-PER-LOC        1\n",
      "4546        molding-ORG-ORG        1\n",
      "4548         lure-MISC-MISC        1\n",
      "4549       compound-PER-ORG        1\n",
      "4550        reuse-MISC-MISC        1\n",
      "4551   administrate-ORG-LOC        1\n",
      "4553       nurtures-ORG-ORG        1\n",
      "4554        justify-ORG-LOC        1\n",
      "4555         harry-LOC-MISC        1\n",
      "4556     liberalize-PER-LOC        1\n",
      "4557           stir-PER-LOC        1\n",
      "4558         load-MISC-MISC        1\n",
      "4559           raid-ORG-ORG        1\n",
      "\n",
      "[3053 rows x 2 columns]\n",
      "                data_index  cluster\n",
      "0            star-PER-MISC        2\n",
      "10             be-PER-MISC        2\n",
      "29        appoint-PER-MISC        2\n",
      "38    established-PER-MISC        2\n",
      "52          carry-PER-MISC        2\n",
      "65           give-PER-MISC        2\n",
      "79           hold-PER-MISC        2\n",
      "85        express-PER-MISC        2\n",
      "95           have-PER-MISC        2\n",
      "105       consist-PER-MISC        2\n",
      "116          play-PER-MISC        2\n",
      "138        remain-PER-MISC        2\n",
      "153          lose-PER-MISC        2\n",
      "174     establish-PER-MISC        2\n",
      "176         leave-PER-MISC        2\n",
      "194   incorporate-PER-MISC        2\n",
      "202        employ-PER-MISC        2\n",
      "214          lead-PER-MISC        2\n",
      "226         write-PER-MISC        2\n",
      "248         serve-PER-MISC        2\n",
      "262          bear-PER-MISC        2\n",
      "265     introduce-PER-MISC        2\n",
      "273       condemn-PER-MISC        2\n",
      "283        oppose-PER-MISC        2\n",
      "286        repeal-PER-MISC        2\n",
      "287       deliver-PER-MISC        2\n",
      "302        follow-PER-MISC        2\n",
      "310         offer-PER-MISC        2\n",
      "333        manage-PER-MISC        2\n",
      "337         cause-PER-MISC        2\n",
      "...                    ...      ...\n",
      "4384  superimpose-PER-MISC        2\n",
      "4385    subjugate-PER-MISC        2\n",
      "4392          fax-PER-MISC        2\n",
      "4397    frontline-PER-MISC        2\n",
      "4400     manifest-PER-MISC        2\n",
      "4401   ghostwrite-PER-MISC        2\n",
      "4407       remake-PER-MISC        2\n",
      "4409      cofound-PER-MISC        2\n",
      "4413         cure-PER-MISC        2\n",
      "4416     position-PER-MISC        2\n",
      "4428       author-PER-MISC        2\n",
      "4429      besiege-PER-MISC        2\n",
      "4435       garner-PER-MISC        2\n",
      "4439        prove-PER-MISC        2\n",
      "4445   displeased-PER-MISC        2\n",
      "4454     retrieve-PER-MISC        2\n",
      "4465       ordain-PER-MISC        2\n",
      "4471     reassure-PER-MISC        2\n",
      "4479       exempt-PER-MISC        2\n",
      "4488   confiscate-PER-MISC        2\n",
      "4495        train-PER-MISC        2\n",
      "4503          hug-PER-MISC        2\n",
      "4512      specify-PER-MISC        2\n",
      "4520         seed-PER-MISC        2\n",
      "4522     resettle-PER-MISC        2\n",
      "4524        rouse-PER-MISC        2\n",
      "4535     overlaid-PER-MISC        2\n",
      "4539         leak-PER-MISC        2\n",
      "4542        grace-PER-MISC        2\n",
      "4552          pen-PER-MISC        2\n",
      "\n",
      "[417 rows x 2 columns]\n",
      "                data_index  cluster\n",
      "1             star-PER-PER        3\n",
      "13              be-PER-PER        3\n",
      "26         appoint-PER-PER        3\n",
      "42     established-PER-PER        3\n",
      "69            give-PER-PER        3\n",
      "81            hold-PER-PER        3\n",
      "90            have-PER-PER        3\n",
      "108        consist-PER-PER        3\n",
      "115           play-PER-PER        3\n",
      "139         remain-PER-PER        3\n",
      "148           lose-PER-PER        3\n",
      "158           beat-PER-PER        3\n",
      "166          clash-PER-PER        3\n",
      "180          leave-PER-PER        3\n",
      "191           bury-PER-PER        3\n",
      "193    incorporate-PER-PER        3\n",
      "208         employ-PER-PER        3\n",
      "221           lead-PER-PER        3\n",
      "227          write-PER-PER        3\n",
      "245          serve-PER-PER        3\n",
      "257           bear-PER-PER        3\n",
      "264      introduce-PER-PER        3\n",
      "281         oppose-PER-PER        3\n",
      "288        deliver-PER-PER        3\n",
      "295         follow-PER-PER        3\n",
      "308          offer-PER-PER        3\n",
      "323       separate-PER-PER        3\n",
      "339          cause-PER-PER        3\n",
      "343           link-PER-PER        3\n",
      "353         become-PER-PER        3\n",
      "...                    ...      ...\n",
      "4410       narrate-PER-PER        3\n",
      "4411          grab-PER-PER        3\n",
      "4412        shower-PER-PER        3\n",
      "4421         alert-PER-PER        3\n",
      "4423          fare-PER-PER        3\n",
      "4424     protected-PER-PER        3\n",
      "4436         lunge-PER-PER        3\n",
      "4446       relayed-PER-PER        3\n",
      "4447  outgeneraled-PER-PER        3\n",
      "4449    relinquish-PER-PER        3\n",
      "4451       singing-PER-PER        3\n",
      "4452   reinterpret-PER-PER        3\n",
      "4460         style-PER-PER        3\n",
      "4466     subscribe-PER-PER        3\n",
      "4469         spell-PER-PER        3\n",
      "4470       imitate-PER-PER        3\n",
      "4473        poison-PER-PER        3\n",
      "4475       refined-PER-PER        3\n",
      "4478      disguise-PER-PER        3\n",
      "4483        evince-PER-PER        3\n",
      "4486    rebuilding-PER-PER        3\n",
      "4499      abdicate-PER-PER        3\n",
      "4506    partnering-PER-PER        3\n",
      "4507         brief-PER-PER        3\n",
      "4509       mislead-PER-PER        3\n",
      "4525      instruct-PER-PER        3\n",
      "4526        seduce-PER-PER        3\n",
      "4532        pepper-PER-PER        3\n",
      "4541       lampoon-PER-PER        3\n",
      "4547         greet-PER-PER        3\n",
      "\n",
      "[665 rows x 2 columns]\n"
     ]
    }
   ],
   "source": [
    "for i in range(0, 4):\n",
    "    print(cluster_map[cluster_map.cluster==i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## analysis explained variance measure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X = StandardScaler().fit_transform(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4560, 14859)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_vec= np.mean(X, axis=0)\n",
    "cov_mat= np.cov(X.T)\n",
    "eig_vals, eig_vecs = np.linalg.eig(cov_mat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14859,)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eig_vals.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(14859, 14859)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eig_vecs.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(21.779132429600566+0j),\n",
       " (15.23378444502121+0j),\n",
       " (10.930851789580757+0j),\n",
       " (9.690065648808035+0j),\n",
       " (8.190966799578705+0j),\n",
       " (5.761317686462058+0j),\n",
       " (4.381647145256029+0j),\n",
       " (4.173206437347235+0j),\n",
       " (3.826808281415417+0j),\n",
       " (3.428680507056521+0j),\n",
       " (3.0524210924015756+0j),\n",
       " (2.6247075354640477+0j),\n",
       " (2.104967618545094+0j),\n",
       " (1.6967483466787159+0j),\n",
       " (1.6011670899345312+0j),\n",
       " (0.040378155756904004+0j),\n",
       " (0.039448444342027635+0j),\n",
       " (0.018367745696577623+0j),\n",
       " (0.01736506213226367+0j),\n",
       " (0.015850100739687956+0j),\n",
       " (0.014694563134223416+0j),\n",
       " (0.013507156299660266+0j),\n",
       " (0.013157879913510036+0j),\n",
       " (0.012984813150927844+0j),\n",
       " (0.011658991186187791+0j),\n",
       " (0.010598790049112931+0j),\n",
       " (0.010272476160700792+0j),\n",
       " (0.009738948769636987+0j),\n",
       " (0.009424796778210454+0j),\n",
       " (0.007543662824265896+0j),\n",
       " (0.00745705027041584+0j),\n",
       " (0.0068432158768995804+0j),\n",
       " (0.006487641381138787+0j),\n",
       " (0.006263815048689331+0j),\n",
       " (0.006103830615728684+0j),\n",
       " (0.006062239598000799+0j),\n",
       " (0.00569805395466136+0j),\n",
       " (0.005647337006397466+0j),\n",
       " (0.005516580643639349+0j),\n",
       " (0.005383197173924382+0j),\n",
       " (0.004990414639351441+0j),\n",
       " (0.004634319265241042+0j),\n",
       " (0.004192332235181924+0j),\n",
       " (0.004105927012679928+0j),\n",
       " (0.004004682062263217+0j),\n",
       " (0.0038480183478361773+0j),\n",
       " (0.0038409869857981994+0j),\n",
       " (0.0037978918339907543+0j),\n",
       " (0.0036432586135385844+0j),\n",
       " (0.003538060312899675+0j),\n",
       " (0.003490782285346203+0j),\n",
       " (0.003459833793590933+0j),\n",
       " (0.0034326326618711036+0j),\n",
       " (0.0033676474849314154+0j),\n",
       " (0.0032752726779418442+0j),\n",
       " (0.0032371793904064855+0j),\n",
       " (0.0031796102801569704+0j),\n",
       " (0.003119744584353369+0j),\n",
       " (0.0030930249861351555+0j),\n",
       " (0.0030500542009735325+0j),\n",
       " (0.002972559018591984+0j),\n",
       " (0.0028486601972909907+0j),\n",
       " (0.0028189687470193854+0j),\n",
       " (0.002793785076112395+0j),\n",
       " (0.0027564951050364+0j),\n",
       " (0.002752082089538471+0j),\n",
       " (0.0026905766696994496+0j),\n",
       " (0.0026217686084001536+0j),\n",
       " (0.0026100412073993516+0j),\n",
       " (0.002600116871384495+0j),\n",
       " (0.002588274917064253+0j),\n",
       " (0.0025710671880059964+0j),\n",
       " (0.0025553571280820795+0j),\n",
       " (0.0025271723772603083+0j),\n",
       " (0.0025152549223577863+0j),\n",
       " (0.002485338056476822+0j),\n",
       " (0.002387416308474329+0j),\n",
       " (0.00234725281967277+0j),\n",
       " (0.0023314746412853855+0j),\n",
       " (0.002298796873834066+0j),\n",
       " (0.002290820612326636+0j),\n",
       " (0.0022796306437112535+0j),\n",
       " (0.0022697992081799887+0j),\n",
       " (0.002265503742043685+0j),\n",
       " (0.0022589776743737145+0j),\n",
       " (0.0022084366225700877+0j),\n",
       " (0.002180757588165848+0j),\n",
       " (0.002173612395410361+0j),\n",
       " (0.002158827805245718+0j),\n",
       " (0.0020972858473519646+0j),\n",
       " (0.002084290346915528+0j),\n",
       " (0.002075726794713398+0j),\n",
       " (0.002070926468048202+0j),\n",
       " (0.002069371979195166+0j),\n",
       " (0.002033994086856743+0j),\n",
       " (0.002024900457947314+0j),\n",
       " (0.0020201230412950914+0j),\n",
       " (0.0019973867114036425+0j),\n",
       " (0.001994090510732158+0j),\n",
       " (0.0019920780226844385+0j),\n",
       " (0.001984639555885319+0j),\n",
       " (0.0019615115611237395+0j),\n",
       " (0.0019513133925866868+0j),\n",
       " (0.0019446918110701346+0j),\n",
       " (0.0019399375609140835+0j),\n",
       " (0.0019351034961393408+0j),\n",
       " (0.0019150744648725035+0j),\n",
       " (0.0019079895444159127+0j),\n",
       " (0.0019068445078411027+0j),\n",
       " (0.0018591166882044952+0j),\n",
       " (0.001852922401025412+0j),\n",
       " (0.0018177022596076043+0j),\n",
       " (0.001786386675009198+0j),\n",
       " (0.0017839856387755204+0j),\n",
       " (0.001755148089755823+0j),\n",
       " (0.0017460653074857894+0j),\n",
       " (0.0017330603287487563+0j),\n",
       " (0.0017282901064936086+0j),\n",
       " (0.0017216444700130413+0j),\n",
       " (0.0017202342510410037+0j),\n",
       " (0.0017136950159721627+0j),\n",
       " (0.0017052894520765275+0j),\n",
       " (0.0017028074743178534+0j),\n",
       " (0.0016877794947790705+0j),\n",
       " (0.001675527598996783+0j),\n",
       " (0.0016735562730917039+0j),\n",
       " (0.0016697213410062628+0j),\n",
       " (0.0016659172627295537+0j),\n",
       " (0.0016649152897211146+0j),\n",
       " (0.0016487602094859308+0j),\n",
       " (0.0016447953887738372+0j),\n",
       " (0.0016360979862136307+0j),\n",
       " (0.0016348545670101715+0j),\n",
       " (0.0016230308413641436+0j),\n",
       " (0.0016046387059612793+0j),\n",
       " (0.001595727603299223+0j),\n",
       " (0.0015899831616236785+0j),\n",
       " (0.0015606765610595306+0j),\n",
       " (0.0015531360710742197+0j),\n",
       " (0.0015408199011328937+0j),\n",
       " (0.0015324154215365115+0j),\n",
       " (0.0015238169158317061+0j),\n",
       " (0.0015099842659909064+0j),\n",
       " (0.001479703376029667+0j),\n",
       " (0.001474807188380023+0j),\n",
       " (0.0014716677259228898+0j),\n",
       " (0.0014652603749297597+0j),\n",
       " (0.001463008523319069+0j),\n",
       " (0.0014553663129262748+0j),\n",
       " (0.001454022055748128+0j),\n",
       " (0.001451520840772301+0j),\n",
       " (0.0014423063722356377+0j),\n",
       " (0.0014348354292624624+0j),\n",
       " (0.0014316922848108121+0j),\n",
       " (0.0014304936634178152+0j),\n",
       " (0.0014226851297443677+0j),\n",
       " (0.0014044301313483457+0j),\n",
       " (0.001402016732574588+0j),\n",
       " (0.0013999788488384532+0j),\n",
       " (0.0013989136495219012+0j),\n",
       " (0.0013981456539992505+0j),\n",
       " (0.0013950675390617124+0j),\n",
       " (0.001393339679227554+0j),\n",
       " (0.001386362618286949+0j),\n",
       " (0.0013672297576897044+0j),\n",
       " (0.0013584200960521803+0j),\n",
       " (0.00135681995220072+0j),\n",
       " (0.001337678079680709+0j),\n",
       " (0.0013293477546919424+0j),\n",
       " (0.0013190262075547874+0j),\n",
       " (0.001307292581531827+0j),\n",
       " (0.0012986440782371067+0j),\n",
       " (0.0012973774794334183+0j),\n",
       " (0.0012940275028065437+0j),\n",
       " (0.0012920071739071692+0j),\n",
       " (0.0012895964052650768+0j),\n",
       " (0.001286804350922138+0j),\n",
       " (0.001274808781071887+0j),\n",
       " (0.0012709056137109242+0j),\n",
       " (0.0012641106631328227+0j),\n",
       " (0.001259480776421901+0j),\n",
       " (0.0012462360501481107+0j),\n",
       " (0.0012405521126153856+0j),\n",
       " (0.001231626855365886+0j),\n",
       " (0.0012199780659530514+0j),\n",
       " (0.0012064145321814984+0j),\n",
       " (0.0012041427672036084+0j),\n",
       " (0.001203105880232522+0j),\n",
       " (0.001202514105611314+0j),\n",
       " (0.0011850282310741447+0j),\n",
       " (0.0011799404932385626+0j),\n",
       " (0.0011784804298098706+0j),\n",
       " (0.00117813971844635+0j),\n",
       " (0.0011761307073352215+0j),\n",
       " (0.0011751167666215136+0j),\n",
       " (0.0011731151994138062+0j),\n",
       " (0.001170113634544711+0j),\n",
       " (0.0011664062473337366+0j),\n",
       " (0.0011644185407438104+0j),\n",
       " (0.0011601926579704882+0j),\n",
       " (0.0011555615402647603+0j),\n",
       " (0.0011375258123918384+0j),\n",
       " (0.001135451054564124+0j),\n",
       " (0.0011256114740480498+0j),\n",
       " (0.0011224777237848896+0j),\n",
       " (0.0011133226110203627+0j),\n",
       " (0.0011127431160137056+0j),\n",
       " (0.001111590078506221+0j),\n",
       " (0.0011111146097868019+0j),\n",
       " (0.0011105727048486107+0j),\n",
       " (0.0011097401929115313+0j),\n",
       " (0.0011081172021577142+0j),\n",
       " (0.0011076497534901492+0j),\n",
       " (0.0010980111873389003+0j),\n",
       " (0.0010854208888180132+0j),\n",
       " (0.0010815869758487007+0j),\n",
       " (0.0010778873107218044+0j),\n",
       " (0.0010755850646185815+0j),\n",
       " (0.0010707756611501938+0j),\n",
       " (0.0010693844429472245+0j),\n",
       " (0.00105964178868973+0j),\n",
       " (0.0010580810103078275+0j),\n",
       " (0.001049781962810516+0j),\n",
       " (0.0010473276010788393+0j),\n",
       " (0.0010451653500199037+0j),\n",
       " (0.001043063682275959+0j),\n",
       " (0.0010422266736363967+0j),\n",
       " (0.0010385257205573046+0j),\n",
       " (0.0010380111892489864+0j),\n",
       " (0.0010373373268836251+0j),\n",
       " (0.0010322188773207066+0j),\n",
       " (0.0010313503727761083+0j),\n",
       " (0.0010300588731328255+0j),\n",
       " (0.0010260585679469443+0j),\n",
       " (0.0010233980759961007+0j),\n",
       " (0.001021833318934081+0j),\n",
       " (0.0010161952682255354+0j),\n",
       " (0.0010140951364559412+0j),\n",
       " (0.0010055246018136348+0j),\n",
       " (0.0010037957688905494+0j),\n",
       " (0.0009948054701069359+0j),\n",
       " (0.000987276041725983+0j),\n",
       " (0.0009847971307768242+0j),\n",
       " (0.0009814423140067545+0j),\n",
       " (0.0009800476142271075+0j),\n",
       " (0.0009722821214732849+0j),\n",
       " (0.0009626690903013265+0j),\n",
       " (0.0009592224093470133+0j),\n",
       " (0.0009566526905152712+0j),\n",
       " (0.0009564560950343062+0j),\n",
       " (0.0009528750837524708+0j),\n",
       " (0.0009512049832953427+0j),\n",
       " (0.0009497359700519912+0j),\n",
       " (0.0009486818647337298+0j),\n",
       " (0.0009482551113238606+0j),\n",
       " (0.0009471936313359094+0j),\n",
       " (0.0009442490324503991+0j),\n",
       " (0.0009422409856013898+0j),\n",
       " (0.0009392803375786631+0j),\n",
       " (0.0009362784419713822+0j),\n",
       " (0.0009317232971012664+0j),\n",
       " (0.0009291044406642663+0j),\n",
       " (0.0009288622212530446+0j),\n",
       " (0.0009234423920251152+0j),\n",
       " (0.0009200908996747216+0j),\n",
       " (0.0009147764682192032+0j),\n",
       " (0.0009095992222605392+0j),\n",
       " (0.0009087444139872545+0j),\n",
       " (0.0009063615030465281+0j),\n",
       " (0.0009051360951537372+0j),\n",
       " (0.0008911436090035944+0j),\n",
       " (0.0008888159087087056+0j),\n",
       " (0.0008872196816584857+0j),\n",
       " (0.0008841753779866272+0j),\n",
       " (0.0008828514632255868+0j),\n",
       " (0.0008805273265027643+0j),\n",
       " (0.0008797663998166244+0j),\n",
       " (0.0008775828969198005+0j),\n",
       " (0.0008774301452236158+0j),\n",
       " (0.0008727829356202124+0j),\n",
       " (0.0008700190934324016+0j),\n",
       " (0.0008694393022596449+0j),\n",
       " (0.0008683009931617125+0j),\n",
       " (0.0008679626587268258+0j),\n",
       " (0.0008669424822835492+0j),\n",
       " (0.00086647281232757+0j),\n",
       " (0.0008656201674209704+0j),\n",
       " (0.0008629991803780735+0j),\n",
       " (0.0008608676481300355+0j),\n",
       " (0.0008585266663248798+0j),\n",
       " (0.0008569244549014826+0j),\n",
       " (0.0008509791762374508+0j),\n",
       " (0.0008485057201768054+0j),\n",
       " (0.000847498171880024+0j),\n",
       " (0.0008459071906997062+0j),\n",
       " (0.0008422869043124529+0j),\n",
       " (0.0008416739534627898+0j),\n",
       " (0.0008411906438664571+0j),\n",
       " (0.0008406771640578682+0j),\n",
       " (0.0008319427392851059+0j),\n",
       " (0.0008286116018798984+0j),\n",
       " (0.0008212252736146773+0j),\n",
       " (0.0008198502781341541+0j),\n",
       " (0.0008182212116661783+0j),\n",
       " (0.0008172018744218025+0j),\n",
       " (0.0008132674546800745+0j),\n",
       " (0.0008109494257186846+0j),\n",
       " (0.0008102788032456654+0j),\n",
       " (0.0008091749751854517+0j),\n",
       " (0.0008077926919721158+0j),\n",
       " (0.0008009509554439705+0j),\n",
       " (0.0008001997811172741+0j),\n",
       " (0.0007983852008122147+0j),\n",
       " (0.0007944543078900827+0j),\n",
       " (0.0007919238015725234+0j),\n",
       " (0.0007900023237182833+0j),\n",
       " (0.0007891647362915931+0j),\n",
       " (0.0007878827686023587+0j),\n",
       " (0.0007873818568345604+0j),\n",
       " (0.0007865855028828469+0j),\n",
       " (0.0007814137731578139+0j),\n",
       " (0.000780646915634136+0j),\n",
       " (0.0007797372250463625+0j),\n",
       " (0.0007792252060142862+0j),\n",
       " (0.0007784479267730854+0j),\n",
       " (0.0007782894297931109+0j),\n",
       " (0.0007781950862003421+0j),\n",
       " (0.0007778263400115354+0j),\n",
       " (0.0007775735040967107+0j),\n",
       " (0.000777309750637187+0j),\n",
       " (0.0007757424641768565+0j),\n",
       " (0.0007751337793079421+0j),\n",
       " (0.0007729193942359041+0j),\n",
       " (0.000772746815731923+0j),\n",
       " (0.000772413931096305+0j),\n",
       " (0.000771646171284943+0j),\n",
       " (0.0007699927016323705+0j),\n",
       " (0.0007689280826982159+0j),\n",
       " (0.0007664111007595089+0j),\n",
       " (0.0007637291690994061+0j),\n",
       " (0.0007617305555134606+0j),\n",
       " (0.0007581728971983773+0j),\n",
       " (0.0007557425816194677+0j),\n",
       " (0.0007547207940602583+0j),\n",
       " (0.0007541167234779582+0j),\n",
       " (0.0007484972085707783+0j),\n",
       " (0.0007473651997194468+0j),\n",
       " (0.0007428394838456912+0j),\n",
       " (0.0007426211807454124+0j),\n",
       " (0.0007424914847657496+0j),\n",
       " (0.0007420487678440839+0j),\n",
       " (0.0007420078733107121+0j),\n",
       " (0.0007409925347181096+0j),\n",
       " (0.0007404527136002274+0j),\n",
       " (0.0007398103751106753+0j),\n",
       " (0.000739545432616464+0j),\n",
       " (0.0007390943172758997+0j),\n",
       " (0.0007386059136360596+0j),\n",
       " (0.0007379304169756822+0j),\n",
       " (0.0007375625167242499+0j),\n",
       " (0.0007371987733609822+0j),\n",
       " (0.0007362451564374398+0j),\n",
       " (0.0007350956232145956+0j),\n",
       " (0.0007345043625534966+0j),\n",
       " (0.0007340028297206242+0j),\n",
       " (0.0007335061855253714+0j),\n",
       " (0.0007314379349382761+0j),\n",
       " (0.0007288268421259366+0j),\n",
       " (0.0007281734883504916+0j),\n",
       " (0.0007274151492113999+0j),\n",
       " (0.0007249773931048068+0j),\n",
       " (0.000724269945100261+0j),\n",
       " (0.0007239774967140145+0j),\n",
       " (0.0007204131082987117+0j),\n",
       " (0.0007201988170426309+0j),\n",
       " (0.0007173078178224282+0j),\n",
       " (0.0007162954562480295+0j),\n",
       " (0.0007140659320501528+0j),\n",
       " (0.000713189376181285+0j),\n",
       " (0.0007108984240259786+0j),\n",
       " (0.0007080981383180286+0j),\n",
       " (0.0007074307411609287+0j),\n",
       " (0.0007067383608315486+0j),\n",
       " (0.0007051259883153742+0j),\n",
       " (0.0007039238019281739+0j),\n",
       " (0.0007016961069927202+0j),\n",
       " (0.0007003158875065183+0j),\n",
       " (0.0006995482687851566+0j),\n",
       " (0.0006993028181126243+0j),\n",
       " (0.0006991658011935628+0j),\n",
       " (0.0006988348145374965+0j),\n",
       " (0.0006985104148279342+0j),\n",
       " (0.000698356646776427+0j),\n",
       " (0.0006983230850519638+0j),\n",
       " (0.0006978588784145242+0j),\n",
       " (0.000697527663689789+0j),\n",
       " (0.0006967148079193924+0j),\n",
       " (0.0006966640669123241+0j),\n",
       " (0.0006958804064895918+0j),\n",
       " (0.0006951627194027954+0j),\n",
       " (0.000694777514968505+0j),\n",
       " (0.0006944440191021027+0j),\n",
       " (0.0006940483938390189+0j),\n",
       " (0.0006938086928504055+0j),\n",
       " (0.0006928989781661562+0j),\n",
       " (0.0006927242381276828+0j),\n",
       " (0.0006921109397468127+0j),\n",
       " (0.0006917384233249813+0j),\n",
       " (0.0006910211236139943+0j),\n",
       " (0.0006905258865076491+0j),\n",
       " (0.0006901715253294801+0j),\n",
       " (0.0006893508300742715+0j),\n",
       " (0.000688549583958792+0j),\n",
       " (0.0006881730020366538+0j),\n",
       " (0.000688153871958139+0j),\n",
       " (0.0006874617953287342+0j),\n",
       " (0.0006868381810211032+0j),\n",
       " (0.0006863630875795616+0j),\n",
       " (0.0006847401037314353+0j),\n",
       " (0.000681677919367223+0j),\n",
       " (0.0006797099308595632+0j),\n",
       " (0.0006775831660862979+0j),\n",
       " (0.000677042860160948+0j),\n",
       " (0.0006759999865512703+0j),\n",
       " (0.000675868718329136+0j),\n",
       " (0.0006720548853336606+0j),\n",
       " (0.0006690446100667492+0j),\n",
       " (0.0006675014629971725+0j),\n",
       " (0.0006643202025081954+0j),\n",
       " (0.000658819642749812+0j),\n",
       " (0.0006538780381492718+0j),\n",
       " (0.0006520902865030577+0j),\n",
       " (0.0006510663361088141+0j),\n",
       " (0.0006507706661155895+0j),\n",
       " (0.0006501885057886468+0j),\n",
       " (0.0006494392697574722+0j),\n",
       " (0.0006489076207252889+0j),\n",
       " (0.0006484174894212054+0j),\n",
       " (0.0006481796388928371+0j),\n",
       " (0.0006475868031876138+0j),\n",
       " (0.0006472076049703179+0j),\n",
       " (0.0006471009476478666+0j),\n",
       " (0.000640980703318545+0j),\n",
       " (0.000639113505825164+0j),\n",
       " (0.0006369639466870507+0j),\n",
       " (0.0006362138232832739+0j),\n",
       " (0.0006359331641376907+0j),\n",
       " (0.000635365426216839+0j),\n",
       " (0.0006349771201953902+0j),\n",
       " (0.0006333242047803535+0j),\n",
       " (0.0006309093485123902+0j),\n",
       " (0.0006304751568124738+0j),\n",
       " (0.0006288611618864907+0j),\n",
       " (0.0006278561458107566+0j),\n",
       " (0.0006253238853124446+0j),\n",
       " (0.0006241080339239205+0j),\n",
       " (0.0006199852129761722+0j),\n",
       " (0.0006183121196771397+0j),\n",
       " (0.0006178133458394378+0j),\n",
       " (0.0006166006544199819+0j),\n",
       " (0.0006145178918652472+0j),\n",
       " (0.0006137373260546949+0j),\n",
       " (0.0006134407777698818+0j),\n",
       " (0.0006107386292088572+0j),\n",
       " (0.0006079506796402857+0j),\n",
       " (0.0006076045504955893+0j),\n",
       " (0.0006073613979714043+0j),\n",
       " (0.0006065990916365954+0j),\n",
       " (0.0006036248278919609+0j),\n",
       " (0.000602107334904861+0j),\n",
       " (0.0005966760700888108+0j),\n",
       " (0.0005956952675661683+0j),\n",
       " (0.00059458220854947+0j),\n",
       " (0.00059210195475975+0j),\n",
       " (0.000590591239798801+0j),\n",
       " (0.0005903296194022529+0j),\n",
       " (0.000590203183641941+0j),\n",
       " (0.0005899024109805376+0j),\n",
       " (0.0005896525246784424+0j),\n",
       " (0.0005895956503192566+0j),\n",
       " (0.0005894951520033315+0j),\n",
       " (0.0005891717797723312+0j),\n",
       " (0.0005891004674483148+0j),\n",
       " (0.0005887618467629792+0j),\n",
       " (0.0005883540472744878+0j),\n",
       " (0.0005877949927592977+0j),\n",
       " (0.000587477874587333+0j),\n",
       " (0.0005871941830184333+0j),\n",
       " (0.0005869888861574537+0j),\n",
       " (0.0005869542266319248+0j),\n",
       " (0.0005866400157160557+0j),\n",
       " (0.0005854569727899609+0j),\n",
       " (0.0005848174901379192+0j),\n",
       " (0.0005816469068162652+0j),\n",
       " (0.0005812411759726778+0j),\n",
       " (0.0005811021502184862+0j),\n",
       " (0.000580375593851473+0j),\n",
       " (0.0005795585078777903+0j),\n",
       " (0.0005788831716877174+0j),\n",
       " (0.0005776931561479658+0j),\n",
       " (0.000577262533670736+0j),\n",
       " (0.0005754202620691754+0j),\n",
       " (0.0005735330260143296+0j),\n",
       " (0.0005726068147936986+0j),\n",
       " (0.0005714616631872079+0j),\n",
       " (0.000571131074173027+0j),\n",
       " (0.0005697728644099472+0j),\n",
       " (0.0005685849260399643+0j),\n",
       " (0.0005676086487309717+0j),\n",
       " (0.0005671993144667431+0j),\n",
       " (0.0005669846095087656+0j),\n",
       " (0.0005655397823517966+0j),\n",
       " (0.0005629183929965419+0j),\n",
       " (0.0005625673834725734+0j),\n",
       " (0.0005613812532241063+0j),\n",
       " (0.0005585035758819134+0j),\n",
       " (0.0005575219956754444+0j),\n",
       " (0.0005571522538519571+0j),\n",
       " (0.0005559937677200523+0j),\n",
       " (0.0005542002435960156+0j),\n",
       " (0.0005532635734933965+0j),\n",
       " (0.0005507374849706831+0j),\n",
       " (0.0005493935236953607+0j),\n",
       " (0.0005487696336411849+0j),\n",
       " (0.0005475766569373777+0j),\n",
       " (0.0005439927045804497+0j),\n",
       " (0.0005430600049623589+0j),\n",
       " (0.0005421797144324115+0j),\n",
       " (0.0005368552221140397+0j),\n",
       " (0.000535136511704456+0j),\n",
       " (0.0005345320658920276+0j),\n",
       " (0.0005324223095689993+0j),\n",
       " (0.0005319689942021895+0j),\n",
       " (0.0005315596533982862+0j),\n",
       " (0.000529302342032241+0j),\n",
       " (0.0005278515273152923+0j),\n",
       " (0.0005275245411253373+0j),\n",
       " (0.0005273674408245691+0j),\n",
       " (0.000524000858131758+0j),\n",
       " (0.000523094737302828+0j),\n",
       " (0.0005215929605514296+0j),\n",
       " (0.0005209939540531074+0j),\n",
       " (0.0005206291937993913+0j),\n",
       " (0.0005197797131922245+0j),\n",
       " (0.0005196753119326904+0j),\n",
       " (0.0005186989398935337+0j),\n",
       " (0.0005182911466300856+0j),\n",
       " (0.0005176593041790031+0j),\n",
       " (0.0005175400830883882+0j),\n",
       " (0.0005171756625843968+0j),\n",
       " (0.000515454371900103+0j),\n",
       " (0.0005152757570370483+0j),\n",
       " (0.0005124222727367669+0j),\n",
       " (0.0005120573834995808+0j),\n",
       " (0.0005112203056931126+0j),\n",
       " (0.0005108025922137441+0j),\n",
       " (0.0005106698711266224+0j),\n",
       " (0.0005099824313865009+0j),\n",
       " (0.0005077791225224816+0j),\n",
       " (0.0005071827393694426+0j),\n",
       " (0.0005062629711619039+0j),\n",
       " (0.0005060225523272927+0j),\n",
       " (0.000505838274044517+0j),\n",
       " (0.0005056675561420515+0j),\n",
       " (0.0005053222395376776+0j),\n",
       " (0.0005052635634341232+0j),\n",
       " (0.0005052016636279784+0j),\n",
       " (0.0005051396267608385+0j),\n",
       " (0.0005048473436676206+0j),\n",
       " (0.0005047182507117662+0j),\n",
       " (0.000504613148807601+0j),\n",
       " (0.0005043327092411182+0j),\n",
       " (0.0005037578868082184+0j),\n",
       " (0.0005032461583476005+0j),\n",
       " (0.0005031372974851939+0j),\n",
       " (0.0005027925014442997+0j),\n",
       " (0.0005020807885052005+0j),\n",
       " (0.0004968859309968777+0j),\n",
       " (0.0004950203126217125+0j),\n",
       " (0.0004936345385482334+0j),\n",
       " (0.0004901130905558072+0j),\n",
       " (0.0004896588560945902+0j),\n",
       " (0.0004886499124046811+0j),\n",
       " (0.0004858500072897817+0j),\n",
       " (0.00048507649231340784+0j),\n",
       " (0.0004841769763414937+0j),\n",
       " (0.0004810488644457424+0j),\n",
       " (0.00047962690362130265+0j),\n",
       " (0.000479035275115462+0j),\n",
       " (0.00047891863010777693+0j),\n",
       " (0.000478894178850584+0j),\n",
       " (0.000478617013379426+0j),\n",
       " (0.0004784982569427877+0j),\n",
       " (0.0004784748368074506+0j),\n",
       " (0.00047833971753004616+0j),\n",
       " (0.0004779517839118485+0j),\n",
       " (0.00047758516416304166+0j),\n",
       " (0.00047755797940120216+0j),\n",
       " (0.0004771804520374088+0j),\n",
       " (0.00047708687681740736+0j),\n",
       " (0.00047704158581719585+0j),\n",
       " (0.00047697274501680187+0j),\n",
       " (0.0004767875808474701+0j),\n",
       " (0.000476743430903568+0j),\n",
       " (0.0004766598762262647+0j),\n",
       " (0.0004761898908215284+0j),\n",
       " (0.00047606798855772536+0j),\n",
       " (0.00047583748708586253+0j),\n",
       " (0.0004750653089939407+0j),\n",
       " (0.0004749268660486798+0j),\n",
       " (0.00047425303689988755+0j),\n",
       " (0.00047407124338436334+0j),\n",
       " (0.0004734927016251334+0j),\n",
       " (0.0004727730468231283+0j),\n",
       " (0.00047241913739476746+0j),\n",
       " (0.0004719869975265954+0j),\n",
       " (0.00047131526126125393+0j),\n",
       " (0.00047081603809814594+0j),\n",
       " (0.0004690769450754277+0j),\n",
       " (0.0004667313237312974+0j),\n",
       " (0.0004658327885136526+0j),\n",
       " (0.00046512193918565604+0j),\n",
       " (0.0004642512407182829+0j),\n",
       " (0.0004641000238720619+0j),\n",
       " (0.00046286385896721316+0j),\n",
       " (0.00046238428600957126+0j),\n",
       " (0.00046125475576942493+0j),\n",
       " (0.0004605921298550239+0j),\n",
       " (0.0004599327424172305+0j),\n",
       " (0.00045972561629968747+0j),\n",
       " (0.00045957441599009597+0j),\n",
       " (0.00045945695866296917+0j),\n",
       " (0.00045925231589717956+0j),\n",
       " (0.0004591237070415176+0j),\n",
       " (0.00045890928838845983+0j),\n",
       " (0.0004588430891328968+0j),\n",
       " (0.0004587269615488785+0j),\n",
       " (0.0004584877236654925+0j),\n",
       " (0.00045823321960834865+0j),\n",
       " (0.00045816056883906607+0j),\n",
       " (0.0004578566840402808+0j),\n",
       " (0.000457689269181713+0j),\n",
       " (0.00045746648059105796+0j),\n",
       " (0.0004574294312874968+0j),\n",
       " (0.00045726320136495397+0j),\n",
       " (0.0004571928096065581+0j),\n",
       " (0.00045711237273172913+0j),\n",
       " (0.00045702293900216944+0j),\n",
       " (0.0004567013275707155+0j),\n",
       " (0.00045660393335407207+0j),\n",
       " (0.0004564674680373144+0j),\n",
       " (0.000456255538977761+0j),\n",
       " (0.00045601160662973357+0j),\n",
       " (0.0004556646895983605+0j),\n",
       " (0.00045525772652127113+0j),\n",
       " (0.00045470505705956727+0j),\n",
       " (0.0004545058193095974+0j),\n",
       " (0.00045347133697271785+0j),\n",
       " (0.00045329985953272706+0j),\n",
       " (0.0004521016801985791+0j),\n",
       " (0.0004513752827592596+0j),\n",
       " (0.000450630680328546+0j),\n",
       " (0.0004501951132251714+0j),\n",
       " (0.00044902791905390384+0j),\n",
       " (0.00044840810609476353+0j),\n",
       " (0.00044816583686543875+0j),\n",
       " (0.0004469026131759293+0j),\n",
       " (0.0004460930346431846+0j),\n",
       " (0.00044588399671591965+0j),\n",
       " (0.000445611430259852+0j),\n",
       " (0.00044384292566153636+0j),\n",
       " (0.00044167460986458166+0j),\n",
       " (0.0004410340874402034+0j),\n",
       " (0.0004401553719861011+0j),\n",
       " (0.00043851356737811353+0j),\n",
       " (0.00043786438280283055+0j),\n",
       " (0.00043455756473748884+0j),\n",
       " (0.0004344551474449157+0j),\n",
       " (0.0004342967526101264+0j),\n",
       " (0.000434170246778641+0j),\n",
       " (0.0004341254760667057+0j),\n",
       " (0.00043409475600147046+0j),\n",
       " (0.0004339358562931648+0j),\n",
       " (0.0004336822688628522+0j),\n",
       " (0.000433474845564479+0j),\n",
       " (0.0004331206169330181+0j),\n",
       " (0.0004326788659884022+0j),\n",
       " (0.00043223722355696334+0j),\n",
       " (0.00043028069430692603+0j),\n",
       " (0.00043026779094391813+0j),\n",
       " (0.0004299905408375388+0j),\n",
       " (0.00042800210083615624+0j),\n",
       " (0.0004260428141464047+0j),\n",
       " (0.00042579964464117576+0j),\n",
       " (0.0004249764983554982+0j),\n",
       " (0.0004248611291079438+0j),\n",
       " (0.00042483804998098444+0j),\n",
       " (0.0004246181499765058+0j),\n",
       " (0.0004237979167682874+0j),\n",
       " (0.00042342349116460197+0j),\n",
       " (0.00042266512483419167+0j),\n",
       " (0.0004225491881404977+0j),\n",
       " (0.00041943882176904876+0j),\n",
       " (0.0004190366735170018+0j),\n",
       " (0.0004175045774095505+0j),\n",
       " (0.0004164165527656779+0j),\n",
       " (0.0004141092925871911+0j),\n",
       " (0.0004139183983616316+0j),\n",
       " (0.00041354210858698416+0j),\n",
       " (0.00041315070200801903+0j),\n",
       " (0.0004127194869427301+0j),\n",
       " (0.0004114319003564087+0j),\n",
       " (0.0004110186993321569+0j),\n",
       " (0.0004099317474872008+0j),\n",
       " (0.0004084659599903036+0j),\n",
       " (0.0004080579876949346+0j),\n",
       " (0.0004079251501482471+0j),\n",
       " (0.0004078679120438201+0j),\n",
       " (0.0004075505569522092+0j),\n",
       " (0.0004060712320616412+0j),\n",
       " (0.0004054702395449308+0j),\n",
       " (0.00040512717393271824+0j),\n",
       " (0.00040286656806568075+0j),\n",
       " (0.0004009793433247729+0j),\n",
       " (0.0004007724503493369+0j),\n",
       " (0.0004003755491738685+0j),\n",
       " (0.0003993247085534792+0j),\n",
       " (0.0003986256960371213+0j),\n",
       " (0.0003984252407219338+0j),\n",
       " (0.00039821084501254016+0j),\n",
       " (0.0003977474204451984+0j),\n",
       " (0.00039696021687427957+0j),\n",
       " (0.00039419364133842354+0j),\n",
       " (0.00039416032748100467+0j),\n",
       " (0.0003933136478465559+0j),\n",
       " (0.0003912828195159257+0j),\n",
       " (0.0003898006647561954+0j),\n",
       " (0.000389121320622999+0j),\n",
       " (0.000388959358620608+0j),\n",
       " (0.00038875391407531764+0j),\n",
       " (0.0003886768749086758+0j),\n",
       " (0.00038858091976773444+0j),\n",
       " (0.0003884881712912588+0j),\n",
       " (0.00038832870841225503+0j),\n",
       " (0.0003882156267803746+0j),\n",
       " (0.00038800012742305415+0j),\n",
       " (0.0003879414213535891+0j),\n",
       " (0.0003877022225056032+0j),\n",
       " (0.0003875869394476442+0j),\n",
       " (0.0003865468812342234+0j),\n",
       " (0.000386449117586453+0j),\n",
       " (0.0003863642297937459+0j),\n",
       " (0.00038630254018676423+0j),\n",
       " (0.00038618451971948747+0j),\n",
       " (0.00038613522673497424+0j),\n",
       " (0.0003860681368424319+0j),\n",
       " (0.00038599361580891217+0j),\n",
       " (0.00038581931211997773+0j),\n",
       " (0.0003857024152608887+0j),\n",
       " (0.00038562030008463285+0j),\n",
       " (0.0003855731475583025+0j),\n",
       " (0.00038546710046149385+0j),\n",
       " (0.0003854096361986858+0j),\n",
       " (0.00038534063816937273+0j),\n",
       " (0.00038524600221488117+0j),\n",
       " (0.0003852057221911016+0j),\n",
       " (0.00038518992892455276+0j),\n",
       " (0.0003850845663148563+0j),\n",
       " (0.0003848911090223355+0j),\n",
       " (0.0003847373183110875+0j),\n",
       " (0.00038454876435866633+0j),\n",
       " (0.0003843396414502888+0j),\n",
       " (0.0003842792656751962+0j),\n",
       " (0.00038412305869709053+0j),\n",
       " (0.0003840182859302775+0j),\n",
       " (0.0003837880350302186+0j),\n",
       " (0.0003833224146994569+0j),\n",
       " (0.00038299583695830064+0j),\n",
       " (0.0003829237411460784+0j),\n",
       " (0.00038135351209414196+0j),\n",
       " (0.00038124458948758823+0j),\n",
       " (0.000381046382887313+0j),\n",
       " (0.0003798242963346835+0j),\n",
       " (0.0003790941159568033+0j),\n",
       " (0.0003782722480520422+0j),\n",
       " (0.0003763850553395872+0j),\n",
       " (0.00037564060377008357+0j),\n",
       " (0.0003754648537151638+0j),\n",
       " (0.00037474072721995947+0j),\n",
       " (0.00037393264529356294+0j),\n",
       " (0.0003738066341942738+0j),\n",
       " (0.000373613605414961+0j),\n",
       " (0.0003730839856914714+0j),\n",
       " (0.00037207920308101564+0j),\n",
       " (0.0003719207413322422+0j),\n",
       " (0.0003718529574887832+0j),\n",
       " (0.0003717547992946663+0j),\n",
       " (0.0003717288231298394+0j),\n",
       " (0.0003717012558501992+0j),\n",
       " (0.0003716561789717613+0j),\n",
       " (0.00037162623954670274+0j),\n",
       " (0.0003716143669765379+0j),\n",
       " (0.00037159808957142434+0j),\n",
       " (0.00037157008168587746+0j),\n",
       " (0.00037155010160344246+0j),\n",
       " (0.0003715325087900019+0j),\n",
       " (0.00037146470445719153+0j),\n",
       " (0.000371455092792311+0j),\n",
       " (0.00037143788228697185+0j),\n",
       " (0.00037139210566541474+0j),\n",
       " (0.00037138073812599526+0j),\n",
       " (0.00037136709494275174+0j),\n",
       " (0.000371332564063797+0j),\n",
       " (0.0003713128238716605+0j),\n",
       " (0.00037128622081391844+0j),\n",
       " (0.0003712638752100893+0j),\n",
       " (0.00037124723662304424+0j),\n",
       " (0.0003712157371725905+0j),\n",
       " (0.0003712133377268071+0j),\n",
       " (0.0003711933158373536+0j),\n",
       " (0.00037118558592499597+0j),\n",
       " (0.0003711811663762369+0j),\n",
       " (0.0003711423063772637+0j),\n",
       " (0.0003710992039400213+0j),\n",
       " (0.0003710972810969987+0j),\n",
       " (0.00037107405704012825+0j),\n",
       " (0.00037102232303582773+0j),\n",
       " (0.0003709849746351391+0j),\n",
       " (0.00037097183710500765+0j),\n",
       " (0.00037094930460986503+0j),\n",
       " (0.00037092945063826925+0j),\n",
       " (0.00037084341120237726+0j),\n",
       " (0.00037081052395697775+0j),\n",
       " (0.0003708032044547833+0j),\n",
       " (0.0003707870596101649+0j),\n",
       " (0.00037076861644499065+0j),\n",
       " (0.0003707456316659315+0j),\n",
       " (0.00037070164885430283+0j),\n",
       " (0.00037065677789825094+0j),\n",
       " (0.0003706495613477701+0j),\n",
       " (0.0003705989012972453+0j),\n",
       " (0.00037057356955014415+0j),\n",
       " (0.0003705495827929977+0j),\n",
       " (0.0003705062018827933+0j),\n",
       " (0.0003704603976342755+0j),\n",
       " (0.0003704142996238139+0j),\n",
       " (0.0003703737963442419+0j),\n",
       " (0.00037024963731712503+0j),\n",
       " (0.00036999643789457725+0j),\n",
       " (0.0003698279307080261+0j),\n",
       " (0.00036972979675367016+0j),\n",
       " (0.0003695557792545159+0j),\n",
       " (0.00036932244007602494+0j),\n",
       " (0.00036926647425420946+0j),\n",
       " (0.00036916785827381215+0j),\n",
       " (0.00036896886470308074+0j),\n",
       " (0.0003688433016454584+0j),\n",
       " (0.00036861614969547476+0j),\n",
       " (0.0003683819815099498+0j),\n",
       " (0.00036835152315306725+0j),\n",
       " (0.00036802893341973886+0j),\n",
       " (0.0003675005824233198+0j),\n",
       " (0.00036739825146466453+0j),\n",
       " (0.0003670794097372873+0j),\n",
       " (0.00036697438307576463+0j),\n",
       " (0.00036659666609395203+0j),\n",
       " (0.0003664194710210658+0j),\n",
       " (0.0003662176008562839+0j),\n",
       " (0.00036609862108526756+0j),\n",
       " (0.00036574522049454884+0j),\n",
       " (0.00036547062967387734+0j),\n",
       " (0.0003650689536735338+0j),\n",
       " (0.00036498595904755607+0j),\n",
       " (0.00036371540116202284+0j),\n",
       " (0.00036309055597575013+0j),\n",
       " (0.0003628956938411604+0j),\n",
       " (0.00036240539473513563+0j),\n",
       " (0.0003621213306052604+0j),\n",
       " (0.0003617300014705773+0j),\n",
       " (0.00036127040822976587+0j),\n",
       " (0.00035906941069293826+0j),\n",
       " (0.00035876220191526624+0j),\n",
       " (0.00035791097642814306+0j),\n",
       " (0.00035782510934116507+0j),\n",
       " (0.00035777225144572727+0j),\n",
       " (0.00035770037502283247+0j),\n",
       " (0.0003576205101521943+0j),\n",
       " (0.00035753873449334393+0j),\n",
       " (0.00035742513863695877+0j),\n",
       " (0.0003573902743093329+0j),\n",
       " (0.00035732211602239333+0j),\n",
       " (0.0003570694976990159+0j),\n",
       " (0.00035694176533542237+0j),\n",
       " (0.0003569052035956851+0j),\n",
       " (0.00035683696771665374+0j),\n",
       " (0.00035676875660572364+0j),\n",
       " (0.00035666640459422515+0j),\n",
       " (0.00035661848035810936+0j),\n",
       " (0.00035627444721063784+0j),\n",
       " (0.0003561834601723419+0j),\n",
       " (0.0003555818259261282+0j),\n",
       " (0.00035551949621058125+0j),\n",
       " (0.00035401213675989433+0j),\n",
       " (0.0003527557212798319+0j),\n",
       " (0.0003526736449875587+0j),\n",
       " (0.00035267364498753486+0j),\n",
       " (0.0003525224489644132+0j),\n",
       " (0.0003523752985904786+0j),\n",
       " (0.00035178539311665454+0j),\n",
       " (0.00035164501577086247+0j),\n",
       " (0.0003510024040377747+0j),\n",
       " (0.0003506324429520036+0j),\n",
       " (0.00035049524391388967+0j),\n",
       " (0.00035041007796350094+0j),\n",
       " (0.00035032215854628143+0j),\n",
       " (0.00035030930933719245+0j),\n",
       " (0.00035027639124646226+0j),\n",
       " (0.000350193057113659+0j),\n",
       " (0.00035018244270174553+0j),\n",
       " (0.00035012126883214866+0j),\n",
       " (0.0003500963733525586+0j),\n",
       " (0.00035006847961204527+0j),\n",
       " (0.00035003553088731686+0j),\n",
       " (0.00034998765003660447+0j),\n",
       " (0.00034992946724557385+0j),\n",
       " (0.00034987924450464016+0j),\n",
       " (0.0003498618117136732+0j),\n",
       " (0.0003498204294803872+0j),\n",
       " (0.0003498043037885637+0j),\n",
       " (0.00034971171999548836+0j),\n",
       " (0.0003496966874810424+0j),\n",
       " (0.00034963394993532665+0j),\n",
       " (0.00034957559575163626+0j),\n",
       " (0.0003495498804653883+0j),\n",
       " (0.0003494902458337884+0j),\n",
       " (0.000349472166656944+0j),\n",
       " (0.000349364155476274+0j),\n",
       " (0.00034931406305593426+0j),\n",
       " (0.00034924613579313715+0j),\n",
       " (0.00034918885550796856+0j),\n",
       " (0.00034911647318088576+0j),\n",
       " (0.000349032239951028+0j),\n",
       " (0.00034898089334538967+0j),\n",
       " (0.0003489412262865443+0j),\n",
       " (0.0003489087733196953+0j),\n",
       " (0.00034877700623857616+0j),\n",
       " (0.00034872780954956157+0j),\n",
       " (0.0003487047310010813+0j),\n",
       " (0.0003486782896341064+0j),\n",
       " (0.00034858405741792283+0j),\n",
       " (0.0003484901134211959+0j),\n",
       " (0.00034843113807574936+0j),\n",
       " (0.0003483061808949826+0j),\n",
       " (0.00034813452589497134+0j),\n",
       " (0.00034807479127003574+0j),\n",
       " (0.00034803498568364397+0j),\n",
       " (0.0003479153979874116+0j),\n",
       " (0.00034776164074647035+0j),\n",
       " (0.00034760413931046404+0j),\n",
       " (0.00034751115753140063+0j),\n",
       " (0.0003473561841206486+0j),\n",
       " (0.0003472259616235773+0j),\n",
       " (0.0003469886048804169+0j),\n",
       " (0.000346735891283455+0j),\n",
       " (0.000346711135910123+0j),\n",
       " (0.00034648812043073797+0j),\n",
       " (0.0003462381309831544+0j),\n",
       " (0.0003461221344043711+0j),\n",
       " (0.0003460869000388905+0j),\n",
       " (0.00034594318189597515+0j),\n",
       " (0.0003459132341806745+0j),\n",
       " (0.00034584597803251794+0j),\n",
       " (0.00034581102064791606+0j),\n",
       " (0.0003457364116008534+0j),\n",
       " (0.0003454323003218004+0j),\n",
       " (0.0003453754694013796+0j),\n",
       " (0.0003452620430182674+0j),\n",
       " (0.0003451380513649038+0j),\n",
       " (0.0003450894186419741+0j),\n",
       " (0.00034496569847081024+0j),\n",
       " (0.0003448788052787746+0j),\n",
       " (0.00034476958776473335+0j),\n",
       " (0.00034471188944268177+0j),\n",
       " (0.00034464642195287574+0j),\n",
       " (0.0003445763660877955+0j),\n",
       " (0.0003444530506820285+0j),\n",
       " (0.0003442947912439222+0j),\n",
       " (0.000344265612709813+0j),\n",
       " (0.00034365676330989784+0j),\n",
       " (0.00034272500913619267+0j),\n",
       " (0.00034253956654636196+0j),\n",
       " (0.00034234424403810864+0j),\n",
       " (0.0003420722226983972+0j),\n",
       " (0.00034195419611715013+0j),\n",
       " (0.0003418437777829063+0j),\n",
       " (0.00034174322567351344+0j),\n",
       " (0.00034153952180681057+0j),\n",
       " (0.0003413483752187511+0j),\n",
       " (0.00034123905047897193+0j),\n",
       " (0.0003411950982910209+0j),\n",
       " ...]"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eig_pairs = [(np.abs(eig_vals[i]), eig_vecs[:,i]) for i in range(len(eig_vals))]\n",
    "\n",
    "#sort from high to low\n",
    "eig_pairs.sort(key=lambda  x:x[0], reverse=True)\n",
    "tot= sum(eig_vals)\n",
    "var_exp=[(i/tot)*100  for i in sorted(eig_vals, reverse=True)]\n",
    "cum_var_exp = np.cumsum(var_exp) # cumulative explained variance\n",
    "var_exp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/matplotlib/transforms.py:796: ComplexWarning: Casting complex values to real discards the imaginary part\n",
      "  points = np.array(args, dtype=float).reshape(2, 2)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/numpy/core/numeric.py:538: ComplexWarning: Casting complex values to real discards the imaginary part\n",
      "  return array(a, dtype, copy=False, order=order)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize= (10,5))\n",
    "plt.bar(range(14859), var_exp, alpha=0.3333, align='center', label='individual explained variance', color='g')\n",
    "plt.step(range(14859), cum_var_exp, where='mid', label='cumulative explained variance')\n",
    "plt.ylabel('Explained variance ration')\n",
    "plt.xlabel('Principle Components')\n",
    "plt.legend(loc='best')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((4560, 16), (4560,))"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_7d.shape, X_clustered.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/seaborn/axisgrid.py:2065: UserWarning: The `size` parameter has been renamed to `height`; pleaes update your code.\n",
      "  warnings.warn(msg, UserWarning)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kde.py:487: RuntimeWarning: invalid value encountered in true_divide\n",
      "  binned = fast_linbin(X, a, b, gridsize) / (delta * nobs)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kdetools.py:34: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  FAC1 = 2*(np.pi*bw/RANGE)**2\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.PairGrid at 0x7f2090793b38>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 586.8x532.8 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create a temp dataframe from our PCA projection data \"x_9d\"\n",
    "df = pd.DataFrame(x_7d)\n",
    "df = df[[0,1,2]] # only want to visualise relationships between first 3 projections\n",
    "df['X_cluster'] = X_clustered\n",
    "\n",
    "# Call Seaborn's pairplot to visualize our KMeans clustering on the PCA projected data\n",
    "sns.pairplot(df, hue='X_cluster', palette= 'Dark2', diag_kind='kde',size=1.85)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 3)\n",
    "\n",
    "#Compute cluster centers and predict cluster indices\n",
    "X_clustered = kmeans.fit_predict(x_4d)\n",
    "\n",
    "#Define our own color map\n",
    "LABEL_COLOR_MAP = {0:'r', 1: 'g', 2: 'b'}\n",
    "label_color = [LABEL_COLOR_MAP[l] for l in X_clustered]\n",
    "\n",
    "# Plot the scatter digram\n",
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,2], c= label_color, alpha=0.5) \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a temp dataframe from our PCA projection data \"x_9d\"\n",
    "df = pd.DataFrame(x_4d)\n",
    "df = df[[0,1,2]] # only want to visualise relationships between first 3 projections\n",
    "df['X_cluster'] = X_clustered"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/seaborn/axisgrid.py:2065: UserWarning: The `size` parameter has been renamed to `height`; pleaes update your code.\n",
      "  warnings.warn(msg, UserWarning)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kde.py:487: RuntimeWarning: invalid value encountered in true_divide\n",
      "  binned = fast_linbin(X, a, b, gridsize) / (delta * nobs)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kdetools.py:34: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  FAC1 = 2*(np.pi*bw/RANGE)**2\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.PairGrid at 0x7fa4d3fe0dd8>"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 586.8x532.8 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Call Seaborn's pairplot to visualize our KMeans clustering on the PCA projected data\n",
    "sns.pairplot(df, hue='X_cluster', palette= 'Dark2', diag_kind='kde',size=1.85)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## no-typed data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "scrolled": true
   },
   "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>(boc tower, hong kong)</th>\n",
       "      <th>(chaplin, chaplin)</th>\n",
       "      <th>(walter gropius, bauhaus)</th>\n",
       "      <th>(hawaii state supreme court, ali ʻ iōlani hale)</th>\n",
       "      <th>(thurgot, saint andrews)</th>\n",
       "      <th>(santiniketan, indian)</th>\n",
       "      <th>(goldwater, reagan)</th>\n",
       "      <th>(dostum, rabbani)</th>\n",
       "      <th>(coloman, béla)</th>\n",
       "      <th>(rockies, oakland athletics)</th>\n",
       "      <th>...</th>\n",
       "      <th>(bali strait, java)</th>\n",
       "      <th>(brown, sarah macaulay)</th>\n",
       "      <th>(statenvertaling, dutch)</th>\n",
       "      <th>(sagan, aaas 's)</th>\n",
       "      <th>(spain, coimbra)</th>\n",
       "      <th>(france, cascade mountains)</th>\n",
       "      <th>(love, brian wilson)</th>\n",
       "      <th>(lenin, reds)</th>\n",
       "      <th>(kelly, daisy bell)</th>\n",
       "      <th>(mclean, lisa kindred)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>star</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>be</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>appoint</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>established</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>carry</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 14859 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             (boc tower, hong kong)  (chaplin, chaplin)  \\\n",
       "star                              0                   0   \n",
       "be                                1                   0   \n",
       "appoint                           0                   0   \n",
       "established                       0                   0   \n",
       "carry                             0                   0   \n",
       "\n",
       "             (walter gropius, bauhaus)  \\\n",
       "star                                 0   \n",
       "be                                   0   \n",
       "appoint                              0   \n",
       "established                          0   \n",
       "carry                                0   \n",
       "\n",
       "             (hawaii state supreme court, ali ʻ iōlani hale)  \\\n",
       "star                                                       0   \n",
       "be                                                         0   \n",
       "appoint                                                    0   \n",
       "established                                                0   \n",
       "carry                                                      0   \n",
       "\n",
       "             (thurgot, saint andrews)  (santiniketan, indian)  \\\n",
       "star                                0                       0   \n",
       "be                                  0                       0   \n",
       "appoint                             0                       0   \n",
       "established                         0                       0   \n",
       "carry                               0                       0   \n",
       "\n",
       "             (goldwater, reagan)  (dostum, rabbani)  (coloman, béla)  \\\n",
       "star                           0                  0                0   \n",
       "be                             0                  0                0   \n",
       "appoint                        0                  0                0   \n",
       "established                    0                  0                0   \n",
       "carry                          0                  0                0   \n",
       "\n",
       "             (rockies, oakland athletics)  ...  (bali strait, java)  \\\n",
       "star                                    0  ...                    0   \n",
       "be                                      0  ...                    0   \n",
       "appoint                                 0  ...                    0   \n",
       "established                             0  ...                    0   \n",
       "carry                                   0  ...                    0   \n",
       "\n",
       "             (brown, sarah macaulay)  (statenvertaling, dutch)  \\\n",
       "star                               0                         0   \n",
       "be                                 0                         1   \n",
       "appoint                            0                         0   \n",
       "established                        0                         0   \n",
       "carry                              0                         0   \n",
       "\n",
       "             (sagan, aaas 's)  (spain, coimbra)  (france, cascade mountains)  \\\n",
       "star                        0                 0                            0   \n",
       "be                          1                 0                            0   \n",
       "appoint                     0                 0                            0   \n",
       "established                 0                 0                            0   \n",
       "carry                       0                 0                            0   \n",
       "\n",
       "             (love, brian wilson)  (lenin, reds)  (kelly, daisy bell)  \\\n",
       "star                            0              0                    0   \n",
       "be                              0              0                    0   \n",
       "appoint                         0              0                    0   \n",
       "established                     0              0                    0   \n",
       "carry                           0              0                    0   \n",
       "\n",
       "             (mclean, lisa kindred)  \n",
       "star                              0  \n",
       "be                                0  \n",
       "appoint                           0  \n",
       "established                       0  \n",
       "carry                             0  \n",
       "\n",
       "[5 rows x 14859 columns]"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_df_pred = pd.DataFrame(features_pred)\n",
    "features_df_pred = features_df_pred.rename(columns= id2args)\n",
    "features_df_pred = features_df_pred.rename(index= id2label_pred)\n",
    "features_df_pred.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "features_df.where(features_df>1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAa8AAAGbCAYAAABzgB+6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAYXUlEQVR4nO3df4zcd53f8ecrdpwQfsaXdeSznbNBhtapyI9bRSB66GjoxaQpDkVBGwG1SqS0p6QC0ap1ilToH5E4qt7Rqs2h3EExNIfPd4BioeuVyKW9a3UQNmAuOInrTQLJ1j57SYJCmpzBzrt/zNcwttf2rr3jmY/3+ZBW853PfGfmPV82fnp+eEhVIUlSSy4Y9gCSJM2X8ZIkNcd4SZKaY7wkSc0xXpKk5iwd9gAAl112Wa1du3bYY0iSRshDDz30o6oam+2ykYjX2rVrmZycHPYYkqQRkuSHJ7vMlw0lSc0xXpKk5hgvSVJzjJckqTmnjVeSNyXZ1ffzfJKPJFme5IEke7vTS/uuc1eSqSR7ktww2IcgSVpsThuvqtpTVVdX1dXArwIvAl8FtgA7q2o9sLM7T5INwARwJbARuCfJkgHNL0lahOb7suH1wONV9UNgE7C1W98K3NxtbwK2VdWhqnoSmAKuW4hhJUmC+cdrAvhSt315Ve0H6E5XdOurgKf7rjPdrR0jye1JJpNMzszMzHMMSdJiNud4JVkGvBv4o9PtOsvaCf+nYVV1b1WNV9X42Nis/4BakqRZzeeZ17uA71TVge78gSQrAbrTg936NLCm73qrgX1nO6gkSUfNJ1638ouXDAF2AJu77c3A/X3rE0kuSrIOWA88eLaDSpJ01Jy+2zDJJcDfBf5x3/Inge1JbgOeAm4BqKrdSbYDjwCHgTuq6siCTi1JWtTmFK+qehH4pePWnqH36cPZ9r8buPusp5ur6Wn4nd+Bb3wDXvta+NCH4P3vhwv8N9iSdD4aiW+VPys/+hG85z3wzDPw6lfDgQPw8Y/D44/DJz4x7OkkSQPQ/lOTbdvg2Wdh+XK48EK45JLes6/77oODB09/fUlSc9qP1+QkLD3uCeSSJb21J54YzkySpIFqP15vfCMcPnzsWlVv7Zd/eTgzSZIGqv14feADcPHF8Pzzv4jWc8/BO94BV1wx7OkkSQPQfryuuAK++MXeM7DnnoOXXoJbb4VPf3rYk0mSBqT9TxsCXHstfO1r8MILsGxZ70eSdN46P+J11KteNewJJEnnQPsvG0qSFh3jJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5c4pXktcl+eMkjyV5NMlbkyxP8kCSvd3ppX3735VkKsmeJDcMbnxJ0mI012de/x7406r6G8BVwKPAFmBnVa0HdnbnSbIBmACuBDYC9yRZstCDS5IWr9PGK8lrgLcDnwWoqp9W1Y+BTcDWbretwM3d9iZgW1UdqqongSnguoUeXJK0eM3lmdfrgRngPyf5bpLfT/JK4PKq2g/Qna7o9l8FPN13/eluTZKkBTGXeC0FrgV+t6quAf4f3UuEJ5FZ1uqEnZLbk0wmmZyZmZnTsJIkwdziNQ1MV9W3uvN/TC9mB5KsBOhOD/btv6bv+quBfcffaFXdW1XjVTU+NjZ2pvNLkhah08arqv4KeDrJm7ql64FHgB3A5m5tM3B/t70DmEhyUZJ1wHrgwQWdWpK0qC2d437/FLgvyTLgCeAf0Qvf9iS3AU8BtwBU1e4k2+kF7jBwR1UdWfDJJUmL1pziVVW7gPFZLrr+JPvfDdx9FnNJknRSfsOGJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5xkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOYYL0lSc4yXJKk5c4pXkh8keTjJriST3dryJA8k2dudXtq3/11JppLsSXLDoIaXJC1O83nm9Y6qurqqxrvzW4CdVbUe2NmdJ8kGYAK4EtgI3JNkyQLOLEla5M7mZcNNwNZueytwc9/6tqo6VFVPAlPAdWdxP5IkHWOu8Srg60keSnJ7t3Z5Ve0H6E5XdOurgKf7rjvdrR0jye1JJpNMzszMnNn0kqRFaekc93tbVe1LsgJ4IMljp9g3s6zVCQtV9wL3AoyPj59wuSRJJzOnZ15Vta87PQh8ld7LgAeSrAToTg92u08Da/quvhrYt1ADS5J02ngleWWSVx/dBn4D+D6wA9jc7bYZuL/b3gFMJLkoyTpgPfDgQg8uSVq85vKy4eXAV5Mc3f8PqupPk3wb2J7kNuAp4BaAqtqdZDvwCHAYuKOqjgxkeknSonTaeFXVE8BVs6w/A1x/kuvcDdx91tNJkjQLv2FDktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktScOccryZIk303yte788iQPJNnbnV7at+9dSaaS7ElywyAGlyQtXvN55vVh4NG+81uAnVW1HtjZnSfJBmACuBLYCNyTZMnCjCtJ0hzjlWQ18PeA3+9b3gRs7ba3Ajf3rW+rqkNV9SQwBVy3MONKkjT3Z16fBv4F8HLf2uVVtR+gO13Rra8Cnu7bb7pbO0aS25NMJpmcmZmZ9+CSpMXrtPFKchNwsKoemuNtZpa1OmGh6t6qGq+q8bGxsTnetCRJsHQO+7wNeHeSG4GLgdck+S/AgSQrq2p/kpXAwW7/aWBN3/VXA/sWcmhJ0uJ22mdeVXVXVa2uqrX0Pojx36vqA8AOYHO322bg/m57BzCR5KIk64D1wIMLPrkkadGayzOvk/kksD3JbcBTwC0AVbU7yXbgEeAwcEdVHTnrSSVJ6qTqhLejzrnx8fGanJwc9hiSpBGS5KGqGp/tMr9hQ5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUnNPGK8nFSR5M8r0ku5P8m259eZIHkuztTi/tu85dSaaS7ElywyAfgCRp8ZnLM69DwN+pqquAq4GNSd4CbAF2VtV6YGd3niQbgAngSmAjcE+SJYMYXpK0OJ02XtXzQnf2wu6ngE3A1m59K3Bzt70J2FZVh6rqSWAKuG5Bp5YkLWpzes8ryZIku4CDwANV9S3g8qraD9Cdruh2XwU83Xf16W7t+Nu8PclkksmZmZmzeQySpEVmTvGqqiNVdTWwGrguyd86xe6Z7SZmuc17q2q8qsbHxsbmNq0kSczz04ZV9WPgf9B7L+tAkpUA3enBbrdpYE3f1VYD+856UkmSOnP5tOFYktd1268A3gk8BuwANne7bQbu77Z3ABNJLkqyDlgPPLjQg0uSFq+lc9hnJbC1+8TgBcD2qvpakr8Atie5DXgKuAWgqnYn2Q48AhwG7qiqI4MZX5K0GKXqhLejzrnx8fGanJwc9hiSpBGS5KGqGp/tMr9hQ5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUnNPGK8maJN9I8miS3Uk+3K0vT/JAkr3d6aV917kryVSSPUluGOQDkCQtPnN55nUY+GdV9TeBtwB3JNkAbAF2VtV6YGd3nu6yCeBKYCNwT5IlgxhekrQ4nTZeVbW/qr7Tbf8EeBRYBWwCtna7bQVu7rY3Aduq6lBVPQlMAdct9OCSpMVrXu95JVkLXAN8C7i8qvZDL3DAim63VcDTfVeb7taOv63bk0wmmZyZmZn/5JKkRWvO8UryKuDLwEeq6vlT7TrLWp2wUHVvVY1X1fjY2Nhcx5AkaW7xSnIhvXDdV1Vf6ZYPJFnZXb4SONitTwNr+q6+Gti3MONKkjS3TxsG+CzwaFX9dt9FO4DN3fZm4P6+9YkkFyVZB6wHHly4kSVJi93SOezzNuCDwMNJdnVr/wr4JLA9yW3AU8AtAFW1O8l24BF6n1S8o6qOLPjkkqRF67Txqqr/xezvYwFcf5Lr3A3cfRZzSZJ0Un7DhiSpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOaeNV5LPJTmY5Pt9a8uTPJBkb3d6ad9ldyWZSrInyQ2DGlyStHjN5ZnX54GNx61tAXZW1XpgZ3eeJBuACeDK7jr3JFmyYNNKksQc4lVVfwY8e9zyJmBrt70VuLlvfVtVHaqqJ4Ep4LoFmlWSJODM3/O6vKr2A3SnK7r1VcDTfftNd2snSHJ7kskkkzMzM2c4hiRpMVroD2xklrWabcequreqxqtqfGxsbIHHkCSdz840XgeSrAToTg9269PAmr79VgP7znw8SZJOdKbx2gFs7rY3A/f3rU8kuSjJOmA98ODZjShJ0rGWnm6HJF8Cfh24LMk08HHgk8D2JLcBTwG3AFTV7iTbgUeAw8AdVXVkQLNLkhap08arqm49yUXXn2T/u4G7z2YoSZJOxW/YkCQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktQc4yVJao7xkiQ1x3hJkppjvCRJzTFekqTmGC9JUnOMlySpOcZLktSc8yNeVbBrF+zcCT/72bCnkSQNWPvx+sxn4IIL4Jpr4J3vhGXL4M1vHvZUkrQovfwyHD48+PsZWLySbEyyJ8lUki0DuZOXXoLf/E0Aqu+Hhx+Gt751IHcpSTrRiy/Cxz8OGzbA+vUwMQF79gzu/gYSryRLgP8EvAvYANyaZMOC39Ell/wiWH0KqG9+c8HvTpI0uzvvhC98AS6+GJYvh8lJeN/74ODBwdzfoJ55XQdMVdUTVfVTYBuwaUD3JUkaoscfhz//8160li6FBF73OnjhBdi+fTD3Oah4rQKe7js/3a39XJLbk0wmmZyZmTmjOzn+GZck6dx76ilYsqQXrX4XXACPPTaY+xxUvDLL2jGtqap7q2q8qsbHxsbO6E5ePsVlhk2Szo03vKH3IY2Xj/tD+eWX4aqrBnOfg4rXNLCm7/xqYN9C38mSOnmi/vdC35kkaVZXXAE33QQ//jEcOtQL2bPP9l5GfO97B3Ofg4rXt4H1SdYlWQZMADsW/F6q+Otf33jCs6wfARt2PbXgdydJmt2nPgUf/Shccknvn9vedBN85Su9gA3C0kHcaFUdTnIn8N+AJcDnqmr3gt/Rd7/LK6anOLzhzbz4zEv87K8Pc+ElF7J82c9YsvdbcNWa09+GJOmsLVvW+8ThnXeem/sbSLwAqupPgD8Z1O0D8OSTACx9xYW8ZvWFv1h/9tnB/gMDSdJQtf0NG+vW9U6Pf+/rggvgTW869/NIks6JtuN1zTW9j7I891zvRdYjR3rbK1bAjTcOezpJ0oC0Ha8EPv95+OAHex9veeGFXrS+/OXeu4aSpPNS6hQfNz9XxsfHa3JycthjSJJGSJKHqmp8tsvafuYlSVqUjJckqTnGS5LUHOMlSWqO8ZIkNcd4SZKaY7wkSc0xXpKk5hgvSVJzjJckqTnGS5LUHOMlSWqO8ZIkNWckvlU+yQzwwwW4qcuAHy3A7QyDs597rc4N7c7e6tzQ7uytzg3wK1U1NtsFIxGvhZJk8mRfnz/qnP3ca3VuaHf2VueGdmdvde7T8WVDSVJzjJckqTnnW7zuHfYAZ8HZz71W54Z2Z291bmh39lbnPqXz6j0vSdLicL4985IkLQLGS5LUnPMmXkk2JtmTZCrJlmHP0y/JmiTfSPJokt1JPtytfyLJ/02yq/u5se86d3WPZU+SG4Y3PST5QZKHuxknu7XlSR5Isrc7vbRv/6HPnuRNfcd1V5Lnk3xkVI95ks8lOZjk+31r8z7GSX61+99qKsl/SJIhzf5vkzyW5C+TfDXJ67r1tUle6jv+nxnW7CeZe96/HyN0zP+wb+4fJNnVrY/MMV9QVdX8D7AEeBx4PbAM+B6wYdhz9c23Eri223418H+ADcAngH8+y/4busdwEbCue2xLhjj/D4DLjlv7FLCl294C/NYozt73+/FXwK+M6jEH3g5cC3z/bI4x8CDwViDAfwXeNaTZfwNY2m3/Vt/sa/v3O+52zunsJ5l73r8fo3LMj7v83wH/etSO+UL+nC/PvK4Dpqrqiar6KbAN2DTkmX6uqvZX1Xe67Z8AjwKrTnGVTcC2qjpUVU8CU/Qe4yjZBGzttrcCN/etj9rs1wOPV9WpvsVlqHNX1Z8Bz84y05yPcZKVwGuq6i+q9yfTF/quc05nr6qvV9Xh7uw3gdWnuo1hzH6SY34yI3/Mj+qePb0P+NKpbmNYsy+U8yVeq4Cn+85Pc+o4DE2StcA1wLe6pTu7l1Y+1/ey0Kg9ngK+nuShJLd3a5dX1X7oxRlY0a2P2uwAExz7H3ILxxzmf4xXddvHrw/bh+j9rf6odUm+m+R/Jvm1bm2UZp/P78cozX3UrwEHqmpv39qoH/N5O1/iNdvrtCP3bwCSvAr4MvCRqnoe+F3gDcDVwH56T/Vh9B7P26rqWuBdwB1J3n6KfUdq9iTLgHcDf9QttXLMT+Vks47cY0jyMeAwcF+3tB+4oqquAT4K/EGS1zA6s8/392NU5u53K8f+ZW3Uj/kZOV/iNQ2s6Tu/Gtg3pFlmleRCeuG6r6q+AlBVB6rqSFW9DPwev3iZaqQeT1Xt604PAl+lN+eB7mWHoy8/HOx2H6nZ6QX3O1V1ANo55p35HuNpjn15bqiPIclm4Cbg/d3LUnQvuz3TbT9E772jNzIis5/B78dIzH1UkqXAPwD+8OjaqB/zM3W+xOvbwPok67q/aU8AO4Y80891r0F/Fni0qn67b31l327vAY5+cmgHMJHkoiTrgPX03lg955K8Msmrj27TeyP++92Mm7vdNgP3d9sjM3vnmL+FtnDM+8zrGHcvLf4kyVu637l/2HedcyrJRuBfAu+uqhf71seSLOm2X09v9idGZfb5/n6Mytx93gk8VlU/fzlw1I/5GRv2J0YW6ge4kd6n+B4HPjbseY6b7W/Tezr+l8Cu7udG4IvAw936DmBl33U+1j2WPQzxE0D0PsH5ve5n99FjC/wSsBPY250uH8HZLwGeAV7btzaSx5xeYPcDP6P3N+LbzuQYA+P0/sB9HPiPdN+iM4TZp+i9R3T09/0z3b7v7X6Pvgd8B/j7w5r9JHPP+/djVI55t/554J8ct+/IHPOF/PHroSRJzTlfXjaUJC0ixkuS1BzjJUlqjvGSJDXHeEmSmmO8JEnNMV6SpOb8f9qWMkGu6I7nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X_pred = StandardScaler().fit_transform(features_pred)\n",
    "\n",
    "\n",
    "pca4 = PCA(n_components=4)\n",
    "x_4d = pca4.fit_transform(X_pred)\n",
    "\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans_pred = KMeans(n_clusters = 3)\n",
    "\n",
    "#Compute cluster centers and predict cluster indices\n",
    "X_clustered = kmeans_pred.fit_predict(x_4d)\n",
    "\n",
    "#Define our own color map\n",
    "LABEL_COLOR_MAP = {0:'r', 1: 'b', 2: 'purple'}\n",
    "label_color = [LABEL_COLOR_MAP[l] for l in X_clustered]\n",
    "\n",
    "# Plot the scatter digram\n",
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,2], c= label_color, alpha=0.8) \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_pred = StandardScaler().fit_transform(features_pred)\n",
    "\n",
    "pca4 = PCA(n_components=1)\n",
    "x_4d = pca4.fit_transform(X_pred)\n",
    "\n",
    "X_pred = StandardScaler().fit_transform(features_pred)\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 3)\n",
    "\n",
    "X_clustered= kmeans.fit(x_4d)\n",
    "cluster_map = pd.DataFrame()\n",
    "cluster_map['data_index']=features_df_pred.index.values\n",
    "cluster_map['cluster']=X_clustered.labels_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1418, 2)"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cluster_map.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        data_index  cluster\n",
      "0             star        0\n",
      "3      established        0\n",
      "4            carry        0\n",
      "5             give        0\n",
      "7          express        0\n",
      "10            play        0\n",
      "13            beat        0\n",
      "14           clash        0\n",
      "15       establish        0\n",
      "16           leave        0\n",
      "17            snub        0\n",
      "18            bury        0\n",
      "19     incorporate        0\n",
      "20          employ        0\n",
      "22           write        0\n",
      "23           serve        0\n",
      "24            bear        0\n",
      "25       introduce        0\n",
      "26         condemn        0\n",
      "27          oppose        0\n",
      "28          repeal        0\n",
      "29         deliver        0\n",
      "31           offer        0\n",
      "32            sink        0\n",
      "33        separate        0\n",
      "35           cause        0\n",
      "39        continue        0\n",
      "40       recommend        0\n",
      "42         feature        0\n",
      "43      contribute        0\n",
      "...            ...      ...\n",
      "1388    trailblaze        0\n",
      "1389         judge        0\n",
      "1390        pepper        0\n",
      "1391          land        0\n",
      "1392     corrupted        0\n",
      "1393      overlaid        0\n",
      "1394      postpone        0\n",
      "1395    substitute        0\n",
      "1396         slate        0\n",
      "1397          leak        0\n",
      "1398     hitchhike        0\n",
      "1399       lampoon        0\n",
      "1400         grace        0\n",
      "1401   proposition        0\n",
      "1402     patterned        0\n",
      "1403       elevate        0\n",
      "1404       molding        0\n",
      "1405         greet        0\n",
      "1406          lure        0\n",
      "1407      compound        0\n",
      "1408         reuse        0\n",
      "1409  administrate        0\n",
      "1410           pen        0\n",
      "1411      nurtures        0\n",
      "1412       justify        0\n",
      "1413         harry        0\n",
      "1414    liberalize        0\n",
      "1415          stir        0\n",
      "1416          load        0\n",
      "1417          raid        0\n",
      "\n",
      "[1345 rows x 2 columns]\n",
      "  data_index  cluster\n",
      "1         be        1\n",
      "       data_index  cluster\n",
      "2         appoint        2\n",
      "6            hold        2\n",
      "8            have        2\n",
      "9         consist        2\n",
      "11         remain        2\n",
      "12           lose        2\n",
      "21           lead        2\n",
      "30         follow        2\n",
      "34         manage        2\n",
      "36           link        2\n",
      "37         become        2\n",
      "38        provide        2\n",
      "41           form        2\n",
      "55          begin        2\n",
      "61         return        2\n",
      "63          found        2\n",
      "68           move        2\n",
      "69           name        2\n",
      "70           take        2\n",
      "72         accept        2\n",
      "83           vote        2\n",
      "97          build        2\n",
      "111          join        2\n",
      "123          know        2\n",
      "125        define        2\n",
      "129         merge        2\n",
      "147         learn        2\n",
      "153       connect        2\n",
      "182          seek        2\n",
      "186           see        2\n",
      "...           ...      ...\n",
      "356        center        2\n",
      "357         annex        2\n",
      "368          kill        2\n",
      "388     transport        2\n",
      "407          view        2\n",
      "418         speak        2\n",
      "460       rebuild        2\n",
      "468       surpass        2\n",
      "510        reveal        2\n",
      "529         drain        2\n",
      "530     surrender        2\n",
      "582          tell        2\n",
      "592     associate        2\n",
      "594   standardize        2\n",
      "632     transform        2\n",
      "645      contrast        2\n",
      "732       exclude        2\n",
      "748         title        2\n",
      "801     subsidize        2\n",
      "808        bypass        2\n",
      "847       reflect        2\n",
      "922         wound        2\n",
      "943        demand        2\n",
      "981          hide        2\n",
      "1037         shut        2\n",
      "1105       stress        2\n",
      "1117       centre        2\n",
      "1124       insist        2\n",
      "1193     suppress        2\n",
      "1235       sculpt        2\n",
      "\n",
      "[72 rows x 2 columns]\n"
     ]
    }
   ],
   "source": [
    "for i in range(0, 3):\n",
    "    print(cluster_map[cluster_map.cluster==i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "#Set a 3 KMeans clustering\n",
    "kmeans = KMeans(n_clusters = 3)\n",
    "\n",
    "#Compute cluster centers and predict cluster indices\n",
    "X_clustered = kmeans.fit_predict(x_4d)\n",
    "\n",
    "#Define our own color map\n",
    "LABEL_COLOR_MAP = {0:'r', 1: 'g', 2: 'b'}\n",
    "label_color = [LABEL_COLOR_MAP[l] for l in X_clustered]\n",
    "\n",
    "# Plot the scatter digram\n",
    "plt.figure(figsize = (7,7))\n",
    "plt.scatter(x_4d[:,0],x_4d[:,2], c= label_color, alpha=0.5) \n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "# Create a temp dataframe from our PCA projection data \"x_9d\"\n",
    "df = pd.DataFrame(x_4d)\n",
    "df = df[[0,1,2]] # only want to visualise relationships between first 3 projections\n",
    "df['X_cluster'] = X_clustered\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/seaborn/axisgrid.py:2065: UserWarning: The `size` parameter has been renamed to `height`; pleaes update your code.\n",
      "  warnings.warn(msg, UserWarning)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/numpy/core/_methods.py:140: RuntimeWarning: Degrees of freedom <= 0 for slice\n",
      "  keepdims=keepdims)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/numpy/core/_methods.py:132: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  ret = ret.dtype.type(ret / rcount)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kde.py:487: RuntimeWarning: invalid value encountered in true_divide\n",
      "  binned = fast_linbin(X, a, b, gridsize) / (delta * nobs)\n",
      "/home/yiyi/anaconda3/envs/semantik/lib/python3.6/site-packages/statsmodels/nonparametric/kdetools.py:34: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  FAC1 = 2*(np.pi*bw/RANGE)**2\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.PairGrid at 0x7fa4968baa90>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 586.8x532.8 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Call Seaborn's pairplot to visualize our KMeans clustering on the PCA projected data\n",
    "sns.pairplot(df, hue='X_cluster', palette= 'Dark2', diag_kind='kde',size=1.85)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# toy demo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>(Shakespeare, Macbeth)</th>\n",
       "      <th>(Dickens, Oliver Twist)</th>\n",
       "      <th>(Rowling, Harry Potter)</th>\n",
       "      <th>(Council, Dublin)</th>\n",
       "      <th>(Sharon, Ronnie)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>write_{arg0:PER,arg1:Book}</th>\n",
       "      <td>10.63</td>\n",
       "      <td>20.63</td>\n",
       "      <td>10.63</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>suggest_{arg0:ORG, arg1:LOC}</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>10.12</td>\n",
       "      <td>0.63</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>suggest_{arg0:PER, arg1:PER}</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.12</td>\n",
       "      <td>5.63</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                              (Shakespeare, Macbeth)  (Dickens, Oliver Twist)  \\\n",
       "write_{arg0:PER,arg1:Book}                     10.63                    20.63   \n",
       "suggest_{arg0:ORG, arg1:LOC}                    0.25                     0.25   \n",
       "suggest_{arg0:PER, arg1:PER}                    0.25                     0.25   \n",
       "\n",
       "                              (Rowling, Harry Potter)  (Council, Dublin)  \\\n",
       "write_{arg0:PER,arg1:Book}                      10.63               0.00   \n",
       "suggest_{arg0:ORG, arg1:LOC}                     0.25              10.12   \n",
       "suggest_{arg0:PER, arg1:PER}                     0.25               0.12   \n",
       "\n",
       "                              (Sharon, Ronnie)  \n",
       "write_{arg0:PER,arg1:Book}                0.27  \n",
       "suggest_{arg0:ORG, arg1:LOC}              0.63  \n",
       "suggest_{arg0:PER, arg1:PER}              5.63  "
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lists=[[10.63,20.63,10.63,0,0.27],[0.25,0.25,0.25,10.12,0.63],[0.25,0.25,0.25,0.12,5.63]]\n",
    "darray= np.asarray(lists)\n",
    "df = pd.DataFrame(darray)\n",
    "# df\n",
    "# df = df.rename(index=)\n",
    "args= [ (\"Shakespeare\", \"Macbeth\") , (\"Dickens\", \"Oliver Twist\") , (\"Rowling\", \"Harry Potter\") , (\"Council\", \"Dublin\"), (\"Sharon\", \"Ronnie\")]\n",
    "deargs={idx:args for idx, args in enumerate(args)}\n",
    "df=df.rename(columns=deargs)\n",
    "df=df.rename(index= {0:\"write_{arg0:PER,arg1:Book}\",1:\"suggest_{arg0:ORG, arg1:LOC}\",2:\"suggest_{arg0:PER, arg1:PER}\"})\n",
    "\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}