hyp-functionality.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Hypothesis Functionality\n",
    "\n",
    "## Goals\n",
    "We want a hypothesis that is:\n",
    "- Simple. As few nodes as possible.\n",
    "- Probable. Should have a high probability of being true\n",
    "\n",
    "## Important Functions\n",
    "- Write the Graph\n",
    "- Learn the CPT Table\n",
    "- ObserveData: Should be done after CPT Table is created but before hypothesis testing\n",
    "- findBestExplanation given the observed data and the graph\n",
    "\n",
    "## TO DO:\n",
    "- USE SIMILARITY TO JUDGE MULTI-NODE HYPOTHESIS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from similarityfunctions import *\n",
    "import networkx as nx\n",
    "from itertools import permutations, combinations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# add observations to a graph\n",
    "def observeData(graph, true_nodes, false_nodes):\n",
    "    nodes = graph.nodes()\n",
    "    for i in nodes:\n",
    "        if i in true_nodes:\n",
    "            graph.nodes[i]['value'] = 1\n",
    "        elif i in false_nodes:\n",
    "            graph.nodes[i]['value'] = 0\n",
    "        else:\n",
    "            graph.nodes[i]['value'] = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CPT(object):\n",
    "    \"\"\"\n",
    "    Defines a CPT Class\n",
    "    \"\"\"\n",
    "    def __init__(self, num_parents):\n",
    "        self.num_parents = num_parents\n",
    "        self.CPTable = self.make_table()\n",
    "        \n",
    "    def make_table(self):\n",
    "        CPTable = {}\n",
    "        if self.num_parents > 0:\n",
    "            for i in range(2**self.num_parents):\n",
    "                CPTable[bin(i)] = 0.0\n",
    "        else:\n",
    "            CPTable['self'] = 0.0\n",
    "        return CPTable\n",
    "    \n",
    "    def add_entry(self, parent_values, prob):\n",
    "        # values of nodes sorted in alphabetical order\n",
    "        key = ''\n",
    "        for i in parent_values:\n",
    "            key += str(i)\n",
    "        self.CPTable[bin(int(key, 2))] = prob \n",
    "    \n",
    "    def add_entry_self(self, prob):\n",
    "        self.CPTable['self'] = prob\n",
    "    \n",
    "    def get_entry(self, parent_values, value):\n",
    "        key = ''\n",
    "        for i in parent_values:\n",
    "            key += str(i)\n",
    "        return self.CPTable[bin(int(key, 2))][value]\n",
    "    \n",
    "    def get_entry_bin(self, parent_bin, value):\n",
    "        return self.CPTable[parent_bin][value]\n",
    "    \n",
    "    def get_entry_self(self, value):\n",
    "        return self.CPTable['self'][value]\n",
    "    \n",
    "    def get_table(self):\n",
    "        return self.CPTable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from itertools import product\n",
    "def calculateMarginalProbability(blanket, knodes):\n",
    "    sorted_nodes = sorted(blanket.nodes())\n",
    "    # want to calculate P(hyp | obs)\n",
    "    # first, calculate marginal probability of P(hyp, obs, nodes)\n",
    "    \n",
    "    unodes = [item for item in sorted_nodes if (not item in knodes)]\n",
    "    keys = list(product(range(2), repeat=len(unodes)))\n",
    "    prob = 0\n",
    "    for i in keys:\n",
    "        p = 1\n",
    "        for x in sorted_nodes:\n",
    "            # find the parent values of a node\n",
    "            parents = sorted(list(blanket.predecessors(x)))\n",
    "            # if it has parents, get the key and get the probability\n",
    "            if parents:\n",
    "                # get the parents:\n",
    "                parent_values = []\n",
    "                for z in parents:\n",
    "                    if z in knodes:\n",
    "                        parent_values.append(blanket.node[z]['value'])\n",
    "                    else:\n",
    "                        parent_values.append(i[unodes.index(z)])\n",
    "                \n",
    "                parents_key = bin(int(''.join(map(str, parent_values)), 2))\n",
    "\n",
    "                if x in knodes:\n",
    "                    p *= blanket.node[x]['CPT'].get_entry_bin(parents_key, blanket.node[x]['value'])\n",
    "                else:\n",
    "                    p *= blanket.node[x]['CPT'].get_entry_bin(parents_key, i[unodes.index(x)])\n",
    "            # if it doesn't have parents, get the entry for itself based on the node value\n",
    "            else:\n",
    "                if x in knodes:\n",
    "                    p *= blanket.node[x]['CPT'].get_entry_self(blanket.node[x]['value'])\n",
    "                else:\n",
    "                    p *= blanket.node[x]['CPT'].get_entry_self(i[unodes.index(x)])\n",
    "        prob += p\n",
    "    return prob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculateTotalMarginalProbability(blanket, hyp, obs, bnodes):\n",
    "    # hyp | obs\n",
    "    # obs | hyp\n",
    "    all_obs = []\n",
    "    all_obs += obs\n",
    "    all_obs += hyp\n",
    "    all_obs += bnodes\n",
    "    new_obs = list(obs) + list(bnodes)\n",
    "    num_prob = calculateMarginalProbability(blanket, all_obs)\n",
    "    denom_prob = calculateMarginalProbability(blanket, new_obs)\n",
    "    return num_prob / denom_prob\n",
    "\n",
    "def calculateTotalMarginalProbabilityObs(blanket, hyp, obs, bnodes):\n",
    "    all_obs = []\n",
    "    all_obs += obs\n",
    "    all_obs += hyp\n",
    "    all_obs += bnodes\n",
    "    new_hyp = list(hyp) + list(bnodes)\n",
    "    num_prob = calculateMarginalProbability(blanket, all_obs)\n",
    "    denom_prob = calculateMarginalProbability(blanket, new_hyp)\n",
    "    return num_prob / denom_prob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Finds all possible hypothesis\n",
    "def combinationsList(my_list, start):\n",
    "    total = []\n",
    "    for i in range(start, len(my_list) + 1):\n",
    "        total += list(combinations(my_list, i))\n",
    "    return total\n",
    "\n",
    "def findHypotheses(graph, observed_nodes, bnodes):\n",
    "    all_node_combos = combinationsList(graph.nodes(), 1)\n",
    "    hyps = []\n",
    "    for i in all_node_combos:\n",
    "        children = {}\n",
    "        for x in i:\n",
    "            children.update(get_all_children(graph, x, {}, 1))\n",
    "        if all(item in children.keys() for item in observed_nodes):\n",
    "            if not any(item in observed_nodes for item in i) and not any(item in bnodes for item in i):\n",
    "                observed_children = {}\n",
    "                for x in observed_nodes:\n",
    "                    observed_children.update(get_all_children(graph, x, {}, 1))\n",
    "                if not any(item in observed_children for item in i) and not too_similar(graph, i):\n",
    "                    hyps.append(list(i))\n",
    "    return hyps\n",
    "\n",
    "def too_similar(graph, hyp_list, thresh=.5):\n",
    "    pairs = list(combinations(hyp_list, 2))\n",
    "    for pair in pairs:\n",
    "        # if they're too similar\n",
    "        sim_1 = get_full_similarity(graph, pair[0], pair[1])\n",
    "        # print(pair, sim_1)\n",
    "        if sim_1 > thresh or sim_1 == 0:\n",
    "            return True\n",
    "    return False\n",
    "\n",
    "# Finds a bunch of hypothesis. Does nothing yet\n",
    "def findBestExplanation(graph, observed_nodes, bnodes, flipped=False, hyps=None):\n",
    "    if hyps is None:\n",
    "        hyps = findHypotheses(graph, observed_nodes, bnodes)\n",
    "    current_best = (None, 0)\n",
    "    hyp_dict = {}\n",
    "    for i in hyps:\n",
    "        if isinstance(i, list) or isinstance(i, tuple): \n",
    "            for x in i:\n",
    "                graph.nodes[x]['value'] = 1\n",
    "            if flipped:\n",
    "                prob = calculateTotalMarginalProbabilityObs(graph, list(i), observed_nodes, bnodes)\n",
    "            else:\n",
    "                prob = calculateTotalMarginalProbability(graph, list(i), observed_nodes, bnodes)\n",
    "            for x in i:\n",
    "                graph.nodes[x]['value'] = None\n",
    "        else:\n",
    "            graph.nodes[i]['value'] = 1\n",
    "            if flipped:\n",
    "                prob = calculateTotalMarginalProbabilityObs(graph, list(i), observed_nodes, bnodes)\n",
    "            else:\n",
    "                prob = calculateTotalMarginalProbability(graph, list(i), observed_nodes, bnodes)\n",
    "            graph.nodes[i]['value'] = None\n",
    "        if prob > current_best[1]:\n",
    "            current_best = (i, prob)\n",
    "        hyp_dict.update({tuple(i): prob})\n",
    "    return (hyp_dict, current_best)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def hypothesisSelection(graph, observed_nodes, bnodes, n):\n",
    "    # observe the data\n",
    "    \n",
    "    # first, o | h to\n",
    "    hyp_dict, current_best = findBestExplanation(graph, observed_nodes, bnodes, flipped=True)\n",
    "    # get the nth percent (4th, third, half)\n",
    "    updated_hyps = {}\n",
    "    lim = int(len(hyp_dict)/n)\n",
    "    for i in range(lim):\n",
    "        maximum = max(hyp_dict.values())\n",
    "        max_key = next((key for key, val in hyp_dict.items() if maximum == val), None)\n",
    "        updated_hyps.update({max_key: maximum})\n",
    "        del hyp_dict[max_key]\n",
    "    new_hyps, final_best = findBestExplanation(graph, observed_nodes, bnodes, hyps=updated_hyps.keys())\n",
    "    return final_best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def graph_tests(graph, n=2):\n",
    "    obs_nodes = [item for item in graph.nodes() if len(nx.ancestors(graph, item)) > 0]\n",
    "    obs_nodes_combos = combinationsList(obs_nodes, 1)\n",
    "    back_nodes_combos = combinationsList(graph.nodes(), 0)\n",
    "    \n",
    "    final_list = []\n",
    "    for i in back_nodes_combos:\n",
    "        for x in obs_nodes_combos:\n",
    "            if not (any(item in i for item in x) and any(item in x for item in i)):\n",
    "                children = get_all_children(graph, x, {}, 1)\n",
    "                for i_bin in product(range(2), repeat=len(i)):\n",
    "                    for x_bin in product(range(2), repeat=len(x)):\n",
    "                        true_i = [i[z] for z in range(len(i_bin)) if i_bin[z] == 1] \n",
    "                        false_i = [i[z] for z in range(len(i_bin)) if i_bin[z] == 0] \n",
    "                        true_x = [x[z] for z in range(len(x_bin)) if x_bin[z] == 1] \n",
    "                        false_x = [x[z] for z in range(len(x_bin)) if x_bin[z] == 0] \n",
    "                        observeData(graph, true_i + true_x, false_i + false_x)\n",
    "                        hyp, prob = hypothesisSelection(graph, x, i, n)\n",
    "                        final_list.append([(true_x, false_x, true_i, false_i, hyp), prob])\n",
    "    return final_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Burgler Tests\n",
    "Based off the example found [HERE](https://www.ics.uci.edu/~rickl/courses/cs-171/2012-wq-cs171/2012-wq-cs171-lecture-slides/2012wq171-17-BayesianNetworks.pdf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "bdg = nx.DiGraph()\n",
    "bdg.add_nodes_from(['B', 'E', 'A', 'M', 'J'])\n",
    "bdg.add_weighted_edges_from([('B','A', .94002),('E','A', .29066),('A', 'M',.7),('A','J',.9)])\n",
    "nx.draw_networkx(bdg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# learn the CPTable\n",
    "\n",
    "# burglar\n",
    "cpt_B = CPT(0)\n",
    "cpt_B.add_entry_self((1-.001, .001))\n",
    "bdg.node['B']['CPT'] = cpt_B\n",
    "# earthquake\n",
    "cpt_E = CPT(0)\n",
    "cpt_E.add_entry_self((1-.002, .002))\n",
    "bdg.node['E']['CPT'] = cpt_E\n",
    "#alarm\n",
    "cpt_A = CPT(2)\n",
    "cpt_A.add_entry([1, 1], (1-.95, .95))\n",
    "cpt_A.add_entry([1, 0], (1-.94, .94))\n",
    "cpt_A.add_entry([0, 1], (1-.29, .29))\n",
    "cpt_A.add_entry([0, 0], (1-.001, .001))\n",
    "bdg.node['A']['CPT'] = cpt_A\n",
    "#john calls\n",
    "cpt_J = CPT(1)\n",
    "cpt_J.add_entry([1], (1-.9, .9))\n",
    "cpt_J.add_entry([0], (1-.05, .05))\n",
    "bdg.node['J']['CPT'] = cpt_J\n",
    "#mary calls\n",
    "cpt_M = CPT(1)\n",
    "cpt_M.add_entry([1], (1-.7, .7))\n",
    "cpt_M.add_entry([0], (1-.01, .01))\n",
    "bdg.node['M']['CPT'] = cpt_M"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "observeData(bdg, ['J', 'M'], [])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[['B'], ['E'], ['A'], ['B', 'A']]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "findHypotheses(bdg, ['M', 'J'], [])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('B',): 0.2841718353643929,\n",
       "  ('E',): 0.17606683840507922,\n",
       "  ('A',): 0.7606920388631078,\n",
       "  ('B', 'A'): 0.284157445461528},\n",
       " (['A'], 0.7606920388631078))"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "findBestExplanation(bdg, ['M', 'J'], [])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('B',): 0.5922425899999999,\n",
       "  ('E',): 0.18347047,\n",
       "  ('A',): 0.63,\n",
       "  ('B', 'A'): 0.6299999999999999},\n",
       " (['A'], 0.63))"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "findBestExplanation(bdg, ['M', 'J'], [], flipped=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(('A',), 0.7606920388631078)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hypothesisSelection(bdg, ['M', 'J'], [], n=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "bdg_results_unclean = graph_tests(bdg)\n",
    "bdg_results_clean = [item for item in bdg_results_unclean if item[1] != 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[([], ['A'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [(['A'], [], [], [], ('B',)), 0.373551228281836],\n",
       " [([], ['M'], [], [], ('E',)), 0.0015976396500871493],\n",
       " [(['M'], [], [], [], ('A',)), 0.15009011774975964],\n",
       " [([], ['J'], [], [], ('E',)), 0.0014832111079134708],\n",
       " [(['J'], [], [], [], ('A',)), 0.043437711799927066],\n",
       " [([], ['A', 'M'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [(['M'], ['A'], [], [], ('E',)), 0.0014222590323639199],\n",
       " [(['A'], ['M'], [], [], ('B',)), 0.3735512282818359],\n",
       " [(['A', 'M'], [], [], [], ('B',)), 0.373551228281836],\n",
       " [([], ['A', 'J'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [(['J'], ['A'], [], [], ('E',)), 0.0014222590323639199],\n",
       " [(['A'], ['J'], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A', 'J'], [], [], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['M', 'J'], [], [], ('E',)), 0.0014407327765951661],\n",
       " [(['J'], ['M'], [], [], ('A',)), 0.013573889331307635],\n",
       " [(['M'], ['J'], [], [], ('A',)), 0.01824973824500748],\n",
       " [(['M', 'J'], [], [], [], ('A',)), 0.7606920388631078],\n",
       " [([], ['A', 'M', 'J'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [(['J'], ['A', 'M'], [], [], ('E',)), 0.0014222590323639196],\n",
       " [(['M'], ['A', 'J'], [], [], ('E',)), 0.00142225903236392],\n",
       " [(['M', 'J'], ['A'], [], [], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], ['M', 'J'], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A', 'J'], ['M'], [], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'M'], ['J'], [], [], ('B',)), 0.3735512282818359],\n",
       " [(['A', 'M', 'J'], [], [], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['A'], [], ['B'], None), 0],\n",
       " [(['A'], [], [], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], [], None), 0],\n",
       " [(['A'], [], ['B'], [], None), 0],\n",
       " [([], ['M'], [], ['B'], ('E',)), 0.0015975145513068226],\n",
       " [(['M'], [], [], ['B'], ('A',)), 0.09961384529643373],\n",
       " [([], ['M'], ['B'], [], ('E',)), 0.0019596574202472156],\n",
       " [(['M'], [], ['B'], [], ('A',)), 0.9990892993739275],\n",
       " [([], ['J'], [], ['B'], ('E',)), 0.0014831466785683834],\n",
       " [(['J'], [], [], ['B'], ('A',)), 0.027661940776723615],\n",
       " [([], ['J'], ['B'], [], ('E',)), 0.0018876297331487648],\n",
       " [(['J'], [], ['B'], [], ('A',)), 0.9964676796813255],\n",
       " [([], ['A', 'M'], [], ['B'], None), 0],\n",
       " [(['M'], ['A'], [], ['B'], None), 0],\n",
       " [(['A'], ['M'], [], ['B'], None), 0],\n",
       " [(['A', 'M'], [], [], ['B'], None), 0],\n",
       " [([], ['A', 'M'], ['B'], [], None), 0],\n",
       " [(['M'], ['A'], ['B'], [], None), 0],\n",
       " [(['A'], ['M'], ['B'], [], None), 0],\n",
       " [(['A', 'M'], [], ['B'], [], None), 0],\n",
       " [([], ['A', 'J'], [], ['B'], None), 0],\n",
       " [(['J'], ['A'], [], ['B'], None), 0],\n",
       " [(['A'], ['J'], [], ['B'], None), 0],\n",
       " [(['A', 'J'], [], [], ['B'], None), 0],\n",
       " [([], ['A', 'J'], ['B'], [], None), 0],\n",
       " [(['J'], ['A'], ['B'], [], None), 0],\n",
       " [(['A'], ['J'], ['B'], [], None), 0],\n",
       " [(['A', 'J'], [], ['B'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['B'], ('E',)), 0.001440701707124981],\n",
       " [(['J'], ['M'], [], ['B'], ('A',)), 0.008547192182370435],\n",
       " [(['M'], ['J'], [], ['B'], ('A',)), 0.011511683274678913],\n",
       " [(['M', 'J'], [], [], ['B'], ('A',)), 0.6657108744026019],\n",
       " [([], ['M', 'J'], ['B'], [], ('E',)), 0.0017852122026965744],\n",
       " [(['J'], ['M'], ['B'], [], ('A',)), 0.9884372823600297],\n",
       " [(['M'], ['J'], ['B'], [], ('A',)), 0.9914148016888648],\n",
       " [(['M', 'J'], [], ['B'], [], ('A',)), 0.99994936196669],\n",
       " [([], ['A', 'M', 'J'], [], ['B'], None), 0],\n",
       " [(['J'], ['A', 'M'], [], ['B'], None), 0],\n",
       " [(['M'], ['A', 'J'], [], ['B'], None), 0],\n",
       " [(['M', 'J'], ['A'], [], ['B'], None), 0],\n",
       " [(['A'], ['M', 'J'], [], ['B'], None), 0],\n",
       " [(['A', 'J'], ['M'], [], ['B'], None), 0],\n",
       " [(['A', 'M'], ['J'], [], ['B'], None), 0],\n",
       " [(['A', 'M', 'J'], [], [], ['B'], None), 0],\n",
       " [([], ['A', 'M', 'J'], ['B'], [], None), 0],\n",
       " [(['J'], ['A', 'M'], ['B'], [], None), 0],\n",
       " [(['M'], ['A', 'J'], ['B'], [], None), 0],\n",
       " [(['M', 'J'], ['A'], ['B'], [], None), 0],\n",
       " [(['A'], ['M', 'J'], ['B'], [], None), 0],\n",
       " [(['A', 'J'], ['M'], ['B'], [], None), 0],\n",
       " [(['A', 'M'], ['J'], ['B'], [], None), 0],\n",
       " [(['A', 'M', 'J'], [], ['B'], [], None), 0],\n",
       " [([], ['A'], [], ['E'], None), 0],\n",
       " [(['A'], [], [], ['E'], None), 0],\n",
       " [([], ['A'], ['E'], [], None), 0],\n",
       " [(['A'], [], ['E'], [], None), 0],\n",
       " [([], ['M'], [], ['E'], ('B',)), 0.0003453151521163314],\n",
       " [(['M'], [], [], ['E'], ('A',)), 0.11971342160945007],\n",
       " [([], ['M'], ['E'], [], ('B',)), 0.0004237156096830609],\n",
       " [(['M'], [], ['E'], [], ('B', 'A')), 0.003158313678965251],\n",
       " [([], ['J'], [], ['E'], ('B',)), 0.00015922360461467972],\n",
       " [(['J'], [], [], ['E'], ('A',)), 0.033788238300887834],\n",
       " [([], ['J'], ['E'], [], ('B',)), 0.00020272029294149277],\n",
       " [(['J'], [], ['E'], [], ('A',)), 0.8806070133743574],\n",
       " [([], ['A', 'M'], [], ['E'], None), 0],\n",
       " [(['M'], ['A'], [], ['E'], None), 0],\n",
       " [(['A'], ['M'], [], ['E'], None), 0],\n",
       " [(['A', 'M'], [], [], ['E'], None), 0],\n",
       " [([], ['A', 'M'], ['E'], [], None), 0],\n",
       " [(['M'], ['A'], ['E'], [], None), 0],\n",
       " [(['A'], ['M'], ['E'], [], None), 0],\n",
       " [(['A', 'M'], [], ['E'], [], None), 0],\n",
       " [([], ['A', 'J'], [], ['E'], None), 0],\n",
       " [(['J'], ['A'], [], ['E'], None), 0],\n",
       " [(['A'], ['J'], [], ['E'], None), 0],\n",
       " [(['A', 'J'], [], [], ['E'], None), 0],\n",
       " [([], ['A', 'J'], ['E'], [], None), 0],\n",
       " [(['J'], ['A'], ['E'], [], None), 0],\n",
       " [(['A'], ['J'], ['E'], [], None), 0],\n",
       " [(['A', 'J'], [], ['E'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['E'], ('B',)), 9.015328226329397e-05],\n",
       " [(['J'], ['M'], [], ['E'], ('A',)), 0.010485794020393492],\n",
       " [(['M'], ['J'], [], ['E'], ('A',)), 0.014113094727792633],\n",
       " [(['M', 'J'], [], [], ['E'], ('A',)), 0.7099672469001376],\n",
       " [([], ['M', 'J'], ['E'], [], ('B',)), 0.00011174749602380887],\n",
       " [(['J'], ['M'], ['E'], [], ('B', 'A')), 0.002258110777368501],\n",
       " [(['M'], ['J'], ['E'], [], ('A',)), 0.7512000215618058],\n",
       " [(['M', 'J'], [], ['E'], [], ('B', 'A')), 0.003262105340439799],\n",
       " [([], ['A', 'M', 'J'], [], ['E'], None), 0],\n",
       " [(['J'], ['A', 'M'], [], ['E'], None), 0],\n",
       " [(['M'], ['A', 'J'], [], ['E'], None), 0],\n",
       " [(['M', 'J'], ['A'], [], ['E'], None), 0],\n",
       " [(['A'], ['M', 'J'], [], ['E'], None), 0],\n",
       " [(['A', 'J'], ['M'], [], ['E'], None), 0],\n",
       " [(['A', 'M'], ['J'], [], ['E'], None), 0],\n",
       " [(['A', 'M', 'J'], [], [], ['E'], None), 0],\n",
       " [([], ['A', 'M', 'J'], ['E'], [], None), 0],\n",
       " [(['J'], ['A', 'M'], ['E'], [], None), 0],\n",
       " [(['M'], ['A', 'J'], ['E'], [], None), 0],\n",
       " [(['M', 'J'], ['A'], ['E'], [], None), 0],\n",
       " [(['A'], ['M', 'J'], ['E'], [], None), 0],\n",
       " [(['A', 'J'], ['M'], ['E'], [], None), 0],\n",
       " [(['A', 'M'], ['J'], ['E'], [], None), 0],\n",
       " [(['A', 'M', 'J'], [], ['E'], [], None), 0],\n",
       " [([], ['M'], [], ['A'], ('B',)), 6.013131697154258e-05],\n",
       " [(['M'], [], [], ['A'], ('B',)), 6.013131697154256e-05],\n",
       " [([], ['M'], ['A'], [], ('B',)), 0.3735512282818359],\n",
       " [(['M'], [], ['A'], [], ('B',)), 0.373551228281836],\n",
       " [([], ['J'], [], ['A'], ('E',)), 0.0014222590323639203],\n",
       " [(['J'], [], [], ['A'], ('B',)), 6.0131316971542564e-05],\n",
       " [([], ['J'], ['A'], [], ('E',)), 0.23100870196889103],\n",
       " [(['J'], [], ['A'], [], ('E',)), 0.23100870196889098],\n",
       " [([], ['M', 'J'], [], ['A'], ('B',)), 6.013131697154257e-05],\n",
       " [(['J'], ['M'], [], ['A'], ('B',)), 6.013131697154255e-05],\n",
       " [(['M'], ['J'], [], ['A'], ('B',)), 6.013131697154256e-05],\n",
       " [(['M', 'J'], [], [], ['A'], ('B',)), 6.0131316971542544e-05],\n",
       " [([], ['M', 'J'], ['A'], [], ('B',)), 0.373551228281836],\n",
       " [(['J'], ['M'], ['A'], [], ('B',)), 0.37355122828183596],\n",
       " [(['M'], ['J'], ['A'], [], ('E',)), 0.23100870196889095],\n",
       " [(['M', 'J'], [], ['A'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['A'], [], ['M'], ('E',)), 0.0014222590323639203],\n",
       " [(['A'], [], [], ['M'], ('B',)), 0.3735512282818359],\n",
       " [([], ['A'], ['M'], [], ('E',)), 0.0014222590323639199],\n",
       " [(['A'], [], ['M'], [], ('B',)), 0.373551228281836],\n",
       " [([], ['J'], [], ['M'], ('E',)), 0.0014407327765951661],\n",
       " [(['J'], [], [], ['M'], ('A',)), 0.013573889331307635],\n",
       " [([], ['J'], ['M'], [], ('B',)), 0.006876246073421025],\n",
       " [(['J'], [], ['M'], [], ('A',)), 0.7606920388631078],\n",
       " [([], ['A', 'J'], [], ['M'], ('E',)), 0.0014222590323639203],\n",
       " [(['J'], ['A'], [], ['M'], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], ['J'], [], ['M'], ('B',)), 0.373551228281836],\n",
       " [(['A', 'J'], [], [], ['M'], ('B',)), 0.37355122828183596],\n",
       " [([], ['A', 'J'], ['M'], [], ('E',)), 0.00142225903236392],\n",
       " [(['J'], ['A'], ['M'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], ['J'], ['M'], [], ('B',)), 0.3735512282818359],\n",
       " [(['A', 'J'], [], ['M'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['A'], [], ['J'], ('E',)), 0.0014222590323639203],\n",
       " [(['A'], [], [], ['J'], ('B',)), 0.373551228281836],\n",
       " [([], ['A'], ['J'], [], ('E',)), 0.0014222590323639199],\n",
       " [(['A'], [], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['M'], [], ['J'], ('E',)), 0.0014407327765951661],\n",
       " [(['M'], [], [], ['J'], ('A',)), 0.01824973824500748],\n",
       " [([], ['M'], ['J'], [], ('B',)), 0.005129858133401301],\n",
       " [(['M'], [], ['J'], [], ('A',)), 0.7606920388631078],\n",
       " [([], ['A', 'M'], [], ['J'], ('E',)), 0.0014222590323639203],\n",
       " [(['M'], ['A'], [], ['J'], ('E',)), 0.00142225903236392],\n",
       " [(['A'], ['M'], [], ['J'], ('B',)), 0.373551228281836],\n",
       " [(['A', 'M'], [], [], ['J'], ('B',)), 0.3735512282818359],\n",
       " [([], ['A', 'M'], ['J'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['M'], ['A'], ['J'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], ['M'], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'M'], [], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['A'], [], ['B', 'E'], None), 0],\n",
       " [(['A'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['A'], ['E'], ['B'], None), 0],\n",
       " [(['A'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['E'], None), 0],\n",
       " [(['A'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['A'], ['B', 'E'], [], None), 0],\n",
       " [(['A'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'E'], None), 0],\n",
       " [(['M'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['M'], ['E'], ['B'], None), 0],\n",
       " [(['M'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['E'], None), 0],\n",
       " [(['M'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['M'], ['B', 'E'], [], None), 0],\n",
       " [(['M'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'E'], None), 0],\n",
       " [(['J'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['J'], ['E'], ['B'], None), 0],\n",
       " [(['J'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['E'], None), 0],\n",
       " [(['J'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['J'], ['B', 'E'], [], None), 0],\n",
       " [(['J'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['A', 'M'], [], ['B', 'E'], None), 0],\n",
       " [(['M'], ['A'], [], ['B', 'E'], None), 0],\n",
       " [(['A'], ['M'], [], ['B', 'E'], None), 0],\n",
       " [(['A', 'M'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['A', 'M'], ['E'], ['B'], None), 0],\n",
       " [(['M'], ['A'], ['E'], ['B'], None), 0],\n",
       " [(['A'], ['M'], ['E'], ['B'], None), 0],\n",
       " [(['A', 'M'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['A', 'M'], ['B'], ['E'], None), 0],\n",
       " [(['M'], ['A'], ['B'], ['E'], None), 0],\n",
       " [(['A'], ['M'], ['B'], ['E'], None), 0],\n",
       " [(['A', 'M'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['A', 'M'], ['B', 'E'], [], None), 0],\n",
       " [(['M'], ['A'], ['B', 'E'], [], None), 0],\n",
       " [(['A'], ['M'], ['B', 'E'], [], None), 0],\n",
       " [(['A', 'M'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['A', 'J'], [], ['B', 'E'], None), 0],\n",
       " [(['J'], ['A'], [], ['B', 'E'], None), 0],\n",
       " [(['A'], ['J'], [], ['B', 'E'], None), 0],\n",
       " [(['A', 'J'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['A', 'J'], ['E'], ['B'], None), 0],\n",
       " [(['J'], ['A'], ['E'], ['B'], None), 0],\n",
       " [(['A'], ['J'], ['E'], ['B'], None), 0],\n",
       " [(['A', 'J'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['A', 'J'], ['B'], ['E'], None), 0],\n",
       " [(['J'], ['A'], ['B'], ['E'], None), 0],\n",
       " [(['A'], ['J'], ['B'], ['E'], None), 0],\n",
       " [(['A', 'J'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['A', 'J'], ['B', 'E'], [], None), 0],\n",
       " [(['J'], ['A'], ['B', 'E'], [], None), 0],\n",
       " [(['A'], ['J'], ['B', 'E'], [], None), 0],\n",
       " [(['A', 'J'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['B', 'E'], None), 0],\n",
       " [(['J'], ['M'], [], ['B', 'E'], None), 0],\n",
       " [(['M'], ['J'], [], ['B', 'E'], None), 0],\n",
       " [(['M', 'J'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['M', 'J'], ['E'], ['B'], None), 0],\n",
       " [(['J'], ['M'], ['E'], ['B'], None), 0],\n",
       " [(['M'], ['J'], ['E'], ['B'], None), 0],\n",
       " [(['M', 'J'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['M', 'J'], ['B'], ['E'], None), 0],\n",
       " [(['J'], ['M'], ['B'], ['E'], None), 0],\n",
       " [(['M'], ['J'], ['B'], ['E'], None), 0],\n",
       " [(['M', 'J'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['M', 'J'], ['B', 'E'], [], None), 0],\n",
       " [(['J'], ['M'], ['B', 'E'], [], None), 0],\n",
       " [(['M'], ['J'], ['B', 'E'], [], None), 0],\n",
       " [(['M', 'J'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['A', 'M', 'J'], [], ['B', 'E'], None), 0],\n",
       " [(['J'], ['A', 'M'], [], ['B', 'E'], None), 0],\n",
       " [(['M'], ['A', 'J'], [], ['B', 'E'], None), 0],\n",
       " [(['M', 'J'], ['A'], [], ['B', 'E'], None), 0],\n",
       " [(['A'], ['M', 'J'], [], ['B', 'E'], None), 0],\n",
       " [(['A', 'J'], ['M'], [], ['B', 'E'], None), 0],\n",
       " [(['A', 'M'], ['J'], [], ['B', 'E'], None), 0],\n",
       " [(['A', 'M', 'J'], [], [], ['B', 'E'], None), 0],\n",
       " [([], ['A', 'M', 'J'], ['E'], ['B'], None), 0],\n",
       " [(['J'], ['A', 'M'], ['E'], ['B'], None), 0],\n",
       " [(['M'], ['A', 'J'], ['E'], ['B'], None), 0],\n",
       " [(['M', 'J'], ['A'], ['E'], ['B'], None), 0],\n",
       " [(['A'], ['M', 'J'], ['E'], ['B'], None), 0],\n",
       " [(['A', 'J'], ['M'], ['E'], ['B'], None), 0],\n",
       " [(['A', 'M'], ['J'], ['E'], ['B'], None), 0],\n",
       " [(['A', 'M', 'J'], [], ['E'], ['B'], None), 0],\n",
       " [([], ['A', 'M', 'J'], ['B'], ['E'], None), 0],\n",
       " [(['J'], ['A', 'M'], ['B'], ['E'], None), 0],\n",
       " [(['M'], ['A', 'J'], ['B'], ['E'], None), 0],\n",
       " [(['M', 'J'], ['A'], ['B'], ['E'], None), 0],\n",
       " [(['A'], ['M', 'J'], ['B'], ['E'], None), 0],\n",
       " [(['A', 'J'], ['M'], ['B'], ['E'], None), 0],\n",
       " [(['A', 'M'], ['J'], ['B'], ['E'], None), 0],\n",
       " [(['A', 'M', 'J'], [], ['B'], ['E'], None), 0],\n",
       " [([], ['A', 'M', 'J'], ['B', 'E'], [], None), 0],\n",
       " [(['J'], ['A', 'M'], ['B', 'E'], [], None), 0],\n",
       " [(['M'], ['A', 'J'], ['B', 'E'], [], None), 0],\n",
       " [(['M', 'J'], ['A'], ['B', 'E'], [], None), 0],\n",
       " [(['A'], ['M', 'J'], ['B', 'E'], [], None), 0],\n",
       " [(['A', 'J'], ['M'], ['B', 'E'], [], None), 0],\n",
       " [(['A', 'M'], ['J'], ['B', 'E'], [], None), 0],\n",
       " [(['A', 'M', 'J'], [], ['B', 'E'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'A'], None), 0],\n",
       " [(['M'], [], [], ['B', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['B'], None), 0],\n",
       " [(['M'], [], ['A'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['A'], None), 0],\n",
       " [(['M'], [], ['B'], ['A'], None), 0],\n",
       " [([], ['M'], ['B', 'A'], [], None), 0],\n",
       " [(['M'], [], ['B', 'A'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'A'], None), 0],\n",
       " [(['J'], [], [], ['B', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['B'], None), 0],\n",
       " [(['J'], [], ['A'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['A'], None), 0],\n",
       " [(['J'], [], ['B'], ['A'], None), 0],\n",
       " [([], ['J'], ['B', 'A'], [], None), 0],\n",
       " [(['J'], [], ['B', 'A'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['B', 'A'], None), 0],\n",
       " [(['J'], ['M'], [], ['B', 'A'], None), 0],\n",
       " [(['M'], ['J'], [], ['B', 'A'], None), 0],\n",
       " [(['M', 'J'], [], [], ['B', 'A'], None), 0],\n",
       " [([], ['M', 'J'], ['A'], ['B'], None), 0],\n",
       " [(['J'], ['M'], ['A'], ['B'], None), 0],\n",
       " [(['M'], ['J'], ['A'], ['B'], None), 0],\n",
       " [(['M', 'J'], [], ['A'], ['B'], None), 0],\n",
       " [([], ['M', 'J'], ['B'], ['A'], None), 0],\n",
       " [(['J'], ['M'], ['B'], ['A'], None), 0],\n",
       " [(['M'], ['J'], ['B'], ['A'], None), 0],\n",
       " [(['M', 'J'], [], ['B'], ['A'], None), 0],\n",
       " [([], ['M', 'J'], ['B', 'A'], [], None), 0],\n",
       " [(['J'], ['M'], ['B', 'A'], [], None), 0],\n",
       " [(['M'], ['J'], ['B', 'A'], [], None), 0],\n",
       " [(['M', 'J'], [], ['B', 'A'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'M'], None), 0],\n",
       " [(['A'], [], [], ['B', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['B'], None), 0],\n",
       " [(['A'], [], ['M'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['M'], None), 0],\n",
       " [(['A'], [], ['B'], ['M'], None), 0],\n",
       " [([], ['A'], ['B', 'M'], [], None), 0],\n",
       " [(['A'], [], ['B', 'M'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'M'], ('E',)), 0.001440701707124981],\n",
       " [(['J'], [], [], ['B', 'M'], ('A',)), 0.008547192182370435],\n",
       " [([], ['J'], ['M'], ['B'], ('E',)), 0.005637035563347659],\n",
       " [(['J'], [], ['M'], ['B'], ('A',)), 0.6657108744026019],\n",
       " [([], ['J'], ['B'], ['M'], ('E',)), 0.0017852122026965744],\n",
       " [(['J'], [], ['B'], ['M'], ('A',)), 0.9884372823600297],\n",
       " [([], ['J'], ['B', 'M'], [], ('E',)), 0.0020181943345616267],\n",
       " [(['J'], [], ['B', 'M'], [], ('A',)), 0.99994936196669],\n",
       " [([], ['A', 'J'], [], ['B', 'M'], None), 0],\n",
       " [(['J'], ['A'], [], ['B', 'M'], None), 0],\n",
       " [(['A'], ['J'], [], ['B', 'M'], None), 0],\n",
       " [(['A', 'J'], [], [], ['B', 'M'], None), 0],\n",
       " [([], ['A', 'J'], ['M'], ['B'], None), 0],\n",
       " [(['J'], ['A'], ['M'], ['B'], None), 0],\n",
       " [(['A'], ['J'], ['M'], ['B'], None), 0],\n",
       " [(['A', 'J'], [], ['M'], ['B'], None), 0],\n",
       " [([], ['A', 'J'], ['B'], ['M'], None), 0],\n",
       " [(['J'], ['A'], ['B'], ['M'], None), 0],\n",
       " [(['A'], ['J'], ['B'], ['M'], None), 0],\n",
       " [(['A', 'J'], [], ['B'], ['M'], None), 0],\n",
       " [([], ['A', 'J'], ['B', 'M'], [], None), 0],\n",
       " [(['J'], ['A'], ['B', 'M'], [], None), 0],\n",
       " [(['A'], ['J'], ['B', 'M'], [], None), 0],\n",
       " [(['A', 'J'], [], ['B', 'M'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'J'], None), 0],\n",
       " [(['A'], [], [], ['B', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['B'], None), 0],\n",
       " [(['A'], [], ['J'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['J'], None), 0],\n",
       " [(['A'], [], ['B'], ['J'], None), 0],\n",
       " [([], ['A'], ['B', 'J'], [], None), 0],\n",
       " [(['A'], [], ['B', 'J'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'J'], ('E',)), 0.001440701707124981],\n",
       " [(['M'], [], [], ['B', 'J'], ('A',)), 0.011511683274678913],\n",
       " [([], ['M'], ['J'], ['B'], ('E',)), 0.004551641633239515],\n",
       " [(['M'], [], ['J'], ['B'], ('A',)), 0.6657108744026019],\n",
       " [([], ['M'], ['B'], ['J'], ('E',)), 0.0017852122026965744],\n",
       " [(['M'], [], ['B'], ['J'], ('A',)), 0.9914148016888648],\n",
       " [([], ['M'], ['B', 'J'], [], ('E',)), 0.0020171402594207116],\n",
       " [(['M'], [], ['B', 'J'], [], ('A',)), 0.99994936196669],\n",
       " [([], ['A', 'M'], [], ['B', 'J'], None), 0],\n",
       " [(['M'], ['A'], [], ['B', 'J'], None), 0],\n",
       " [(['A'], ['M'], [], ['B', 'J'], None), 0],\n",
       " [(['A', 'M'], [], [], ['B', 'J'], None), 0],\n",
       " [([], ['A', 'M'], ['J'], ['B'], None), 0],\n",
       " [(['M'], ['A'], ['J'], ['B'], None), 0],\n",
       " [(['A'], ['M'], ['J'], ['B'], None), 0],\n",
       " [(['A', 'M'], [], ['J'], ['B'], None), 0],\n",
       " [([], ['A', 'M'], ['B'], ['J'], None), 0],\n",
       " [(['M'], ['A'], ['B'], ['J'], None), 0],\n",
       " [(['A'], ['M'], ['B'], ['J'], None), 0],\n",
       " [(['A', 'M'], [], ['B'], ['J'], None), 0],\n",
       " [([], ['A', 'M'], ['B', 'J'], [], None), 0],\n",
       " [(['M'], ['A'], ['B', 'J'], [], None), 0],\n",
       " [(['A'], ['M'], ['B', 'J'], [], None), 0],\n",
       " [(['A', 'M'], [], ['B', 'J'], [], None), 0],\n",
       " [([], ['M'], [], ['E', 'A'], None), 0],\n",
       " [(['M'], [], [], ['E', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['E'], None), 0],\n",
       " [(['M'], [], ['A'], ['E'], None), 0],\n",
       " [([], ['M'], ['E'], ['A'], None), 0],\n",
       " [(['M'], [], ['E'], ['A'], None), 0],\n",
       " [([], ['M'], ['E', 'A'], [], None), 0],\n",
       " [(['M'], [], ['E', 'A'], [], None), 0],\n",
       " [([], ['J'], [], ['E', 'A'], None), 0],\n",
       " [(['J'], [], [], ['E', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['E'], None), 0],\n",
       " [(['J'], [], ['A'], ['E'], None), 0],\n",
       " [([], ['J'], ['E'], ['A'], None), 0],\n",
       " [(['J'], [], ['E'], ['A'], None), 0],\n",
       " [([], ['J'], ['E', 'A'], [], None), 0],\n",
       " [(['J'], [], ['E', 'A'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['E', 'A'], None), 0],\n",
       " [(['J'], ['M'], [], ['E', 'A'], None), 0],\n",
       " [(['M'], ['J'], [], ['E', 'A'], None), 0],\n",
       " [(['M', 'J'], [], [], ['E', 'A'], None), 0],\n",
       " [([], ['M', 'J'], ['A'], ['E'], None), 0],\n",
       " [(['J'], ['M'], ['A'], ['E'], None), 0],\n",
       " [(['M'], ['J'], ['A'], ['E'], None), 0],\n",
       " [(['M', 'J'], [], ['A'], ['E'], None), 0],\n",
       " [([], ['M', 'J'], ['E'], ['A'], None), 0],\n",
       " [(['J'], ['M'], ['E'], ['A'], None), 0],\n",
       " [(['M'], ['J'], ['E'], ['A'], None), 0],\n",
       " [(['M', 'J'], [], ['E'], ['A'], None), 0],\n",
       " [([], ['M', 'J'], ['E', 'A'], [], None), 0],\n",
       " [(['J'], ['M'], ['E', 'A'], [], None), 0],\n",
       " [(['M'], ['J'], ['E', 'A'], [], None), 0],\n",
       " [(['M', 'J'], [], ['E', 'A'], [], None), 0],\n",
       " [([], ['A'], [], ['E', 'M'], None), 0],\n",
       " [(['A'], [], [], ['E', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['E'], None), 0],\n",
       " [(['A'], [], ['M'], ['E'], None), 0],\n",
       " [([], ['A'], ['E'], ['M'], None), 0],\n",
       " [(['A'], [], ['E'], ['M'], None), 0],\n",
       " [([], ['A'], ['E', 'M'], [], None), 0],\n",
       " [(['A'], [], ['E', 'M'], [], None), 0],\n",
       " [([], ['J'], [], ['E', 'M'], ('B',)), 9.015328226329397e-05],\n",
       " [(['J'], [], [], ['E', 'M'], ('B', 'A')), 0.005083365847947335],\n",
       " [([], ['J'], ['M'], ['E'], ('B',)), 0.006901098482896905],\n",
       " [(['J'], [], ['M'], ['E'], ('A',)), 0.7099672469001376],\n",
       " [([], ['J'], ['E'], ['M'], ('B',)), 0.00011174749602380887],\n",
       " [(['J'], [], ['E'], ['M'], ('B', 'A')), 0.002258110777368501],\n",
       " [([], ['J'], ['E', 'M'], [], ('B',)), 0.0024727772971907254],\n",
       " [(['J'], [], ['E', 'M'], [], ('A',)), 0.9980668823707707],\n",
       " [([], ['A', 'J'], [], ['E', 'M'], None), 0],\n",
       " [(['J'], ['A'], [], ['E', 'M'], None), 0],\n",
       " [(['A'], ['J'], [], ['E', 'M'], None), 0],\n",
       " [(['A', 'J'], [], [], ['E', 'M'], None), 0],\n",
       " [([], ['A', 'J'], ['M'], ['E'], None), 0],\n",
       " [(['J'], ['A'], ['M'], ['E'], None), 0],\n",
       " [(['A'], ['J'], ['M'], ['E'], None), 0],\n",
       " [(['A', 'J'], [], ['M'], ['E'], None), 0],\n",
       " [([], ['A', 'J'], ['E'], ['M'], None), 0],\n",
       " [(['J'], ['A'], ['E'], ['M'], None), 0],\n",
       " [(['A'], ['J'], ['E'], ['M'], None), 0],\n",
       " [(['A', 'J'], [], ['E'], ['M'], None), 0],\n",
       " [([], ['A', 'J'], ['E', 'M'], [], None), 0],\n",
       " [(['J'], ['A'], ['E', 'M'], [], None), 0],\n",
       " [(['A'], ['J'], ['E', 'M'], [], None), 0],\n",
       " [(['A', 'J'], [], ['E', 'M'], [], None), 0],\n",
       " [([], ['A'], [], ['E', 'J'], None), 0],\n",
       " [(['A'], [], [], ['E', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['E'], None), 0],\n",
       " [(['A'], [], ['J'], ['E'], None), 0],\n",
       " [([], ['A'], ['E'], ['J'], None), 0],\n",
       " [(['A'], [], ['E'], ['J'], None), 0],\n",
       " [([], ['A'], ['E', 'J'], [], None), 0],\n",
       " [(['A'], [], ['E', 'J'], [], None), 0],\n",
       " [([], ['M'], [], ['E', 'J'], ('B',)), 9.015328226329397e-05],\n",
       " [(['M'], [], [], ['E', 'J'], ('A',)), 0.014113094727792633],\n",
       " [([], ['M'], ['J'], ['E'], ('B',)), 0.005142852044040336],\n",
       " [(['M'], [], ['J'], ['E'], ('A',)), 0.7099672469001376],\n",
       " [([], ['M'], ['E'], ['J'], ('B',)), 0.00011174749602380887],\n",
       " [(['M'], [], ['E'], ['J'], ('A',)), 0.7512000215618058],\n",
       " [([], ['M'], ['E', 'J'], [], ('B',)), 0.00227989956557118],\n",
       " [(['M'], [], ['E', 'J'], [], ('A',)), 0.9980668823707707],\n",
       " [([], ['A', 'M'], [], ['E', 'J'], None), 0],\n",
       " [(['M'], ['A'], [], ['E', 'J'], None), 0],\n",
       " [(['A'], ['M'], [], ['E', 'J'], None), 0],\n",
       " [(['A', 'M'], [], [], ['E', 'J'], None), 0],\n",
       " [([], ['A', 'M'], ['J'], ['E'], None), 0],\n",
       " [(['M'], ['A'], ['J'], ['E'], None), 0],\n",
       " [(['A'], ['M'], ['J'], ['E'], None), 0],\n",
       " [(['A', 'M'], [], ['J'], ['E'], None), 0],\n",
       " [([], ['A', 'M'], ['E'], ['J'], None), 0],\n",
       " [(['M'], ['A'], ['E'], ['J'], None), 0],\n",
       " [(['A'], ['M'], ['E'], ['J'], None), 0],\n",
       " [(['A', 'M'], [], ['E'], ['J'], None), 0],\n",
       " [([], ['A', 'M'], ['E', 'J'], [], None), 0],\n",
       " [(['M'], ['A'], ['E', 'J'], [], None), 0],\n",
       " [(['A'], ['M'], ['E', 'J'], [], None), 0],\n",
       " [(['A', 'M'], [], ['E', 'J'], [], None), 0],\n",
       " [([], ['J'], [], ['A', 'M'], ('B',)), 6.013131697154257e-05],\n",
       " [(['J'], [], [], ['A', 'M'], ('B',)), 6.013131697154255e-05],\n",
       " [([], ['J'], ['M'], ['A'], ('E',)), 0.00142225903236392],\n",
       " [(['J'], [], ['M'], ['A'], ('B',)), 6.0131316971542544e-05],\n",
       " [([], ['J'], ['A'], ['M'], ('B',)), 0.373551228281836],\n",
       " [(['J'], [], ['A'], ['M'], ('B',)), 0.37355122828183596],\n",
       " [([], ['J'], ['A', 'M'], [], ('B',)), 0.3735512282818359],\n",
       " [(['J'], [], ['A', 'M'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['M'], [], ['A', 'J'], ('B',)), 6.013131697154257e-05],\n",
       " [(['M'], [], [], ['A', 'J'], ('B',)), 6.013131697154256e-05],\n",
       " [([], ['M'], ['J'], ['A'], ('B',)), 6.013131697154255e-05],\n",
       " [(['M'], [], ['J'], ['A'], ('E',)), 0.0014222590323639196],\n",
       " [([], ['M'], ['A'], ['J'], ('B',)), 0.373551228281836],\n",
       " [(['M'], [], ['A'], ['J'], ('B',)), 0.3735512282818359],\n",
       " [([], ['M'], ['A', 'J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['M'], [], ['A', 'J'], [], ('E',)), 0.23100870196889098],\n",
       " [([], ['A'], [], ['M', 'J'], ('E',)), 0.0014222590323639203],\n",
       " [(['A'], [], [], ['M', 'J'], ('B',)), 0.373551228281836],\n",
       " [([], ['A'], ['J'], ['M'], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], [], ['J'], ['M'], ('B',)), 0.37355122828183596],\n",
       " [([], ['A'], ['M'], ['J'], ('E',)), 0.00142225903236392],\n",
       " [(['A'], [], ['M'], ['J'], ('B',)), 0.3735512282818359],\n",
       " [([], ['A'], ['M', 'J'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['A'], [], ['M', 'J'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['M'], [], ['B', 'E', 'A'], None), 0],\n",
       " [(['M'], [], [], ['B', 'E', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['B', 'E'], None), 0],\n",
       " [(['M'], [], ['A'], ['B', 'E'], None), 0],\n",
       " [([], ['M'], ['E'], ['B', 'A'], None), 0],\n",
       " [(['M'], [], ['E'], ['B', 'A'], None), 0],\n",
       " [([], ['M'], ['E', 'A'], ['B'], None), 0],\n",
       " [(['M'], [], ['E', 'A'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['E', 'A'], None), 0],\n",
       " [(['M'], [], ['B'], ['E', 'A'], None), 0],\n",
       " [([], ['M'], ['B', 'A'], ['E'], None), 0],\n",
       " [(['M'], [], ['B', 'A'], ['E'], None), 0],\n",
       " [([], ['M'], ['B', 'E'], ['A'], None), 0],\n",
       " [(['M'], [], ['B', 'E'], ['A'], None), 0],\n",
       " [([], ['M'], ['B', 'E', 'A'], [], None), 0],\n",
       " [(['M'], [], ['B', 'E', 'A'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'E', 'A'], None), 0],\n",
       " [(['J'], [], [], ['B', 'E', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['B', 'E'], None), 0],\n",
       " [(['J'], [], ['A'], ['B', 'E'], None), 0],\n",
       " [([], ['J'], ['E'], ['B', 'A'], None), 0],\n",
       " [(['J'], [], ['E'], ['B', 'A'], None), 0],\n",
       " [([], ['J'], ['E', 'A'], ['B'], None), 0],\n",
       " [(['J'], [], ['E', 'A'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['E', 'A'], None), 0],\n",
       " [(['J'], [], ['B'], ['E', 'A'], None), 0],\n",
       " [([], ['J'], ['B', 'A'], ['E'], None), 0],\n",
       " [(['J'], [], ['B', 'A'], ['E'], None), 0],\n",
       " [([], ['J'], ['B', 'E'], ['A'], None), 0],\n",
       " [(['J'], [], ['B', 'E'], ['A'], None), 0],\n",
       " [([], ['J'], ['B', 'E', 'A'], [], None), 0],\n",
       " [(['J'], [], ['B', 'E', 'A'], [], None), 0],\n",
       " [([], ['M', 'J'], [], ['B', 'E', 'A'], None), 0],\n",
       " [(['J'], ['M'], [], ['B', 'E', 'A'], None), 0],\n",
       " [(['M'], ['J'], [], ['B', 'E', 'A'], None), 0],\n",
       " [(['M', 'J'], [], [], ['B', 'E', 'A'], None), 0],\n",
       " [([], ['M', 'J'], ['A'], ['B', 'E'], None), 0],\n",
       " [(['J'], ['M'], ['A'], ['B', 'E'], None), 0],\n",
       " [(['M'], ['J'], ['A'], ['B', 'E'], None), 0],\n",
       " [(['M', 'J'], [], ['A'], ['B', 'E'], None), 0],\n",
       " [([], ['M', 'J'], ['E'], ['B', 'A'], None), 0],\n",
       " [(['J'], ['M'], ['E'], ['B', 'A'], None), 0],\n",
       " [(['M'], ['J'], ['E'], ['B', 'A'], None), 0],\n",
       " [(['M', 'J'], [], ['E'], ['B', 'A'], None), 0],\n",
       " [([], ['M', 'J'], ['E', 'A'], ['B'], None), 0],\n",
       " [(['J'], ['M'], ['E', 'A'], ['B'], None), 0],\n",
       " [(['M'], ['J'], ['E', 'A'], ['B'], None), 0],\n",
       " [(['M', 'J'], [], ['E', 'A'], ['B'], None), 0],\n",
       " [([], ['M', 'J'], ['B'], ['E', 'A'], None), 0],\n",
       " [(['J'], ['M'], ['B'], ['E', 'A'], None), 0],\n",
       " [(['M'], ['J'], ['B'], ['E', 'A'], None), 0],\n",
       " [(['M', 'J'], [], ['B'], ['E', 'A'], None), 0],\n",
       " [([], ['M', 'J'], ['B', 'A'], ['E'], None), 0],\n",
       " [(['J'], ['M'], ['B', 'A'], ['E'], None), 0],\n",
       " [(['M'], ['J'], ['B', 'A'], ['E'], None), 0],\n",
       " [(['M', 'J'], [], ['B', 'A'], ['E'], None), 0],\n",
       " [([], ['M', 'J'], ['B', 'E'], ['A'], None), 0],\n",
       " [(['J'], ['M'], ['B', 'E'], ['A'], None), 0],\n",
       " [(['M'], ['J'], ['B', 'E'], ['A'], None), 0],\n",
       " [(['M', 'J'], [], ['B', 'E'], ['A'], None), 0],\n",
       " [([], ['M', 'J'], ['B', 'E', 'A'], [], None), 0],\n",
       " [(['J'], ['M'], ['B', 'E', 'A'], [], None), 0],\n",
       " [(['M'], ['J'], ['B', 'E', 'A'], [], None), 0],\n",
       " [(['M', 'J'], [], ['B', 'E', 'A'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'E', 'M'], None), 0],\n",
       " [(['A'], [], [], ['B', 'E', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['B', 'E'], None), 0],\n",
       " [(['A'], [], ['M'], ['B', 'E'], None), 0],\n",
       " [([], ['A'], ['E'], ['B', 'M'], None), 0],\n",
       " [(['A'], [], ['E'], ['B', 'M'], None), 0],\n",
       " [([], ['A'], ['E', 'M'], ['B'], None), 0],\n",
       " [(['A'], [], ['E', 'M'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['E', 'M'], None), 0],\n",
       " [(['A'], [], ['B'], ['E', 'M'], None), 0],\n",
       " [([], ['A'], ['B', 'M'], ['E'], None), 0],\n",
       " [(['A'], [], ['B', 'M'], ['E'], None), 0],\n",
       " [([], ['A'], ['B', 'E'], ['M'], None), 0],\n",
       " [(['A'], [], ['B', 'E'], ['M'], None), 0],\n",
       " [([], ['A'], ['B', 'E', 'M'], [], None), 0],\n",
       " [(['A'], [], ['B', 'E', 'M'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'E', 'M'], None), 0],\n",
       " [(['J'], [], [], ['B', 'E', 'M'], None), 0],\n",
       " [([], ['J'], ['M'], ['B', 'E'], None), 0],\n",
       " [(['J'], [], ['M'], ['B', 'E'], None), 0],\n",
       " [([], ['J'], ['E'], ['B', 'M'], None), 0],\n",
       " [(['J'], [], ['E'], ['B', 'M'], None), 0],\n",
       " [([], ['J'], ['E', 'M'], ['B'], None), 0],\n",
       " [(['J'], [], ['E', 'M'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['E', 'M'], None), 0],\n",
       " [(['J'], [], ['B'], ['E', 'M'], None), 0],\n",
       " [([], ['J'], ['B', 'M'], ['E'], None), 0],\n",
       " [(['J'], [], ['B', 'M'], ['E'], None), 0],\n",
       " [([], ['J'], ['B', 'E'], ['M'], None), 0],\n",
       " [(['J'], [], ['B', 'E'], ['M'], None), 0],\n",
       " [([], ['J'], ['B', 'E', 'M'], [], None), 0],\n",
       " [(['J'], [], ['B', 'E', 'M'], [], None), 0],\n",
       " [([], ['A', 'J'], [], ['B', 'E', 'M'], None), 0],\n",
       " [(['J'], ['A'], [], ['B', 'E', 'M'], None), 0],\n",
       " [(['A'], ['J'], [], ['B', 'E', 'M'], None), 0],\n",
       " [(['A', 'J'], [], [], ['B', 'E', 'M'], None), 0],\n",
       " [([], ['A', 'J'], ['M'], ['B', 'E'], None), 0],\n",
       " [(['J'], ['A'], ['M'], ['B', 'E'], None), 0],\n",
       " [(['A'], ['J'], ['M'], ['B', 'E'], None), 0],\n",
       " [(['A', 'J'], [], ['M'], ['B', 'E'], None), 0],\n",
       " [([], ['A', 'J'], ['E'], ['B', 'M'], None), 0],\n",
       " [(['J'], ['A'], ['E'], ['B', 'M'], None), 0],\n",
       " [(['A'], ['J'], ['E'], ['B', 'M'], None), 0],\n",
       " [(['A', 'J'], [], ['E'], ['B', 'M'], None), 0],\n",
       " [([], ['A', 'J'], ['E', 'M'], ['B'], None), 0],\n",
       " [(['J'], ['A'], ['E', 'M'], ['B'], None), 0],\n",
       " [(['A'], ['J'], ['E', 'M'], ['B'], None), 0],\n",
       " [(['A', 'J'], [], ['E', 'M'], ['B'], None), 0],\n",
       " [([], ['A', 'J'], ['B'], ['E', 'M'], None), 0],\n",
       " [(['J'], ['A'], ['B'], ['E', 'M'], None), 0],\n",
       " [(['A'], ['J'], ['B'], ['E', 'M'], None), 0],\n",
       " [(['A', 'J'], [], ['B'], ['E', 'M'], None), 0],\n",
       " [([], ['A', 'J'], ['B', 'M'], ['E'], None), 0],\n",
       " [(['J'], ['A'], ['B', 'M'], ['E'], None), 0],\n",
       " [(['A'], ['J'], ['B', 'M'], ['E'], None), 0],\n",
       " [(['A', 'J'], [], ['B', 'M'], ['E'], None), 0],\n",
       " [([], ['A', 'J'], ['B', 'E'], ['M'], None), 0],\n",
       " [(['J'], ['A'], ['B', 'E'], ['M'], None), 0],\n",
       " [(['A'], ['J'], ['B', 'E'], ['M'], None), 0],\n",
       " [(['A', 'J'], [], ['B', 'E'], ['M'], None), 0],\n",
       " [([], ['A', 'J'], ['B', 'E', 'M'], [], None), 0],\n",
       " [(['J'], ['A'], ['B', 'E', 'M'], [], None), 0],\n",
       " [(['A'], ['J'], ['B', 'E', 'M'], [], None), 0],\n",
       " [(['A', 'J'], [], ['B', 'E', 'M'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'E', 'J'], None), 0],\n",
       " [(['A'], [], [], ['B', 'E', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['B', 'E'], None), 0],\n",
       " [(['A'], [], ['J'], ['B', 'E'], None), 0],\n",
       " [([], ['A'], ['E'], ['B', 'J'], None), 0],\n",
       " [(['A'], [], ['E'], ['B', 'J'], None), 0],\n",
       " [([], ['A'], ['E', 'J'], ['B'], None), 0],\n",
       " [(['A'], [], ['E', 'J'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['E', 'J'], None), 0],\n",
       " [(['A'], [], ['B'], ['E', 'J'], None), 0],\n",
       " [([], ['A'], ['B', 'J'], ['E'], None), 0],\n",
       " [(['A'], [], ['B', 'J'], ['E'], None), 0],\n",
       " [([], ['A'], ['B', 'E'], ['J'], None), 0],\n",
       " [(['A'], [], ['B', 'E'], ['J'], None), 0],\n",
       " [([], ['A'], ['B', 'E', 'J'], [], None), 0],\n",
       " [(['A'], [], ['B', 'E', 'J'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'E', 'J'], None), 0],\n",
       " [(['M'], [], [], ['B', 'E', 'J'], None), 0],\n",
       " [([], ['M'], ['J'], ['B', 'E'], None), 0],\n",
       " [(['M'], [], ['J'], ['B', 'E'], None), 0],\n",
       " [([], ['M'], ['E'], ['B', 'J'], None), 0],\n",
       " [(['M'], [], ['E'], ['B', 'J'], None), 0],\n",
       " [([], ['M'], ['E', 'J'], ['B'], None), 0],\n",
       " [(['M'], [], ['E', 'J'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['E', 'J'], None), 0],\n",
       " [(['M'], [], ['B'], ['E', 'J'], None), 0],\n",
       " [([], ['M'], ['B', 'J'], ['E'], None), 0],\n",
       " [(['M'], [], ['B', 'J'], ['E'], None), 0],\n",
       " [([], ['M'], ['B', 'E'], ['J'], None), 0],\n",
       " [(['M'], [], ['B', 'E'], ['J'], None), 0],\n",
       " [([], ['M'], ['B', 'E', 'J'], [], None), 0],\n",
       " [(['M'], [], ['B', 'E', 'J'], [], None), 0],\n",
       " [([], ['A', 'M'], [], ['B', 'E', 'J'], None), 0],\n",
       " [(['M'], ['A'], [], ['B', 'E', 'J'], None), 0],\n",
       " [(['A'], ['M'], [], ['B', 'E', 'J'], None), 0],\n",
       " [(['A', 'M'], [], [], ['B', 'E', 'J'], None), 0],\n",
       " [([], ['A', 'M'], ['J'], ['B', 'E'], None), 0],\n",
       " [(['M'], ['A'], ['J'], ['B', 'E'], None), 0],\n",
       " [(['A'], ['M'], ['J'], ['B', 'E'], None), 0],\n",
       " [(['A', 'M'], [], ['J'], ['B', 'E'], None), 0],\n",
       " [([], ['A', 'M'], ['E'], ['B', 'J'], None), 0],\n",
       " [(['M'], ['A'], ['E'], ['B', 'J'], None), 0],\n",
       " [(['A'], ['M'], ['E'], ['B', 'J'], None), 0],\n",
       " [(['A', 'M'], [], ['E'], ['B', 'J'], None), 0],\n",
       " [([], ['A', 'M'], ['E', 'J'], ['B'], None), 0],\n",
       " [(['M'], ['A'], ['E', 'J'], ['B'], None), 0],\n",
       " [(['A'], ['M'], ['E', 'J'], ['B'], None), 0],\n",
       " [(['A', 'M'], [], ['E', 'J'], ['B'], None), 0],\n",
       " [([], ['A', 'M'], ['B'], ['E', 'J'], None), 0],\n",
       " [(['M'], ['A'], ['B'], ['E', 'J'], None), 0],\n",
       " [(['A'], ['M'], ['B'], ['E', 'J'], None), 0],\n",
       " [(['A', 'M'], [], ['B'], ['E', 'J'], None), 0],\n",
       " [([], ['A', 'M'], ['B', 'J'], ['E'], None), 0],\n",
       " [(['M'], ['A'], ['B', 'J'], ['E'], None), 0],\n",
       " [(['A'], ['M'], ['B', 'J'], ['E'], None), 0],\n",
       " [(['A', 'M'], [], ['B', 'J'], ['E'], None), 0],\n",
       " [([], ['A', 'M'], ['B', 'E'], ['J'], None), 0],\n",
       " [(['M'], ['A'], ['B', 'E'], ['J'], None), 0],\n",
       " [(['A'], ['M'], ['B', 'E'], ['J'], None), 0],\n",
       " [(['A', 'M'], [], ['B', 'E'], ['J'], None), 0],\n",
       " [([], ['A', 'M'], ['B', 'E', 'J'], [], None), 0],\n",
       " [(['M'], ['A'], ['B', 'E', 'J'], [], None), 0],\n",
       " [(['A'], ['M'], ['B', 'E', 'J'], [], None), 0],\n",
       " [(['A', 'M'], [], ['B', 'E', 'J'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'A', 'M'], None), 0],\n",
       " [(['J'], [], [], ['B', 'A', 'M'], None), 0],\n",
       " [([], ['J'], ['M'], ['B', 'A'], None), 0],\n",
       " [(['J'], [], ['M'], ['B', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['B', 'M'], None), 0],\n",
       " [(['J'], [], ['A'], ['B', 'M'], None), 0],\n",
       " [([], ['J'], ['A', 'M'], ['B'], None), 0],\n",
       " [(['J'], [], ['A', 'M'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['A', 'M'], None), 0],\n",
       " [(['J'], [], ['B'], ['A', 'M'], None), 0],\n",
       " [([], ['J'], ['B', 'M'], ['A'], None), 0],\n",
       " [(['J'], [], ['B', 'M'], ['A'], None), 0],\n",
       " [([], ['J'], ['B', 'A'], ['M'], None), 0],\n",
       " [(['J'], [], ['B', 'A'], ['M'], None), 0],\n",
       " [([], ['J'], ['B', 'A', 'M'], [], None), 0],\n",
       " [(['J'], [], ['B', 'A', 'M'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'A', 'J'], None), 0],\n",
       " [(['M'], [], [], ['B', 'A', 'J'], None), 0],\n",
       " [([], ['M'], ['J'], ['B', 'A'], None), 0],\n",
       " [(['M'], [], ['J'], ['B', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['B', 'J'], None), 0],\n",
       " [(['M'], [], ['A'], ['B', 'J'], None), 0],\n",
       " [([], ['M'], ['A', 'J'], ['B'], None), 0],\n",
       " [(['M'], [], ['A', 'J'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['A', 'J'], None), 0],\n",
       " [(['M'], [], ['B'], ['A', 'J'], None), 0],\n",
       " [([], ['M'], ['B', 'J'], ['A'], None), 0],\n",
       " [(['M'], [], ['B', 'J'], ['A'], None), 0],\n",
       " [([], ['M'], ['B', 'A'], ['J'], None), 0],\n",
       " [(['M'], [], ['B', 'A'], ['J'], None), 0],\n",
       " [([], ['M'], ['B', 'A', 'J'], [], None), 0],\n",
       " [(['M'], [], ['B', 'A', 'J'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'M', 'J'], None), 0],\n",
       " [(['A'], [], [], ['B', 'M', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['B', 'M'], None), 0],\n",
       " [(['A'], [], ['J'], ['B', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['B', 'J'], None), 0],\n",
       " [(['A'], [], ['M'], ['B', 'J'], None), 0],\n",
       " [([], ['A'], ['M', 'J'], ['B'], None), 0],\n",
       " [(['A'], [], ['M', 'J'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['M', 'J'], None), 0],\n",
       " [(['A'], [], ['B'], ['M', 'J'], None), 0],\n",
       " [([], ['A'], ['B', 'J'], ['M'], None), 0],\n",
       " [(['A'], [], ['B', 'J'], ['M'], None), 0],\n",
       " [([], ['A'], ['B', 'M'], ['J'], None), 0],\n",
       " [(['A'], [], ['B', 'M'], ['J'], None), 0],\n",
       " [([], ['A'], ['B', 'M', 'J'], [], None), 0],\n",
       " [(['A'], [], ['B', 'M', 'J'], [], None), 0],\n",
       " [([], ['J'], [], ['E', 'A', 'M'], None), 0],\n",
       " [(['J'], [], [], ['E', 'A', 'M'], None), 0],\n",
       " [([], ['J'], ['M'], ['E', 'A'], None), 0],\n",
       " [(['J'], [], ['M'], ['E', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['E', 'M'], None), 0],\n",
       " [(['J'], [], ['A'], ['E', 'M'], None), 0],\n",
       " [([], ['J'], ['A', 'M'], ['E'], None), 0],\n",
       " [(['J'], [], ['A', 'M'], ['E'], None), 0],\n",
       " [([], ['J'], ['E'], ['A', 'M'], None), 0],\n",
       " [(['J'], [], ['E'], ['A', 'M'], None), 0],\n",
       " [([], ['J'], ['E', 'M'], ['A'], None), 0],\n",
       " [(['J'], [], ['E', 'M'], ['A'], None), 0],\n",
       " [([], ['J'], ['E', 'A'], ['M'], None), 0],\n",
       " [(['J'], [], ['E', 'A'], ['M'], None), 0],\n",
       " [([], ['J'], ['E', 'A', 'M'], [], None), 0],\n",
       " [(['J'], [], ['E', 'A', 'M'], [], None), 0],\n",
       " [([], ['M'], [], ['E', 'A', 'J'], None), 0],\n",
       " [(['M'], [], [], ['E', 'A', 'J'], None), 0],\n",
       " [([], ['M'], ['J'], ['E', 'A'], None), 0],\n",
       " [(['M'], [], ['J'], ['E', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['E', 'J'], None), 0],\n",
       " [(['M'], [], ['A'], ['E', 'J'], None), 0],\n",
       " [([], ['M'], ['A', 'J'], ['E'], None), 0],\n",
       " [(['M'], [], ['A', 'J'], ['E'], None), 0],\n",
       " [([], ['M'], ['E'], ['A', 'J'], None), 0],\n",
       " [(['M'], [], ['E'], ['A', 'J'], None), 0],\n",
       " [([], ['M'], ['E', 'J'], ['A'], None), 0],\n",
       " [(['M'], [], ['E', 'J'], ['A'], None), 0],\n",
       " [([], ['M'], ['E', 'A'], ['J'], None), 0],\n",
       " [(['M'], [], ['E', 'A'], ['J'], None), 0],\n",
       " [([], ['M'], ['E', 'A', 'J'], [], None), 0],\n",
       " [(['M'], [], ['E', 'A', 'J'], [], None), 0],\n",
       " [([], ['A'], [], ['E', 'M', 'J'], None), 0],\n",
       " [(['A'], [], [], ['E', 'M', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['E', 'M'], None), 0],\n",
       " [(['A'], [], ['J'], ['E', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['E', 'J'], None), 0],\n",
       " [(['A'], [], ['M'], ['E', 'J'], None), 0],\n",
       " [([], ['A'], ['M', 'J'], ['E'], None), 0],\n",
       " [(['A'], [], ['M', 'J'], ['E'], None), 0],\n",
       " [([], ['A'], ['E'], ['M', 'J'], None), 0],\n",
       " [(['A'], [], ['E'], ['M', 'J'], None), 0],\n",
       " [([], ['A'], ['E', 'J'], ['M'], None), 0],\n",
       " [(['A'], [], ['E', 'J'], ['M'], None), 0],\n",
       " [([], ['A'], ['E', 'M'], ['J'], None), 0],\n",
       " [(['A'], [], ['E', 'M'], ['J'], None), 0],\n",
       " [([], ['A'], ['E', 'M', 'J'], [], None), 0],\n",
       " [(['A'], [], ['E', 'M', 'J'], [], None), 0],\n",
       " [([], ['J'], [], ['B', 'E', 'A', 'M'], None), 0],\n",
       " [(['J'], [], [], ['B', 'E', 'A', 'M'], None), 0],\n",
       " [([], ['J'], ['M'], ['B', 'E', 'A'], None), 0],\n",
       " [(['J'], [], ['M'], ['B', 'E', 'A'], None), 0],\n",
       " [([], ['J'], ['A'], ['B', 'E', 'M'], None), 0],\n",
       " [(['J'], [], ['A'], ['B', 'E', 'M'], None), 0],\n",
       " [([], ['J'], ['A', 'M'], ['B', 'E'], None), 0],\n",
       " [(['J'], [], ['A', 'M'], ['B', 'E'], None), 0],\n",
       " [([], ['J'], ['E'], ['B', 'A', 'M'], None), 0],\n",
       " [(['J'], [], ['E'], ['B', 'A', 'M'], None), 0],\n",
       " [([], ['J'], ['E', 'M'], ['B', 'A'], None), 0],\n",
       " [(['J'], [], ['E', 'M'], ['B', 'A'], None), 0],\n",
       " [([], ['J'], ['E', 'A'], ['B', 'M'], None), 0],\n",
       " [(['J'], [], ['E', 'A'], ['B', 'M'], None), 0],\n",
       " [([], ['J'], ['E', 'A', 'M'], ['B'], None), 0],\n",
       " [(['J'], [], ['E', 'A', 'M'], ['B'], None), 0],\n",
       " [([], ['J'], ['B'], ['E', 'A', 'M'], None), 0],\n",
       " [(['J'], [], ['B'], ['E', 'A', 'M'], None), 0],\n",
       " [([], ['J'], ['B', 'M'], ['E', 'A'], None), 0],\n",
       " [(['J'], [], ['B', 'M'], ['E', 'A'], None), 0],\n",
       " [([], ['J'], ['B', 'A'], ['E', 'M'], None), 0],\n",
       " [(['J'], [], ['B', 'A'], ['E', 'M'], None), 0],\n",
       " [([], ['J'], ['B', 'A', 'M'], ['E'], None), 0],\n",
       " [(['J'], [], ['B', 'A', 'M'], ['E'], None), 0],\n",
       " [([], ['J'], ['B', 'E'], ['A', 'M'], None), 0],\n",
       " [(['J'], [], ['B', 'E'], ['A', 'M'], None), 0],\n",
       " [([], ['J'], ['B', 'E', 'M'], ['A'], None), 0],\n",
       " [(['J'], [], ['B', 'E', 'M'], ['A'], None), 0],\n",
       " [([], ['J'], ['B', 'E', 'A'], ['M'], None), 0],\n",
       " [(['J'], [], ['B', 'E', 'A'], ['M'], None), 0],\n",
       " [([], ['J'], ['B', 'E', 'A', 'M'], [], None), 0],\n",
       " [(['J'], [], ['B', 'E', 'A', 'M'], [], None), 0],\n",
       " [([], ['M'], [], ['B', 'E', 'A', 'J'], None), 0],\n",
       " [(['M'], [], [], ['B', 'E', 'A', 'J'], None), 0],\n",
       " [([], ['M'], ['J'], ['B', 'E', 'A'], None), 0],\n",
       " [(['M'], [], ['J'], ['B', 'E', 'A'], None), 0],\n",
       " [([], ['M'], ['A'], ['B', 'E', 'J'], None), 0],\n",
       " [(['M'], [], ['A'], ['B', 'E', 'J'], None), 0],\n",
       " [([], ['M'], ['A', 'J'], ['B', 'E'], None), 0],\n",
       " [(['M'], [], ['A', 'J'], ['B', 'E'], None), 0],\n",
       " [([], ['M'], ['E'], ['B', 'A', 'J'], None), 0],\n",
       " [(['M'], [], ['E'], ['B', 'A', 'J'], None), 0],\n",
       " [([], ['M'], ['E', 'J'], ['B', 'A'], None), 0],\n",
       " [(['M'], [], ['E', 'J'], ['B', 'A'], None), 0],\n",
       " [([], ['M'], ['E', 'A'], ['B', 'J'], None), 0],\n",
       " [(['M'], [], ['E', 'A'], ['B', 'J'], None), 0],\n",
       " [([], ['M'], ['E', 'A', 'J'], ['B'], None), 0],\n",
       " [(['M'], [], ['E', 'A', 'J'], ['B'], None), 0],\n",
       " [([], ['M'], ['B'], ['E', 'A', 'J'], None), 0],\n",
       " [(['M'], [], ['B'], ['E', 'A', 'J'], None), 0],\n",
       " [([], ['M'], ['B', 'J'], ['E', 'A'], None), 0],\n",
       " [(['M'], [], ['B', 'J'], ['E', 'A'], None), 0],\n",
       " [([], ['M'], ['B', 'A'], ['E', 'J'], None), 0],\n",
       " [(['M'], [], ['B', 'A'], ['E', 'J'], None), 0],\n",
       " [([], ['M'], ['B', 'A', 'J'], ['E'], None), 0],\n",
       " [(['M'], [], ['B', 'A', 'J'], ['E'], None), 0],\n",
       " [([], ['M'], ['B', 'E'], ['A', 'J'], None), 0],\n",
       " [(['M'], [], ['B', 'E'], ['A', 'J'], None), 0],\n",
       " [([], ['M'], ['B', 'E', 'J'], ['A'], None), 0],\n",
       " [(['M'], [], ['B', 'E', 'J'], ['A'], None), 0],\n",
       " [([], ['M'], ['B', 'E', 'A'], ['J'], None), 0],\n",
       " [(['M'], [], ['B', 'E', 'A'], ['J'], None), 0],\n",
       " [([], ['M'], ['B', 'E', 'A', 'J'], [], None), 0],\n",
       " [(['M'], [], ['B', 'E', 'A', 'J'], [], None), 0],\n",
       " [([], ['A'], [], ['B', 'E', 'M', 'J'], None), 0],\n",
       " [(['A'], [], [], ['B', 'E', 'M', 'J'], None), 0],\n",
       " [([], ['A'], ['J'], ['B', 'E', 'M'], None), 0],\n",
       " [(['A'], [], ['J'], ['B', 'E', 'M'], None), 0],\n",
       " [([], ['A'], ['M'], ['B', 'E', 'J'], None), 0],\n",
       " [(['A'], [], ['M'], ['B', 'E', 'J'], None), 0],\n",
       " [([], ['A'], ['M', 'J'], ['B', 'E'], None), 0],\n",
       " [(['A'], [], ['M', 'J'], ['B', 'E'], None), 0],\n",
       " [([], ['A'], ['E'], ['B', 'M', 'J'], None), 0],\n",
       " [(['A'], [], ['E'], ['B', 'M', 'J'], None), 0],\n",
       " [([], ['A'], ['E', 'J'], ['B', 'M'], None), 0],\n",
       " [(['A'], [], ['E', 'J'], ['B', 'M'], None), 0],\n",
       " [([], ['A'], ['E', 'M'], ['B', 'J'], None), 0],\n",
       " [(['A'], [], ['E', 'M'], ['B', 'J'], None), 0],\n",
       " [([], ['A'], ['E', 'M', 'J'], ['B'], None), 0],\n",
       " [(['A'], [], ['E', 'M', 'J'], ['B'], None), 0],\n",
       " [([], ['A'], ['B'], ['E', 'M', 'J'], None), 0],\n",
       " [(['A'], [], ['B'], ['E', 'M', 'J'], None), 0],\n",
       " [([], ['A'], ['B', 'J'], ['E', 'M'], None), 0],\n",
       " [(['A'], [], ['B', 'J'], ['E', 'M'], None), 0],\n",
       " [([], ['A'], ['B', 'M'], ['E', 'J'], None), 0],\n",
       " [(['A'], [], ['B', 'M'], ['E', 'J'], None), 0],\n",
       " [([], ['A'], ['B', 'M', 'J'], ['E'], None), 0],\n",
       " [(['A'], [], ['B', 'M', 'J'], ['E'], None), 0],\n",
       " [([], ['A'], ['B', 'E'], ['M', 'J'], None), 0],\n",
       " [(['A'], [], ['B', 'E'], ['M', 'J'], None), 0],\n",
       " [([], ['A'], ['B', 'E', 'J'], ['M'], None), 0],\n",
       " [(['A'], [], ['B', 'E', 'J'], ['M'], None), 0],\n",
       " [([], ['A'], ['B', 'E', 'M'], ['J'], None), 0],\n",
       " [(['A'], [], ['B', 'E', 'M'], ['J'], None), 0],\n",
       " [([], ['A'], ['B', 'E', 'M', 'J'], [], None), 0],\n",
       " [(['A'], [], ['B', 'E', 'M', 'J'], [], None), 0]]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bdg_results_unclean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[([], ['A'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A'], [], ['J'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A'], [], ['M'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A'], [], ['M', 'J'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A'], ['J'], [], ('E',)), 0.0014222590323639199],\n",
       " [([], ['A'], ['J'], ['M'], ('E',)), 0.0014222590323639196],\n",
       " [([], ['A'], ['M'], [], ('E',)), 0.0014222590323639199],\n",
       " [([], ['A'], ['M'], ['J'], ('E',)), 0.00142225903236392],\n",
       " [([], ['A'], ['M', 'J'], [], ('E',)), 0.0014222590323639196],\n",
       " [([], ['A', 'J'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A', 'J'], [], ['M'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A', 'J'], ['M'], [], ('E',)), 0.00142225903236392],\n",
       " [([], ['A', 'M'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A', 'M'], [], ['J'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['A', 'M'], ['J'], [], ('E',)), 0.0014222590323639196],\n",
       " [([], ['A', 'M', 'J'], [], [], ('E',)), 0.0014222590323639203],\n",
       " [([], ['J'], [], [], ('E',)), 0.0014832111079134708],\n",
       " [([], ['J'], [], ['A'], ('E',)), 0.0014222590323639203],\n",
       " [([], ['J'], [], ['A', 'M'], ('B',)), 6.013131697154257e-05],\n",
       " [([], ['J'], [], ['B'], ('E',)), 0.0014831466785683834],\n",
       " [([], ['J'], [], ['B', 'M'], ('E',)), 0.001440701707124981],\n",
       " [([], ['J'], [], ['E'], ('B',)), 0.00015922360461467972],\n",
       " [([], ['J'], [], ['E', 'M'], ('B',)), 9.015328226329397e-05],\n",
       " [([], ['J'], [], ['M'], ('E',)), 0.0014407327765951661],\n",
       " [([], ['J'], ['A'], [], ('E',)), 0.23100870196889103],\n",
       " [([], ['J'], ['A'], ['M'], ('B',)), 0.373551228281836],\n",
       " [([], ['J'], ['A', 'M'], [], ('B',)), 0.3735512282818359],\n",
       " [([], ['J'], ['B'], [], ('E',)), 0.0018876297331487648],\n",
       " [([], ['J'], ['B'], ['M'], ('E',)), 0.0017852122026965744],\n",
       " [([], ['J'], ['B', 'M'], [], ('E',)), 0.0020181943345616267],\n",
       " [([], ['J'], ['E'], [], ('B',)), 0.00020272029294149277],\n",
       " [([], ['J'], ['E'], ['M'], ('B',)), 0.00011174749602380887],\n",
       " [([], ['J'], ['E', 'M'], [], ('B',)), 0.0024727772971907254],\n",
       " [([], ['J'], ['M'], [], ('B',)), 0.006876246073421025],\n",
       " [([], ['J'], ['M'], ['A'], ('E',)), 0.00142225903236392],\n",
       " [([], ['J'], ['M'], ['B'], ('E',)), 0.005637035563347659],\n",
       " [([], ['J'], ['M'], ['E'], ('B',)), 0.006901098482896905],\n",
       " [([], ['M'], [], [], ('E',)), 0.0015976396500871493],\n",
       " [([], ['M'], [], ['A'], ('B',)), 6.013131697154258e-05],\n",
       " [([], ['M'], [], ['A', 'J'], ('B',)), 6.013131697154257e-05],\n",
       " [([], ['M'], [], ['B'], ('E',)), 0.0015975145513068226],\n",
       " [([], ['M'], [], ['B', 'J'], ('E',)), 0.001440701707124981],\n",
       " [([], ['M'], [], ['E'], ('B',)), 0.0003453151521163314],\n",
       " [([], ['M'], [], ['E', 'J'], ('B',)), 9.015328226329397e-05],\n",
       " [([], ['M'], [], ['J'], ('E',)), 0.0014407327765951661],\n",
       " [([], ['M'], ['A'], [], ('B',)), 0.3735512282818359],\n",
       " [([], ['M'], ['A'], ['J'], ('B',)), 0.373551228281836],\n",
       " [([], ['M'], ['A', 'J'], [], ('B',)), 0.37355122828183596],\n",
       " [([], ['M'], ['B'], [], ('E',)), 0.0019596574202472156],\n",
       " [([], ['M'], ['B'], ['J'], ('E',)), 0.0017852122026965744],\n",
       " [([], ['M'], ['B', 'J'], [], ('E',)), 0.0020171402594207116],\n",
       " [([], ['M'], ['E'], [], ('B',)), 0.0004237156096830609],\n",
       " [([], ['M'], ['E'], ['J'], ('B',)), 0.00011174749602380887],\n",
       " [([], ['M'], ['E', 'J'], [], ('B',)), 0.00227989956557118],\n",
       " [([], ['M'], ['J'], [], ('B',)), 0.005129858133401301],\n",
       " [([], ['M'], ['J'], ['A'], ('B',)), 6.013131697154255e-05],\n",
       " [([], ['M'], ['J'], ['B'], ('E',)), 0.004551641633239515],\n",
       " [([], ['M'], ['J'], ['E'], ('B',)), 0.005142852044040336],\n",
       " [([], ['M', 'J'], [], [], ('E',)), 0.0014407327765951661],\n",
       " [([], ['M', 'J'], [], ['A'], ('B',)), 6.013131697154257e-05],\n",
       " [([], ['M', 'J'], [], ['B'], ('E',)), 0.001440701707124981],\n",
       " [([], ['M', 'J'], [], ['E'], ('B',)), 9.015328226329397e-05],\n",
       " [([], ['M', 'J'], ['A'], [], ('B',)), 0.373551228281836],\n",
       " [([], ['M', 'J'], ['B'], [], ('E',)), 0.0017852122026965744],\n",
       " [([], ['M', 'J'], ['E'], [], ('B',)), 0.00011174749602380887],\n",
       " [(['A'], [], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A'], [], [], ['J'], ('B',)), 0.373551228281836],\n",
       " [(['A'], [], [], ['M'], ('B',)), 0.3735512282818359],\n",
       " [(['A'], [], [], ['M', 'J'], ('B',)), 0.373551228281836],\n",
       " [(['A'], [], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A'], [], ['J'], ['M'], ('B',)), 0.37355122828183596],\n",
       " [(['A'], [], ['M'], [], ('B',)), 0.373551228281836],\n",
       " [(['A'], [], ['M'], ['J'], ('B',)), 0.3735512282818359],\n",
       " [(['A'], [], ['M', 'J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A'], ['J'], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A'], ['J'], [], ['M'], ('B',)), 0.373551228281836],\n",
       " [(['A'], ['J'], ['M'], [], ('B',)), 0.3735512282818359],\n",
       " [(['A'], ['M'], [], [], ('B',)), 0.3735512282818359],\n",
       " [(['A'], ['M'], [], ['J'], ('B',)), 0.373551228281836],\n",
       " [(['A'], ['M'], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A'], ['M', 'J'], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A', 'J'], [], [], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'J'], [], [], ['M'], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'J'], [], ['M'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'J'], ['M'], [], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'M'], [], [], [], ('B',)), 0.373551228281836],\n",
       " [(['A', 'M'], [], [], ['J'], ('B',)), 0.3735512282818359],\n",
       " [(['A', 'M'], [], ['J'], [], ('B',)), 0.37355122828183596],\n",
       " [(['A', 'M'], ['J'], [], [], ('B',)), 0.3735512282818359],\n",
       " [(['A', 'M', 'J'], [], [], [], ('B',)), 0.37355122828183596],\n",
       " [(['J'], [], [], [], ('A',)), 0.043437711799927066],\n",
       " [(['J'], [], [], ['A'], ('B',)), 6.0131316971542564e-05],\n",
       " [(['J'], [], [], ['A', 'M'], ('B',)), 6.013131697154255e-05],\n",
       " [(['J'], [], [], ['B'], ('A',)), 0.027661940776723615],\n",
       " [(['J'], [], [], ['B', 'M'], ('A',)), 0.008547192182370435],\n",
       " [(['J'], [], [], ['E'], ('A',)), 0.033788238300887834],\n",
       " [(['J'], [], [], ['E', 'M'], ('B', 'A')), 0.005083365847947335],\n",
       " [(['J'], [], [], ['M'], ('A',)), 0.013573889331307635],\n",
       " [(['J'], [], ['A'], [], ('E',)), 0.23100870196889098],\n",
       " [(['J'], [], ['A'], ['M'], ('B',)), 0.37355122828183596],\n",
       " [(['J'], [], ['A', 'M'], [], ('B',)), 0.37355122828183596],\n",
       " [(['J'], [], ['B'], [], ('A',)), 0.9964676796813255],\n",
       " [(['J'], [], ['B'], ['M'], ('A',)), 0.9884372823600297],\n",
       " [(['J'], [], ['B', 'M'], [], ('A',)), 0.99994936196669],\n",
       " [(['J'], [], ['E'], [], ('A',)), 0.8806070133743574],\n",
       " [(['J'], [], ['E'], ['M'], ('B', 'A')), 0.002258110777368501],\n",
       " [(['J'], [], ['E', 'M'], [], ('A',)), 0.9980668823707707],\n",
       " [(['J'], [], ['M'], [], ('A',)), 0.7606920388631078],\n",
       " [(['J'], [], ['M'], ['A'], ('B',)), 6.0131316971542544e-05],\n",
       " [(['J'], [], ['M'], ['B'], ('A',)), 0.6657108744026019],\n",
       " [(['J'], [], ['M'], ['E'], ('A',)), 0.7099672469001376],\n",
       " [(['J'], ['A'], [], [], ('E',)), 0.0014222590323639199],\n",
       " [(['J'], ['A'], [], ['M'], ('E',)), 0.0014222590323639196],\n",
       " [(['J'], ['A'], ['M'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['J'], ['A', 'M'], [], [], ('E',)), 0.0014222590323639196],\n",
       " [(['J'], ['M'], [], [], ('A',)), 0.013573889331307635],\n",
       " [(['J'], ['M'], [], ['A'], ('B',)), 6.013131697154255e-05],\n",
       " [(['J'], ['M'], [], ['B'], ('A',)), 0.008547192182370435],\n",
       " [(['J'], ['M'], [], ['E'], ('A',)), 0.010485794020393492],\n",
       " [(['J'], ['M'], ['A'], [], ('B',)), 0.37355122828183596],\n",
       " [(['J'], ['M'], ['B'], [], ('A',)), 0.9884372823600297],\n",
       " [(['J'], ['M'], ['E'], [], ('B', 'A')), 0.002258110777368501],\n",
       " [(['M'], [], [], [], ('A',)), 0.15009011774975964],\n",
       " [(['M'], [], [], ['A'], ('B',)), 6.013131697154256e-05],\n",
       " [(['M'], [], [], ['A', 'J'], ('B',)), 6.013131697154256e-05],\n",
       " [(['M'], [], [], ['B'], ('A',)), 0.09961384529643373],\n",
       " [(['M'], [], [], ['B', 'J'], ('A',)), 0.011511683274678913],\n",
       " [(['M'], [], [], ['E'], ('A',)), 0.11971342160945007],\n",
       " [(['M'], [], [], ['E', 'J'], ('A',)), 0.014113094727792633],\n",
       " [(['M'], [], [], ['J'], ('A',)), 0.01824973824500748],\n",
       " [(['M'], [], ['A'], [], ('B',)), 0.373551228281836],\n",
       " [(['M'], [], ['A'], ['J'], ('B',)), 0.3735512282818359],\n",
       " [(['M'], [], ['A', 'J'], [], ('E',)), 0.23100870196889098],\n",
       " [(['M'], [], ['B'], [], ('A',)), 0.9990892993739275],\n",
       " [(['M'], [], ['B'], ['J'], ('A',)), 0.9914148016888648],\n",
       " [(['M'], [], ['B', 'J'], [], ('A',)), 0.99994936196669],\n",
       " [(['M'], [], ['E'], [], ('B', 'A')), 0.003158313678965251],\n",
       " [(['M'], [], ['E'], ['J'], ('A',)), 0.7512000215618058],\n",
       " [(['M'], [], ['E', 'J'], [], ('A',)), 0.9980668823707707],\n",
       " [(['M'], [], ['J'], [], ('A',)), 0.7606920388631078],\n",
       " [(['M'], [], ['J'], ['A'], ('E',)), 0.0014222590323639196],\n",
       " [(['M'], [], ['J'], ['B'], ('A',)), 0.6657108744026019],\n",
       " [(['M'], [], ['J'], ['E'], ('A',)), 0.7099672469001376],\n",
       " [(['M'], ['A'], [], [], ('E',)), 0.0014222590323639199],\n",
       " [(['M'], ['A'], [], ['J'], ('E',)), 0.00142225903236392],\n",
       " [(['M'], ['A'], ['J'], [], ('E',)), 0.0014222590323639196],\n",
       " [(['M'], ['A', 'J'], [], [], ('E',)), 0.00142225903236392],\n",
       " [(['M'], ['J'], [], [], ('A',)), 0.01824973824500748],\n",
       " [(['M'], ['J'], [], ['A'], ('B',)), 6.013131697154256e-05],\n",
       " [(['M'], ['J'], [], ['B'], ('A',)), 0.011511683274678913],\n",
       " [(['M'], ['J'], [], ['E'], ('A',)), 0.014113094727792633],\n",
       " [(['M'], ['J'], ['A'], [], ('E',)), 0.23100870196889095],\n",
       " [(['M'], ['J'], ['B'], [], ('A',)), 0.9914148016888648],\n",
       " [(['M'], ['J'], ['E'], [], ('A',)), 0.7512000215618058],\n",
       " [(['M', 'J'], [], [], [], ('A',)), 0.7606920388631078],\n",
       " [(['M', 'J'], [], [], ['A'], ('B',)), 6.0131316971542544e-05],\n",
       " [(['M', 'J'], [], [], ['B'], ('A',)), 0.6657108744026019],\n",
       " [(['M', 'J'], [], [], ['E'], ('A',)), 0.7099672469001376],\n",
       " [(['M', 'J'], [], ['A'], [], ('B',)), 0.37355122828183596],\n",
       " [(['M', 'J'], [], ['B'], [], ('A',)), 0.99994936196669],\n",
       " [(['M', 'J'], [], ['E'], [], ('B', 'A')), 0.003262105340439799],\n",
       " [(['M', 'J'], ['A'], [], [], ('E',)), 0.0014222590323639196]]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sorted(bdg_results_clean)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Sprinkler Tests\n",
    "Example found [here](https://www.bayesserver.com/examples/networks/sprinkler)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "# make the graph\n",
    "sdg = nx.DiGraph()\n",
    "sdg.add_nodes_from(['C', 'S', 'R', 'W'])\n",
    "sdg.add_weighted_edges_from([('C','S', 0),('C','R', 0),('S', 'W', 0),('R','W', 0)])\n",
    "nx.draw_networkx(sdg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# learn the CPTable\n",
    "\n",
    "# Cloudy\n",
    "cpt_C = CPT(0)\n",
    "cpt_C.add_entry_self((.5, .5))\n",
    "sdg.node['C']['CPT'] = cpt_C\n",
    "\n",
    "# Rain\n",
    "cpt_R = CPT(1)\n",
    "cpt_R.add_entry([1], (1-.8, .8))\n",
    "cpt_R.add_entry([0], (1-.2, .2))\n",
    "sdg.node['R']['CPT'] = cpt_R\n",
    "\n",
    "# Sprinkler\n",
    "cpt_S = CPT(1)\n",
    "cpt_S.add_entry([1], (1-.1, .1))\n",
    "cpt_S.add_entry([0], (1-.5, .5))\n",
    "sdg.node['S']['CPT'] = cpt_S\n",
    "\n",
    "# Grass Wet\n",
    "cpt_W = CPT(2)\n",
    "cpt_W.add_entry([1, 1], (1-.99, .99))\n",
    "cpt_W.add_entry([1, 0], (1-.9, .9))\n",
    "cpt_W.add_entry([0, 1], (1-.9, .9))\n",
    "cpt_W.add_entry([0, 0], (1, 0))\n",
    "sdg.node['W']['CPT'] = cpt_W"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# observe the data\n",
    "observeData(sdg, ['W', 'C'], [])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('S',): 0.13043478260869565, ('R',): 0.9758454106280193},\n",
       " (['R'], 0.9758454106280193))"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculate the total marginal probability (both ways)\n",
    "# hyp | obs\n",
    "findBestExplanation(sdg, ['W'], ['C'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('S',): 0.9720000000000001, ('R',): 0.909}, (['S'], 0.9720000000000001))"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# obs | hyp\n",
    "findBestExplanation(sdg, ['W'], ['C'], flipped=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(('R',), 0.9758454106280193)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hypothesisSelection(sdg, ['W'], ['C'], n=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "sdg_results_unclean = graph_tests(sdg)\n",
    "sdg_results_clean = [item for item in sdg_results_unclean if item[1] != 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[([], ['W'], [], [], ('C',)), 0.36100878435817507],\n",
       " [([], ['W'], [], ['C'], ('S',)), 0.0909090909090909],\n",
       " [([], ['W'], [], ['R'], ('C',)), 0.292604501607717],\n",
       " [([], ['W'], [], ['S'], ('C',)), 0.38066465256797577],\n",
       " [([], ['W'], ['C'], [], ('R',)), 0.2857142857142857],\n",
       " [([], ['W'], ['R'], [], ('C',)), 0.8687350835322196],\n",
       " [([], ['W'], ['S'], [], ('C',)), 0.0639269406392694],\n",
       " [(['W'], [], [], [], ('S',)), 0.4297635605006954],\n",
       " [(['W'], [], [], ['C'], ('R',)), 0.3442622950819672],\n",
       " [(['W'], [], [], ['R'], ('S',)), 1.0],\n",
       " [(['W'], [], [], ['S'], ('R',)), 1.0],\n",
       " [(['W'], [], ['C'], [], ('S',)), 0.13043478260869565],\n",
       " [(['W'], [], ['R'], [], ('S',)), 0.19449901768172886],\n",
       " [(['W'], [], ['S'], [], ('R',)), 0.3203883495145631]]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sorted(sdg_results_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('S',): 0.13043478260869565, ('R',): 0.9758454106280193},\n",
       " (['R'], 0.9758454106280193))"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# test grass is wet and is cloudy\n",
    "observeData(sdg, ['W', 'C'], [])\n",
    "findBestExplanation(sdg, ['W'], ['C'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({('S',): 0.9720000000000001, ('R',): 0.909}, (['S'], 0.9720000000000001))"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "findBestExplanation(sdg, ['W'], ['C'], flipped=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Asia Tests\n",
    "Example found [here](https://www.bayesserver.com/examples/networks/asia)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "# make the graph\n",
    "adg = nx.DiGraph()\n",
    "adg.add_nodes_from(['V', 'T', 'S', 'L', 'B', 'O', 'X', 'D'])\n",
    "adg.add_weighted_edges_from([('V','T', 0),('T','O', 0),('O','X', 0),\n",
    "                             ('O','D', 0), ('S','L', 0), ('S','B', 0),\n",
    "                            ('L','O', 0), ('B','D', 0)])\n",
    "nx.draw_networkx(adg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# learn the CPTable\n",
    "\n",
    "# Visit to Asia\n",
    "cpt_V = CPT(0)\n",
    "cpt_V.add_entry_self((.99, .01))\n",
    "adg.node['V']['CPT'] = cpt_V\n",
    "\n",
    "# Smoker\n",
    "cpt_S = CPT(0)\n",
    "cpt_S.add_entry_self((.5, .5))\n",
    "adg.node['S']['CPT'] = cpt_S\n",
    "\n",
    "# Tuburculosis\n",
    "cpt_T = CPT(1)\n",
    "cpt_T.add_entry([1], (1-.05, .05))\n",
    "cpt_T.add_entry([0], (1-.01, .01))\n",
    "adg.node['T']['CPT'] = cpt_T\n",
    "\n",
    "# Lung Cancer\n",
    "cpt_L = CPT(1)\n",
    "cpt_L.add_entry([1], (1-.1, .1))\n",
    "cpt_L.add_entry([0], (1-.01, .01))\n",
    "adg.node['L']['CPT'] = cpt_L\n",
    "\n",
    "# Bronchitus\n",
    "cpt_B = CPT(1)\n",
    "cpt_B.add_entry([1], (1-.6, .6))\n",
    "cpt_B.add_entry([0], (1-.3, .3))\n",
    "adg.node['B']['CPT'] = cpt_B\n",
    "\n",
    "# Tub or Cancer\n",
    "cpt_O = CPT(2)\n",
    "cpt_O.add_entry([1, 1], (.01, .99))\n",
    "cpt_O.add_entry([1, 0], (.01, .99))\n",
    "cpt_O.add_entry([0, 1], (.01, .99))\n",
    "cpt_O.add_entry([0, 0], (.99, .01))\n",
    "adg.node['O']['CPT'] = cpt_O\n",
    "\n",
    "# Dyspena\n",
    "cpt_D = CPT(2)\n",
    "cpt_D.add_entry([1, 1], (1-.9, .9))\n",
    "cpt_D.add_entry([1, 0], (1-.8, .8))\n",
    "cpt_D.add_entry([0, 1], (1-.7, .7))\n",
    "cpt_D.add_entry([0, 0], (1-.1, .1))\n",
    "adg.node['D']['CPT'] = cpt_D\n",
    "\n",
    "# Xray result\n",
    "cpt_X = CPT(1)\n",
    "cpt_X.add_entry([1], (1-.98, .98))\n",
    "cpt_X.add_entry([0], (1-.05, .05))\n",
    "adg.node['X']['CPT'] = cpt_X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# observe the data\n",
    "observeData(adg, ['O'], ['T', 'L'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# multiple tests\n",
    "adg_results_unclean = graph_tests(adg)\n",
    "adg_results_clean = [item for item in adg_results_unclean if item[1] != 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "final_results = {}\n",
    "for i in bdg_results_clean:\n",
    "    new_i = []\n",
    "    for x in i[0]:\n",
    "        new_i.append(tuple(x))\n",
    "    new_i = tuple(new_i)\n",
    "    if new_i in final_results.keys():\n",
    "        final_results[new_i].append(i)\n",
    "    else:\n",
    "        final_results[new_i] = [i]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0, 0, \"((), ('A',), (), (), ('E',))\"),\n",
       " Text(0, 0, \"(('A',), (), (), (), ('B',))\"),\n",
       " Text(0, 0, \"((), ('M',), (), (), ('E',))\"),\n",
       " Text(0, 0, \"(('M',), (), (), (), ('A',))\"),\n",
       " Text(0, 0, \"((), ('J',), (), (), ('E',))\"),\n",
       " Text(0, 0, \"(('J',), (), (), (), ('A',))\"),\n",
       " Text(0, 0, \"((), ('A', 'M'), (), (), ('E',))\"),\n",
       " Text(0, 0, \"(('M',), ('A',), (), (), ('E',))\"),\n",
       " Text(0, 0, \"(('A',), ('M',), (), (), ('B',))\"),\n",
       " Text(0, 0, \"(('A', 'M'), (), (), (), ('B',))\")]"
      ]
     },
     "execution_count": 78,
     "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": [
    "fig1, ax1 = plt.subplots()\n",
    "ax1.set_ylabel('Probabilities')\n",
    "ax1.set_xlabel('Effect Nodes')\n",
    "ax1.boxplot(all_data[:int(len(all_data)/16)])\n",
    "ax1.set_xticklabels(list(final_results.keys())[:int(len(final_results.keys())/16)],\n",
    "                    rotation=45, fontsize=8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig1, ax1 = plt.subplots()\n",
    "ax1.set_ylabel('Probabilities')\n",
    "ax1.set_xlabel('Effect Nodes')\n",
    "ax1.boxplot(all_data[int(len(all_data)/2):int(3*len(all_data)/4)])\n",
    "ax1.set_xticklabels(list(final_results.keys())[int(len(final_results.keys())/2):int(3*len(final_results.keys())/4)],\n",
    "                    rotation=45, fontsize=8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fig1, ax1 = plt.subplots()\n",
    "ax1.set_ylabel('Probabilities')\n",
    "ax1.set_xlabel('Effect Nodes')\n",
    "ax1.boxplot(all_data[int(3*len(all_data)/4):])\n",
    "ax1.set_xticklabels(list(final_results.keys())[int(3*len(final_results.keys())/4):],\n",
    "                    rotation=45, fontsize=8)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Export data with Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convert(lst):\n",
    "    final_results = {}\n",
    "    for i in lst:\n",
    "        new_i = []\n",
    "        for x in i[0]:\n",
    "            new_i.append(tuple(x))\n",
    "        new_i = tuple(new_i)\n",
    "        final_results[new_i] = i[1]\n",
    "\n",
    "    #res_dct = {tuple(lst[i]): lst[i] for i in range(0, len(lst))} \n",
    "    return final_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_df = pd.DataFrame({ key:pd.Series(value) for key, value in dict_adg_results_clean.items() })\n",
    "dict_df.to_csv(r'asia_graph_hypothesis_results_updated.csv', index = False, header=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_bdg_results_clean = convert(bdg_results_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_df = pd.DataFrame({ key:pd.Series(value) for key, value in dict_bdg_results_clean.items() })\n",
    "dict_df.to_csv(r'burglar_graph_hypothesis_results.csv', index = False, header=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_sdg_results_clean = convert(sdg_results_clean)\n",
    "dict_df = pd.DataFrame({ key:pd.Series(value) for key, value in dict_sdg_results_clean.items() })\n",
    "dict_df.to_csv(r'sprinkler_graph_hypothesis_results.csv', index = False, header=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Unused Stuff"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# adds new true nodes and new false nodes to the graph and then checks for contradictions\n",
    "# only finds direct contraidctions? Should talk with group about this.\n",
    "import copy \n",
    "def findContradictions(graph, new_true_nodes, new_false_nodes, threshold=.5):\n",
    "    new_g = copy.deepcopy(graph)\n",
    "    observeData(new_g, new_true_nodes, new_false_nodes)\n",
    "    edge_attrs = nx.get_edge_attributes(new_g, 'weight')\n",
    "    for i in new_g.nodes():\n",
    "        if 'observed' in new_g.nodes.data()[i]:\n",
    "            if new_g.nodes.data()[i]['observed'] is True:\n",
    "                for x in edge_attrs:\n",
    "                    print(x, x[1], new_g.nodes.data()[x[1]], edge_attrs[x])\n",
    "                    if 'observed' in new_g.nodes.data()[x[1]]:\n",
    "                        if x[0] == i and new_g.nodes.data()[x[1]]['observed'] is False and edge_attrs[x] >= threshold:\n",
    "                            return (i, True, x[1], False, edge_attrs[x])\n",
    "    return \"No contradictions found\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sees if some nodes are independent based on the nodes we are conditioning on\n",
    "# we are only checking for *direct* conditions. chains are not accounted for\n",
    "# since in baysien networks edges are \"direct dependence.\" Should double-check this.\n",
    "def independenceChecker(graph, conditions, nodes):\n",
    "    edges = nx.get_edge_attributes(graph, 'weight')\n",
    "    for i in edges:\n",
    "        if i[1] in nodes and not i[0] in conditions:\n",
    "            return False\n",
    "    return True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getMarkovBlanket(graph, obs, hyp):\n",
    "    #generates a new graph (markov blanket) based on the graph, observations, and hypothesis\n",
    "    #get all parent nodes\n",
    "    edge_attrs = nx.get_edge_attributes(graph, 'weight')\n",
    "    new_graph = nx.DiGraph()\n",
    "    new_graph.add_node(hyp)\n",
    "    new_graph.add_nodes_from(obs)\n",
    "    for i in edge_attrs:\n",
    "        if i[0] in obs or i[0] == hyp:\n",
    "            new_graph.add_node(i[1])\n",
    "            new_graph.add_edge(i[0], i[1], weight=edge_attrs[i])\n",
    "        if i[1] in obs or i[1] == hyp:\n",
    "            new_graph.add_node(i[0])\n",
    "            new_graph.add_edge(i[0], i[1], weight=edge_attrs[i])\n",
    "    return new_graph"
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}