asia_bnn_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load an Example Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pgmpy provides a method pgmpy.utils.get_example_model to quickly load any model from the bnlearn repository: https://www.bnlearn.com/bnrepository/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\adhav\\AppData\\Local\\Programs\\Python\\Python313\\Lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "from pgmpy.utils import get_example_model\n",
    "\n",
    "# Load the model\n",
    "asia_model = get_example_model('asia')\n",
    "\n",
    "# Visualize the network\n",
    "viz = asia_model.to_graphviz()\n",
    "viz.draw('./images/asia.png', prog='neato')\n",
    "Image('./images/asia.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Nodes in the model: ['asia', 'tub', 'smoke', 'lung', 'bronc', 'either', 'xray', 'dysp'] \n",
      "\n",
      "Edges in the model: [('asia', 'tub'), ('tub', 'either'), ('smoke', 'lung'), ('smoke', 'bronc'), ('lung', 'either'), ('bronc', 'dysp'), ('either', 'xray'), ('either', 'dysp')] \n",
      "\n",
      "CPDs in the model: \n",
      "[<TabularCPD representing P(asia:2) at 0x1e5b78bdbe0>,\n",
      " <TabularCPD representing P(bronc:2 | smoke:2) at 0x1e5b77eee90>,\n",
      " <TabularCPD representing P(dysp:2 | bronc:2, either:2) at 0x1e5b77ef110>,\n",
      " <TabularCPD representing P(either:2 | lung:2, tub:2) at 0x1e5b78d5940>,\n",
      " <TabularCPD representing P(lung:2 | smoke:2) at 0x1e5b78d5a70>,\n",
      " <TabularCPD representing P(smoke:2) at 0x1e5b9072450>,\n",
      " <TabularCPD representing P(tub:2 | asia:2) at 0x1e5b788f240>,\n",
      " <TabularCPD representing P(xray:2 | either:2) at 0x1e5b788f020>]\n"
     ]
    }
   ],
   "source": [
    "import pprint\n",
    "\n",
    "# Access attributes of the model\n",
    "nodes = asia_model.nodes()\n",
    "edges = asia_model.edges()\n",
    "cpds = asia_model.get_cpds()\n",
    "\n",
    "print(f\"Nodes in the model: {nodes} \\n\")\n",
    "print(f\"Edges in the model: {edges} \\n\")\n",
    "print(f\"CPDs in the model: \")\n",
    "pprint.pp(cpds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Defining the model manually"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Defining a network in pgmpy involves two steps: 1. Define the network structure 2. Define the parameterization, i.e. CPDs\n",
    "\n",
    "And lastly, CPDs can be associated with the network. Taking an example of the Cancer network from the bnlearn repository as shown below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pgmpy.models import BayesianNetwork\n",
    "from pgmpy.factors.discrete import TabularCPD\n",
    "\n",
    "# Step 1: Define the network structure.\n",
    "cancer_model = BayesianNetwork(\n",
    "    [\n",
    "        (\"Pollution\", \"Cancer\"),\n",
    "        (\"Smoker\", \"Cancer\"),\n",
    "        (\"Cancer\", \"Xray\"),\n",
    "        (\"Cancer\", \"Dyspnoea\"),\n",
    "    ]\n",
    ")\n",
    "\n",
    "# Step 2: Define the CPDs.\n",
    "cpd_poll = TabularCPD(variable=\"Pollution\", variable_card=2, values=[[0.9], [0.1]])\n",
    "cpd_smoke = TabularCPD(variable=\"Smoker\", variable_card=2, values=[[0.3], [0.7]])\n",
    "cpd_cancer = TabularCPD(\n",
    "    variable=\"Cancer\",\n",
    "    variable_card=2,\n",
    "    values=[[0.03, 0.05, 0.001, 0.02], [0.97, 0.95, 0.999, 0.98]],\n",
    "    evidence=[\"Smoker\", \"Pollution\"],\n",
    "    evidence_card=[2, 2],\n",
    ")\n",
    "cpd_xray = TabularCPD(\n",
    "    variable=\"Xray\",\n",
    "    variable_card=2,\n",
    "    values=[[0.9, 0.2], [0.1, 0.8]],\n",
    "    evidence=[\"Cancer\"],\n",
    "    evidence_card=[2],\n",
    ")\n",
    "cpd_dysp = TabularCPD(\n",
    "    variable=\"Dyspnoea\",\n",
    "    variable_card=2,\n",
    "    values=[[0.65, 0.3], [0.35, 0.7]],\n",
    "    evidence=[\"Cancer\"],\n",
    "    evidence_card=[2],\n",
    ")\n",
    "\n",
    "# Step 3: Add the CPDs to the model.\n",
    "cancer_model.add_cpds(cpd_poll, cpd_smoke, cpd_cancer, cpd_xray, cpd_dysp)\n",
    "\n",
    "# Step 4: Check if the model is correctly defined.\n",
    "cancer_model.check_model()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate a complete random model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "pgmpy offers methods to: 1. Generate a completely random network 2. Given a network structure, generate random CPDs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pgmpy.models import BayesianNetwork\n",
    "\n",
    "random_model = BayesianNetwork.get_random(n_nodes=8, edge_prob=0.4, n_states=2, latents=True)\n",
    "\n",
    "# Visualize the network\n",
    "viz = random_model.to_graphviz()\n",
    "viz.draw('random.png', prog='neato')\n",
    "Image('random.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Nodes in the model: [0, 6, 1, 2, 3, 4, 7, 5] \n",
      "\n",
      "Edges in the model: [(0, 6), (1, 2), (1, 3), (1, 4), (1, 7), (2, 3), (2, 4), (2, 6), (3, 7), (4, 5), (4, 6), (5, 7)] \n",
      "\n",
      "CPDs in the model: \n",
      "[<TabularCPD representing P(0:2) at 0x1e5c0938590>,\n",
      " <TabularCPD representing P(6:2 | 0:2, 2:2, 4:2) at 0x1e5b938ca10>,\n",
      " <TabularCPD representing P(1:2) at 0x1e5b796a510>,\n",
      " <TabularCPD representing P(2:2 | 1:2) at 0x1e5b7969850>,\n",
      " <TabularCPD representing P(3:2 | 1:2, 2:2) at 0x1e5b9091d30>,\n",
      " <TabularCPD representing P(4:2 | 1:2, 2:2) at 0x1e5b9091e90>,\n",
      " <TabularCPD representing P(7:2 | 1:2, 3:2, 5:2) at 0x1e5b795bcf0>,\n",
      " <TabularCPD representing P(5:2 | 4:2) at 0x1e5b8ff0230>]\n"
     ]
    }
   ],
   "source": [
    "import pprint\n",
    "\n",
    "# Access attributes of the model\n",
    "nodes = random_model.nodes()\n",
    "edges = random_model.edges()\n",
    "cpds = random_model.get_cpds()\n",
    "\n",
    "print(f\"Nodes in the model: {nodes} \\n\")\n",
    "print(f\"Edges in the model: {edges} \\n\")\n",
    "print(f\"CPDs in the model: \")\n",
    "pprint.pp(cpds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate random CPDs for a given network structure¶\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Nodes in the model: ['Pollution', 'Cancer', 'Smoker', 'Xray', 'Dyspnoea'] \n",
      "\n",
      "Edges in the model: [('Pollution', 'Cancer'), ('Cancer', 'Xray'), ('Cancer', 'Dyspnoea'), ('Smoker', 'Cancer')] \n",
      "\n",
      "CPDs in the model: \n",
      "[<TabularCPD representing P(Pollution:3) at 0x1e5b8fc4ef0>,\n",
      " <TabularCPD representing P(Cancer:3 | Pollution:3, Smoker:3) at 0x1e5b8fc4f80>,\n",
      " <TabularCPD representing P(Smoker:3) at 0x1e5b8fc5010>,\n",
      " <TabularCPD representing P(Xray:3 | Cancer:3) at 0x1e5b8fc50a0>,\n",
      " <TabularCPD representing P(Dyspnoea:3 | Cancer:3) at 0x1e5b8fc5130>]\n"
     ]
    }
   ],
   "source": [
    "# Taking the example of the cancer model.\n",
    "\n",
    "cancer_random = BayesianNetwork(\n",
    "    [\n",
    "        (\"Pollution\", \"Cancer\"),\n",
    "        (\"Smoker\", \"Cancer\"),\n",
    "        (\"Cancer\", \"Xray\"),\n",
    "        (\"Cancer\", \"Dyspnoea\"),\n",
    "    ]\n",
    ")\n",
    "\n",
    "cancer_random.get_random_cpds(n_states=3, inplace=True)\n",
    "\n",
    "# Access attributes of the model\n",
    "nodes = cancer_random.nodes()\n",
    "edges = cancer_random.edges()\n",
    "cpds = cancer_random.get_cpds()\n",
    "\n",
    "print(f\"Nodes in the model: {nodes} \\n\")\n",
    "print(f\"Edges in the model: {edges} \\n\")\n",
    "print(f\"CPDs in the model: \")\n",
    "pprint.pp(cpds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Inference in Discrete Bayesian Network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "a simple example for doing Exact inference in Bayesian Networks using pgmpy. We will be using the Asia network (http://www.bnlearn.com/bnrepository/#asia) for this example."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## initialize inference class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pgmpy.inference import VariableElimination\n",
    "\n",
    "asia_infer = VariableElimination(asia_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Doing inference using hard evidence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+--------------+\n",
      "| bronc      |   phi(bronc) |\n",
      "+============+==============+\n",
      "| bronc(yes) |       0.3000 |\n",
      "+------------+--------------+\n",
      "| bronc(no)  |       0.7000 |\n",
      "+------------+--------------+\n",
      "+------------+-----------+-------------------+\n",
      "| bronc      | asia      |   phi(bronc,asia) |\n",
      "+============+===========+===================+\n",
      "| bronc(yes) | asia(yes) |            0.0060 |\n",
      "+------------+-----------+-------------------+\n",
      "| bronc(yes) | asia(no)  |            0.5940 |\n",
      "+------------+-----------+-------------------+\n",
      "| bronc(no)  | asia(yes) |            0.0040 |\n",
      "+------------+-----------+-------------------+\n",
      "| bronc(no)  | asia(no)  |            0.3960 |\n",
      "+------------+-----------+-------------------+\n",
      "+------------+--------------+\n",
      "| bronc      |   phi(bronc) |\n",
      "+============+==============+\n",
      "| bronc(yes) |       0.3000 |\n",
      "+------------+--------------+\n",
      "| bronc(no)  |       0.7000 |\n",
      "+------------+--------------+\n",
      "+-----------+-------------+\n",
      "| asia      |   phi(asia) |\n",
      "+===========+=============+\n",
      "| asia(yes) |      0.0100 |\n",
      "+-----------+-------------+\n",
      "| asia(no)  |      0.9900 |\n",
      "+-----------+-------------+\n"
     ]
    }
   ],
   "source": [
    "# Computing the probability of bronc given smoke=no.\n",
    "q = asia_infer.query(variables=[\"bronc\"], evidence={\"smoke\": \"no\"})\n",
    "print(q)\n",
    "\n",
    "# Computing the joint probability of bronc and asia given smoke=yes\n",
    "q = asia_infer.query(variables=[\"bronc\", \"asia\"], evidence={\"smoke\": \"yes\"})\n",
    "print(q)\n",
    "\n",
    "# Computing the probabilities (not joint) of bronc and asia given smoke=no\n",
    "q = asia_infer.query(variables=[\"bronc\", \"asia\"], evidence={\"smoke\": \"no\"}, joint=False)\n",
    "for factor in q.values():\n",
    "    print(factor)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Finding Elimination Order: : : 0it [00:00, ?it/s]\n",
      "0it [00:00, ?it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'bronc': 'no'}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Finding Elimination Order: : : 0it [00:00, ?it/s]\n",
      "0it [00:00, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'bronc': 'yes', 'asia': 'no'}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# Computing the MAP of bronc given smoke=no.\n",
    "q = asia_infer.map_query(variables=[\"bronc\"], evidence={\"smoke\": \"no\"})\n",
    "print(q)\n",
    "\n",
    "# Computing the MAP of bronc and asia given smoke=yes\n",
    "q = asia_infer.map_query(variables=[\"bronc\", \"asia\"], evidence={\"smoke\": \"yes\"})\n",
    "print(q)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inferencing from virtual evidence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------+--------------+\n",
      "| bronc      |   phi(bronc) |\n",
      "+============+==============+\n",
      "| bronc(yes) |       0.3000 |\n",
      "+------------+--------------+\n",
      "| bronc(no)  |       0.7000 |\n",
      "+------------+--------------+\n"
     ]
    }
   ],
   "source": [
    "from pgmpy.factors.discrete import TabularCPD\n",
    "\n",
    "# Get state name, which ensure lung evidence has same state names as asia_model\n",
    "state_names = asia_model.get_cpds('lung').state_names\n",
    "\n",
    "lung_virt_evidence = TabularCPD(variable=\"lung\", variable_card=2, values=[[0.4], [0.6]],state_names=state_names)\n",
    "\n",
    "# Query with hard evidence smoke = no and virtual evidence lung = [0.4, 0.6]\n",
    "q = asia_infer.query(\n",
    "    variables=[\"bronc\"], evidence={\"smoke\": \"no\"}, virtual_evidence=[lung_virt_evidence]\n",
    ")\n",
    "print(q)\n",
    "\n",
    "# Query with hard evidence smoke = no and virtual evidences lung = [0.4, 0.6] and bronc = [0.3, 0.7]\n",
    "lung_virt_evidence = TabularCPD(variable=\"lung\", variable_card=2, values=[[0.4], [0.7]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+------------+-----------+\n",
      "| smoke     | smoke(yes) | smoke(no) |\n",
      "+-----------+------------+-----------+\n",
      "| lung(yes) | 0.1        | 0.01      |\n",
      "+-----------+------------+-----------+\n",
      "| lung(no)  | 0.9        | 0.99      |\n",
      "+-----------+------------+-----------+\n"
     ]
    }
   ],
   "source": [
    "print(asia_model.get_cpds(\"lung\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Troubleshooting for slow inference"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the case of large models, or models in which variables have a lot of states, inference can be quite slow. Some of the ways to deal with it are:\n",
    "\n",
    "Reduce the number of states of variables by combining states together.\n",
    "\n",
    "Try a different elimination order by specifying `elimination_order` argument. Possible options are: MinFill, MinNeighbors, MinWeight, WeightedMinFill.\n",
    "\n",
    "Try a custom elimination order: The implemented heuristics for computing the elimination order might not be efficient in every case. If you can think of a more efficient order, you can also pass it as a list to the `elimination_order` argument.\n",
    "\n",
    "If it is still too slow, try using approximate inference using sampling algorithms."
   ]
  }
 ],
 "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": "3.13.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}