notebooks/explore_hytrails.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b421cdab-5ad6-4d5c-bd90-817ff2306d2f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T10:08:40.395101Z",
     "start_time": "2025-02-17T10:08:35.410204Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: prettytable in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (3.14.0)\r\n",
      "Requirement already satisfied: wcwidth in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from prettytable) (0.2.13)\r\n",
      "Note: you may need to restart the kernel to use updated packages.\n",
      "Requirement already satisfied: tables in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (3.9.2)\r\n",
      "Requirement already satisfied: numpy>=1.19.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from tables) (1.23.0)\r\n",
      "Requirement already satisfied: numexpr>=2.6.2 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from tables) (2.10.2)\r\n",
      "Requirement already satisfied: packaging in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from tables) (24.2)\r\n",
      "Requirement already satisfied: py-cpuinfo in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from tables) (9.0.0)\r\n",
      "Requirement already satisfied: blosc2>=2.3.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from tables) (2.5.1)\r\n",
      "Requirement already satisfied: ndindex>=1.4 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from blosc2>=2.3.0->tables) (1.9.2)\r\n",
      "Requirement already satisfied: msgpack in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from blosc2>=2.3.0->tables) (1.1.0)\r\n",
      "Note: you may need to restart the kernel to use updated packages.\n",
      "Requirement already satisfied: seaborn in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (0.13.2)\r\n",
      "Requirement already satisfied: numpy!=1.24.0,>=1.20 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from seaborn) (1.23.0)\r\n",
      "Requirement already satisfied: pandas>=1.2 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from seaborn) (1.5.3)\r\n",
      "Requirement already satisfied: matplotlib!=3.6.1,>=3.4 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from seaborn) (3.9.4)\r\n",
      "Requirement already satisfied: contourpy>=1.0.1 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (1.3.0)\r\n",
      "Requirement already satisfied: cycler>=0.10 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (0.12.1)\r\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (4.56.0)\r\n",
      "Requirement already satisfied: kiwisolver>=1.3.1 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (1.4.7)\r\n",
      "Requirement already satisfied: packaging>=20.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (24.2)\r\n",
      "Requirement already satisfied: pillow>=8 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (11.1.0)\r\n",
      "Requirement already satisfied: pyparsing>=2.3.1 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (3.2.1)\r\n",
      "Requirement already satisfied: python-dateutil>=2.7 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (2.9.0.post0)\r\n",
      "Requirement already satisfied: importlib-resources>=3.2.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from matplotlib!=3.6.1,>=3.4->seaborn) (6.5.2)\r\n",
      "Requirement already satisfied: pytz>=2020.1 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from pandas>=1.2->seaborn) (2025.1)\r\n",
      "Requirement already satisfied: zipp>=3.1.0 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from importlib-resources>=3.2.0->matplotlib!=3.6.1,>=3.4->seaborn) (3.21.0)\r\n",
      "Requirement already satisfied: six>=1.5 in /Users/lisa/RecBole/venv/torstest/lib/python3.9/site-packages (from python-dateutil>=2.7->matplotlib!=3.6.1,>=3.4->seaborn) (1.17.0)\r\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "%pip install prettytable\n",
    "%pip install tables\n",
    "%pip install seaborn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8ccdfa7f-60cd-4c3d-9e7b-3c1bd0311dc0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T10:08:44.654799Z",
     "start_time": "2025-02-17T10:08:42.480963Z"
    }
   },
   "outputs": [],
   "source": [
    "# We use Hyptrails and Pathtools from https://github.com/psinger/HypTrails, ported to Python 3\n",
    "import pandas as pd\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "outputs": [
    {
     "data": {
      "text/plain": "'/Users/lisa/RecBole/hyptrails'"
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.getcwd()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2025-02-17T10:11:33.316957Z",
     "start_time": "2025-02-17T10:11:33.306511Z"
    }
   },
   "id": "eb82749955154d91",
   "execution_count": 4
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0b2fe584-2bd2-477d-a44c-8ec613e311af",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T10:11:31.228054Z",
     "start_time": "2025-02-17T10:11:31.222938Z"
    }
   },
   "outputs": [],
   "source": [
    "os.chdir(\"../hyptrails/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5c5837e6-992d-4d7a-bfe0-2ef8e079d6fa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T10:11:40.663608Z",
     "start_time": "2025-02-17T10:11:37.247799Z"
    }
   },
   "outputs": [],
   "source": [
    "from hyptrails.hyptrails import *\n",
    "from hyptrails.hyptrails.trial_roulette import *\n",
    "from hyptrails.pathtools.markovchain import *\n",
    "from hyptrails.pathtools.markovtools import *\n",
    "from scipy.sparse import lil_matrix, csr_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "61056273-f4b7-4197-961a-b07badd4a23f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T10:04:55.182502Z",
     "start_time": "2025-02-17T10:04:55.178035Z"
    }
   },
   "outputs": [],
   "source": [
    "fig_name = \"coveo\"\n",
    "cat_id = \"category_product_id:token\"\n",
    "cat = \"category_hash:token_seq\"\n",
    "list_name= \"SEARCHLIST\"\n",
    "item_id = 'product_sku_hash:token'\n",
    "session_id ='session_id_hash:token'\n",
    "interactions_path = \"../datasets/final/coveo-sl-search/coveo-sl-search-recbole/coveo-sl-search-recbole.train.inter\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "25ca2be9-b6af-4a7d-97d0-30c483a3049a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:28:41.206059Z",
     "start_time": "2025-02-17T11:28:41.169052Z"
    }
   },
   "outputs": [],
   "source": [
    "fig_name = \"ml-1m-extended\"\n",
    "cat_id = \"title_genres:token\"\n",
    "cat = \"genres:token_seq\"\n",
    "list_name= \"OVERVIEW\"\n",
    "item_id = 'title:token'\n",
    "session_id ='userId:token'\n",
    "interactions_path = \"../datasets/final/ml-1m-extended/ml-1m-extended.train.inter\""
   ]
  },
  {
   "cell_type": "code",
   "outputs": [
    {
     "data": {
      "text/plain": "'/Users/lisa/RecBole/hyptrails'"
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.getcwd()"
   ],
   "metadata": {
    "collapsed": false,
    "ExecuteTime": {
     "end_time": "2025-02-17T11:28:41.933493Z",
     "start_time": "2025-02-17T11:28:41.921864Z"
    }
   },
   "id": "7a258ffcd1456c16",
   "execution_count": 16
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "e620e70f-f837-45b4-8263-105d29b3fae4",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:28:43.407368Z",
     "start_time": "2025-02-17T11:28:42.092159Z"
    }
   },
   "outputs": [],
   "source": [
    "interaction_df = pd.read_csv(interactions_path, sep = \"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "1105a5f6-c723-47f3-a4f1-f7812c16deaf",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:28:43.412888Z",
     "start_time": "2025-02-17T11:28:43.409629Z"
    }
   },
   "outputs": [],
   "source": [
    "#interaction_df =interaction_df.head(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "a0fcee75-007b-4725-b4e7-fd4f9db6417c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:03.772230Z",
     "start_time": "2025-02-17T11:28:48.323400Z"
    }
   },
   "outputs": [],
   "source": [
    "interaction_df[\"states\"] = interaction_df.apply(\n",
    "    lambda row: f\"{row[item_id]} + {row[cat_id]}\" if row[item_id] == list_name else row[cat], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "07f7c3f1-eec3-437d-9ab1-8c0a6fba0055",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:03.801980Z",
     "start_time": "2025-02-17T11:29:03.775261Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": "   timestamp:float  item_id_type:float  userId:token  \\\n0        978300019                   0             1   \n1        978300019                   1             1   \n2        978300055                   0             1   \n3        978300055                   1             1   \n4        978300055                   0             1   \n\n                 title:token         title_genres:token genres:token_seq  \\\n0              OVERVIEW-PAGE                      Drama            Drama   \n1   Girl, Interrupted (1999)   Girl, Interrupted (1999)            Drama   \n2              OVERVIEW-PAGE              Comedy|Sci-Fi    Comedy|Sci-Fi   \n3  Back to the Future (1985)  Back to the Future (1985)    Comedy|Sci-Fi   \n4              OVERVIEW-PAGE              Drama|Romance    Drama|Romance   \n\n          states  \n0          Drama  \n1          Drama  \n2  Comedy|Sci-Fi  \n3  Comedy|Sci-Fi  \n4  Drama|Romance  ",
      "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>timestamp:float</th>\n      <th>item_id_type:float</th>\n      <th>userId:token</th>\n      <th>title:token</th>\n      <th>title_genres:token</th>\n      <th>genres:token_seq</th>\n      <th>states</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>978300019</td>\n      <td>0</td>\n      <td>1</td>\n      <td>OVERVIEW-PAGE</td>\n      <td>Drama</td>\n      <td>Drama</td>\n      <td>Drama</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>978300019</td>\n      <td>1</td>\n      <td>1</td>\n      <td>Girl, Interrupted (1999)</td>\n      <td>Girl, Interrupted (1999)</td>\n      <td>Drama</td>\n      <td>Drama</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>978300055</td>\n      <td>0</td>\n      <td>1</td>\n      <td>OVERVIEW-PAGE</td>\n      <td>Comedy|Sci-Fi</td>\n      <td>Comedy|Sci-Fi</td>\n      <td>Comedy|Sci-Fi</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>978300055</td>\n      <td>1</td>\n      <td>1</td>\n      <td>Back to the Future (1985)</td>\n      <td>Back to the Future (1985)</td>\n      <td>Comedy|Sci-Fi</td>\n      <td>Comedy|Sci-Fi</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>978300055</td>\n      <td>0</td>\n      <td>1</td>\n      <td>OVERVIEW-PAGE</td>\n      <td>Drama|Romance</td>\n      <td>Drama|Romance</td>\n      <td>Drama|Romance</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interaction_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "cc85c72d-5678-4fbd-be7d-acf416038082",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:05.154584Z",
     "start_time": "2025-02-17T11:29:03.805241Z"
    }
   },
   "outputs": [],
   "source": [
    "#Build vocabulary/state\n",
    "vocab = {page: idx for idx, page in enumerate(interaction_df[\"states\"].unique())}\n",
    "reduced_df = interaction_df[[\"states\", cat]].drop_duplicates()\n",
    "overview_genres = interaction_df[interaction_df[item_id] == list_name][\"states\"].unique()\n",
    "user_transitions = interaction_df.groupby(session_id)[\"states\"].apply(list).tolist()\n",
    "\n",
    "states_dict = dict(zip(reduced_df['states'], reduced_df[cat]))\n",
    "user_transitions = interaction_df.groupby(session_id)['states'].apply(list).tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "6fef4295-a3de-455a-bea6-5a27950ee9e1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:05.170394Z",
     "start_time": "2025-02-17T11:29:05.157419Z"
    }
   },
   "outputs": [],
   "source": [
    "def data_prior(paths, vocab, k, ax):\n",
    "    shape_matrix = (max(vocab.values()) + 1, max(vocab.values()) + 1)\n",
    "    evidences = {}\n",
    "    shape = len(vocab)\n",
    "    for i in k:\n",
    "        if i == 0:\n",
    "            prior = csr_matrix((shape,shape), dtype=np.float64)\n",
    "        else:\n",
    "            # We first calculate the MLE for the data\n",
    "            markov = MarkovChain(k=1, use_prior=False, reset = False, modus=\"mle\")\n",
    "            markov.prepare_data(paths)\n",
    "            markov.fit(paths)\n",
    "            mle = markov.transition_dict_norm_\n",
    "            \n",
    "            # Now we use the parameter configuration from the MLE as a hypothesis\n",
    "            i_indices = list()\n",
    "            j_indices = list()\n",
    "            values = list()\n",
    "            for s,targets in mle.items():\n",
    "                for t, v in targets.items():\n",
    "                    i_indices.append(vocab[s[0]])\n",
    "                    j_indices.append(vocab[t])\n",
    "                    values.append(v)\n",
    "\n",
    "            matrix_data = csr_matrix((values, (i_indices, j_indices)),\n",
    "                             shape=shape_matrix, dtype=np.float64)\n",
    "            chips = i*(matrix_data.shape[0]-1.)\n",
    "            prior = distr_chips_row(matrix_data, chips, n_jobs=1)\n",
    "\n",
    "        markov = MarkovChain(k=1, use_prior=True, reset = False, prior=1., specific_prior=prior,\n",
    "                                specific_prior_vocab = vocab, state_count=len(vocab), modus=\"bayes\")\n",
    "        markov.prepare_data(paths)\n",
    "        markov.fit(paths)\n",
    "\n",
    "        evidence = markov.bayesian_evidence()\n",
    "        evidences[i] = evidence\n",
    "\n",
    "    ax.plot(list(evidences.keys()), list(evidences.values()), marker='o', clip_on = False, label=\"data\", linestyle='--')\n",
    "    return evidences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "0c398aab-bfa6-4d41-8e03-513c6e43e871",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:05.181846Z",
     "start_time": "2025-02-17T11:29:05.172106Z"
    }
   },
   "outputs": [],
   "source": [
    "def uniform_prior(paths, vocab, k, ax):\n",
    "    shape = len(vocab)\n",
    "    evidences = {}\n",
    "    for i in k:\n",
    "        if i == 0:\n",
    "            prior = csr_matrix((shape,shape), dtype=np.float64)\n",
    "        else:\n",
    "            matrix_uniform = lil_matrix((shape,shape), dtype=np.float64)\n",
    "            matrix_uniform[:] = 1.\n",
    "            #matrix_uniform.setdiag(0.)\n",
    "            matrix_uniform = matrix_uniform.tocsr()\n",
    "\n",
    "            chips = i*(matrix_uniform.shape[0]-1.)\n",
    "            prior = distr_chips_row(matrix_uniform, chips, n_jobs=1)\n",
    "        markov = MarkovChain(k=1, use_prior=True, reset = False, prior=1.+i, specific_prior=prior,\n",
    "                                specific_prior_vocab = vocab, modus=\"bayes\")\n",
    "        \n",
    "        markov.prepare_data(paths)\n",
    "        markov.fit(paths)\n",
    "        evidence = markov.bayesian_evidence()\n",
    "        evidences[i] = evidence\n",
    "\n",
    "    ax.plot(list(evidences.keys()), list(evidences.values()), marker='o', clip_on = False, label=\"uniform\", linestyle='-.')\n",
    "    return evidences\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "6dd72fa4-df5b-4d8c-9f91-c9021926b3c2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:05.351790Z",
     "start_time": "2025-02-17T11:29:05.184582Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import normalize\n",
    "def struc_prior(paths, vocab, k, ax):\n",
    "    i_indices = list()\n",
    "    j_indices = list()\n",
    "    values = list()\n",
    "    # Let us iterate through all nodes\n",
    "    for s,v in states_dict.items():\n",
    "        overview_id = vocab[s]\n",
    "        for s2,v2 in states_dict.items():\n",
    "            if v2 == v:\n",
    "                node_id = vocab[s2]\n",
    "                i_indices.append(overview_id)\n",
    "                j_indices.append(node_id)\n",
    "                values.append(1)\n",
    "\n",
    "    shape_matrix = (max(vocab.values()) + 1, max(vocab.values()) + 1)\n",
    "    evidences = {}\n",
    "    shape = len(vocab)\n",
    "    for i in k:\n",
    "        if i == 0:\n",
    "            prior = csr_matrix((shape,shape), dtype=np.float64)\n",
    "        else:\n",
    "            # Creating the matrix with above defined values\n",
    "            matrix_struc = csr_matrix((values, (i_indices, j_indices)),\n",
    "                             shape=shape_matrix, dtype=np.float64)\n",
    "\n",
    "            chips = i*(matrix_struc.shape[0]-1.)\n",
    "            prior = distr_chips_row(matrix_struc, chips, n_jobs=1)\n",
    "        markov = MarkovChain(k=1, use_prior=True, reset = False, prior=1., specific_prior=prior,\n",
    "                                specific_prior_vocab = vocab, state_count=len(vocab), modus=\"bayes\")\n",
    "        markov.prepare_data(paths)\n",
    "        markov.fit(paths)\n",
    "        evidence = markov.bayesian_evidence()\n",
    "        evidences[i] = evidence\n",
    "    \n",
    "    ax.plot(list(evidences.keys()), list(evidences.values()), marker='o', clip_on = False, label=\"structural\", linestyle='--')\n",
    "    return evidences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "530ebcf3-cea2-4794-bc1b-0c7191ded2b2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:29:05.373023Z",
     "start_time": "2025-02-17T11:29:05.354841Z"
    }
   },
   "outputs": [],
   "source": [
    "def uni_struct_prior(paths, vocab, k, ax):\n",
    "\n",
    "    i_indices = list()\n",
    "    j_indices = list()\n",
    "    values = list()\n",
    "    # Let us iterate through all nodes\n",
    "    for s,v in states_dict.items():\n",
    "        overview_id = vocab[s]\n",
    "        for s2,v2 in states_dict.items():\n",
    "            if v2 == v:\n",
    "                node_id = vocab[s2]\n",
    "                i_indices.append(overview_id)\n",
    "                j_indices.append(node_id)\n",
    "                values.append(1)\n",
    "\n",
    "    shape_matrix = (max(vocab.values()) + 1, max(vocab.values()) + 1)\n",
    "\n",
    "    \n",
    "    shape = len(vocab)\n",
    "    evidences = {}\n",
    "    for i in k:\n",
    "        if i == 0:\n",
    "            prior = csr_matrix((shape,shape), dtype=np.float64)\n",
    "        else:\n",
    "            matrix_struc = csr_matrix((values, (i_indices, j_indices)),\n",
    "                 shape=shape_matrix, dtype=np.float64)\n",
    "            matrix_uniform = lil_matrix((shape,shape), dtype=np.float64)\n",
    "            matrix_uniform[:] = 1.\n",
    "            matrix_uniform = matrix_uniform+matrix_struc\n",
    "            #matrix_uniform.setdiag(0.)\n",
    "            matrix_uniform = matrix_uniform.tocsr()\n",
    "\n",
    "            chips = i*(matrix_uniform.shape[0]-1.)\n",
    "            prior = distr_chips_row(matrix_uniform, chips, n_jobs=1)\n",
    "        markov = MarkovChain(k=1, use_prior=True, reset = False, prior=1.+i, specific_prior=prior,\n",
    "                                specific_prior_vocab = vocab, modus=\"bayes\")\n",
    "        \n",
    "        markov.prepare_data(paths)\n",
    "        markov.fit(paths)\n",
    "        evidence = markov.bayesian_evidence()\n",
    "        evidences[i] = evidence\n",
    "\n",
    "    ax.plot(list(evidences.keys()), list(evidences.values()), marker='o', clip_on = False, label=\"uniform+structural\", linestyle='--')\n",
    "    return evidences\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "1134f967-70f3-4a92-b2f3-24bb6683014d",
   "metadata": {
    "scrolled": true,
    "ExecuteTime": {
     "end_time": "2025-02-17T11:30:25.236695Z",
     "start_time": "2025-02-17T11:29:05.376389Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uniform done\n",
      "data done\n",
      "struc done\n",
      "mixture done\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure()\n",
    "\n",
    "\n",
    "plt.rc('font', size=16)  # default text size\n",
    "plt.rc('axes', titlesize=25)  # axes title size\n",
    "plt.rc('axes', labelsize=16)  # x and y label size\n",
    "plt.rc('xtick', labelsize=14)  # x tick label size\n",
    "plt.rc('ytick', labelsize=14)  # y tick label size\n",
    "plt.rc('legend', fontsize=14)  # legend font size\n",
    "plt.rc('figure', titlesize=20)  # figure title size\n",
    "\n",
    "k= [0,1,3,5,10,100, 1000]\n",
    "paths= user_transitions\n",
    "ax = fig.add_subplot(111)\n",
    "ax.set_xscale(\"symlog\")\n",
    "\n",
    "# Now, let us calculate the evidences\n",
    "ret = uniform_prior(paths, vocab, k, ax)\n",
    "print(\"Uniform done\")\n",
    "ret = data_prior(paths, vocab, k, ax)\n",
    "print(\"data done\")\n",
    "ret = struc_prior(paths, vocab, k, ax)\n",
    "print(\"struc done\")\n",
    "ret = uni_struct_prior(paths, vocab, k, ax)\n",
    "print(\"mixture done\")\n",
    "\n",
    "# Further plotting\n",
    "ax.set_xlabel(\"hypothesis weighting factor k\")\n",
    "ax.set_ylabel(\"marginal likelihood / evidence\")\n",
    "\n",
    "plt.legend()\n",
    "plt.grid(False)\n",
    "ax.xaxis.grid(True)\n",
    "plt.tight_layout(pad=0.2)\n",
    "#plt.savefig('./hyptrails-'+fig_name+'.png', format='png', dpi=300, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "4b7f37b6-fe16-4c5b-a875-920902bea845",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-02-17T11:30:25.240535Z",
     "start_time": "2025-02-17T11:30:25.238644Z"
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c62dd813-4541-49e2-a37c-9819ba6bd39c",
   "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": "3.11.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}