examples/multi_armed_bandit.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tempfile\n",
    "import seaborn as sns\n",
    "sns.set(style=\"darkgrid\")\n",
    "\n",
    "import logging\n",
    "logging.getLogger().setLevel(logging.INFO)\n",
    "\n",
    "from banditpylib import trials_to_dataframe\n",
    "from banditpylib.arms import BernoulliArm\n",
    "from banditpylib.bandits import MultiArmedBandit\n",
    "from banditpylib.protocols import SinglePlayerProtocol\n",
    "from banditpylib.learners.mab_learner import UCB, EpsGreedy, ThompsonSampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "horizon = 2000\n",
    "means = np.array([0.3, 0.5, 0.7])\n",
    "arms = [BernoulliArm(mean) for mean in means]\n",
    "bandit = MultiArmedBandit(arms=arms)\n",
    "learners = [EpsGreedy(arm_num=len(arms), name='Epsilon Greedy'), \n",
    "            UCB(arm_num=len(arms), name='UCB'),\n",
    "            ThompsonSampling(arm_num=len(arms), name='Thompson Sampling')]\n",
    "# Report intermediate regrets after these horizons\n",
    "intermediate_horizons = list(range(0, horizon+1, 50))\n",
    "temp_file = tempfile.NamedTemporaryFile()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:absl:start Epsilon Greedy's play with multi_armed_bandit\n",
      "INFO:absl:Epsilon Greedy's play with multi_armed_bandit runs 7.42 seconds.\n",
      "INFO:absl:start UCB's play with multi_armed_bandit\n",
      "INFO:absl:UCB's play with multi_armed_bandit runs 7.21 seconds.\n",
      "INFO:absl:start Thompson Sampling's play with multi_armed_bandit\n",
      "INFO:absl:Thompson Sampling's play with multi_armed_bandit runs 7.16 seconds.\n"
     ]
    }
   ],
   "source": [
    "game = SinglePlayerProtocol(bandit, learners)\n",
    "# Start playing the game\n",
    "# Add `debug=True` for debugging purpose\n",
    "game.play(200, temp_file.name, intermediate_horizons=intermediate_horizons, horizon=horizon)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>rounds</th>\n",
       "      <th>total_actions</th>\n",
       "      <th>regret</th>\n",
       "      <th>other</th>\n",
       "      <th>bandit</th>\n",
       "      <th>learner</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>multi_armed_bandit</td>\n",
       "      <td>Epsilon Greedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>50</td>\n",
       "      <td>50</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>multi_armed_bandit</td>\n",
       "      <td>Epsilon Greedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>100</td>\n",
       "      <td>100</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>multi_armed_bandit</td>\n",
       "      <td>Epsilon Greedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>150</td>\n",
       "      <td>150</td>\n",
       "      <td>-2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>multi_armed_bandit</td>\n",
       "      <td>Epsilon Greedy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>200</td>\n",
       "      <td>200</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>multi_armed_bandit</td>\n",
       "      <td>Epsilon Greedy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   rounds  total_actions  regret  other              bandit         learner\n",
       "0       0              0     0.0    0.0  multi_armed_bandit  Epsilon Greedy\n",
       "1      50             50     3.0    0.0  multi_armed_bandit  Epsilon Greedy\n",
       "2     100            100     3.0    0.0  multi_armed_bandit  Epsilon Greedy\n",
       "3     150            150    -2.0    0.0  multi_armed_bandit  Epsilon Greedy\n",
       "4     200            200     1.0    0.0  multi_armed_bandit  Epsilon Greedy"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "trials_df = trials_to_dataframe(temp_file.name)\n",
    "trials_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='total_actions', ylabel='regret'>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.lineplot(x='total_actions', y='regret', hue='learner', data=trials_df)"
   ]
  }
 ],
 "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}