riesling_challenge.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Riesling Challenge\n",
    "\n",
    "We recently stumbled upon a Riesling tasting box and ended up turning last years drinking challenge into a tradition: We invited a bunch of friends and tried to assign the wines to annonymised descriptions (see [form](./data/riesling_challenge.pdf) in German).\n",
    "\n",
    "![WTF](data/riesling_challenge.jpg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Order</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Stef</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Ste</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Cheese</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Fränzi</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Nesa</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Marc</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Lilä</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Lüku</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Michu</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Geri*</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Order</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "      <th>Rating</th>\n",
       "      <th>Guess</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Barth Hattenheimer Schützenhaus 2016</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>3</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>4</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>3</td>\n",
       "      <td>No Sex</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Barth Fructus 2017</th>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>4</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>4.0</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>4</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Stefan Bietighöfer Reserve</th>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>5</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>4</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>5</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>4</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>4</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>4</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Julian Haart Mosel Riesling</th>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>3</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>2</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>2</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>2</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No Sex</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>4</td>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>1</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>3.0</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>2</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>3</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>2</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>2</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>4</td>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>3</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Steinterrassen Reh</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>4</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>2</td>\n",
       "      <td>No Sex</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>2</td>\n",
       "      <td>Barth Hattenheimer Schützenhaus 2016</td>\n",
       "      <td>5</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>3</td>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     Order   Stef  \\\n",
       "                                     Order Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016     2      3   \n",
       "Barth Fructus 2017                       5      4   \n",
       "Stefan Bietighöfer Reserve               1      4   \n",
       "Julian Haart Mosel Riesling              6      3   \n",
       "No Sex                                   3      3   \n",
       "Steinterrassen Reh                       4      4   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016           Julian Haart Mosel Riesling   \n",
       "Barth Fructus 2017                                                  No Sex   \n",
       "Stefan Bietighöfer Reserve                              Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling                             Steinterrassen Reh   \n",
       "No Sex                                Barth Hattenheimer Schützenhaus 2016   \n",
       "Steinterrassen Reh                              Stefan Bietighöfer Reserve   \n",
       "\n",
       "                                        Ste  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        4   \n",
       "Stefan Bietighöfer Reserve                5   \n",
       "Julian Haart Mosel Riesling               3   \n",
       "No Sex                                    4   \n",
       "Steinterrassen Reh                        3   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                                No Sex   \n",
       "Barth Fructus 2017                             Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                              Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling                     Stefan Bietighöfer Reserve   \n",
       "No Sex                                                  Steinterrassen Reh   \n",
       "Steinterrassen Reh                    Barth Hattenheimer Schützenhaus 2016   \n",
       "\n",
       "                                     Cheese  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      4   \n",
       "Barth Fructus 2017                        3   \n",
       "Stefan Bietighöfer Reserve                3   \n",
       "Julian Haart Mosel Riesling               3   \n",
       "No Sex                                    1   \n",
       "Steinterrassen Reh                        4   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Steinterrassen Reh   \n",
       "Barth Fructus 2017                             Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                              Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling           Barth Hattenheimer Schützenhaus 2016   \n",
       "No Sex                                                              No Sex   \n",
       "Steinterrassen Reh                              Stefan Bietighöfer Reserve   \n",
       "\n",
       "                                     Fränzi  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016    3.0   \n",
       "Barth Fructus 2017                      4.0   \n",
       "Stefan Bietighöfer Reserve              1.0   \n",
       "Julian Haart Mosel Riesling             NaN   \n",
       "No Sex                                  3.0   \n",
       "Steinterrassen Reh                      3.0   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Steinterrassen Reh   \n",
       "Barth Fructus 2017                             Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                              Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling                     Stefan Bietighöfer Reserve   \n",
       "No Sex                                                              No Sex   \n",
       "Steinterrassen Reh                    Barth Hattenheimer Schützenhaus 2016   \n",
       "\n",
       "                                       Nesa  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        3   \n",
       "Stefan Bietighöfer Reserve                4   \n",
       "Julian Haart Mosel Riesling               3   \n",
       "No Sex                                    2   \n",
       "Steinterrassen Reh                        2   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Barth Fructus 2017   \n",
       "Barth Fructus 2017                                      Steinterrassen Reh   \n",
       "Stefan Bietighöfer Reserve                     Julian Haart Mosel Riesling   \n",
       "Julian Haart Mosel Riesling                     Stefan Bietighöfer Reserve   \n",
       "No Sex                                Barth Hattenheimer Schützenhaus 2016   \n",
       "Steinterrassen Reh                                                  No Sex   \n",
       "\n",
       "                                       Marc  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        4   \n",
       "Stefan Bietighöfer Reserve                5   \n",
       "Julian Haart Mosel Riesling               2   \n",
       "No Sex                                    3   \n",
       "Steinterrassen Reh                        3   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Steinterrassen Reh   \n",
       "Barth Fructus 2017                                      Barth Fructus 2017   \n",
       "Stefan Bietighöfer Reserve                                          No Sex   \n",
       "Julian Haart Mosel Riesling           Barth Hattenheimer Schützenhaus 2016   \n",
       "No Sex                                          Stefan Bietighöfer Reserve   \n",
       "Steinterrassen Reh                             Julian Haart Mosel Riesling   \n",
       "\n",
       "                                       Lilä  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        3   \n",
       "Stefan Bietighöfer Reserve                4   \n",
       "Julian Haart Mosel Riesling               2   \n",
       "No Sex                                    2   \n",
       "Steinterrassen Reh                        2   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Steinterrassen Reh   \n",
       "Barth Fructus 2017                             Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                              Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling                     Stefan Bietighöfer Reserve   \n",
       "No Sex                                                              No Sex   \n",
       "Steinterrassen Reh                    Barth Hattenheimer Schützenhaus 2016   \n",
       "\n",
       "                                       Lüku  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      4   \n",
       "Barth Fructus 2017                        4   \n",
       "Stefan Bietighöfer Reserve                4   \n",
       "Julian Haart Mosel Riesling               3   \n",
       "No Sex                                    2   \n",
       "Steinterrassen Reh                        5   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Barth Fructus 2017   \n",
       "Barth Fructus 2017                                                  No Sex   \n",
       "Stefan Bietighöfer Reserve                      Stefan Bietighöfer Reserve   \n",
       "Julian Haart Mosel Riesling                             Steinterrassen Reh   \n",
       "No Sex                                Barth Hattenheimer Schützenhaus 2016   \n",
       "Steinterrassen Reh                             Julian Haart Mosel Riesling   \n",
       "\n",
       "                                      Michu  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        3   \n",
       "Stefan Bietighöfer Reserve                4   \n",
       "Julian Haart Mosel Riesling               2   \n",
       "No Sex                                    4   \n",
       "Steinterrassen Reh                        3   \n",
       "\n",
       "                                                                            \\\n",
       "                                                                     Guess   \n",
       "Barth Hattenheimer Schützenhaus 2016                    Steinterrassen Reh   \n",
       "Barth Fructus 2017                             Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                                          No Sex   \n",
       "Julian Haart Mosel Riesling           Barth Hattenheimer Schützenhaus 2016   \n",
       "No Sex                                                  Barth Fructus 2017   \n",
       "Steinterrassen Reh                              Stefan Bietighöfer Reserve   \n",
       "\n",
       "                                      Geri*  \\\n",
       "                                     Rating   \n",
       "Barth Hattenheimer Schützenhaus 2016      3   \n",
       "Barth Fructus 2017                        4   \n",
       "Stefan Bietighöfer Reserve                4   \n",
       "Julian Haart Mosel Riesling               3   \n",
       "No Sex                                    3   \n",
       "Steinterrassen Reh                        3   \n",
       "\n",
       "                                                                            \n",
       "                                                                     Guess  \n",
       "Barth Hattenheimer Schützenhaus 2016                                No Sex  \n",
       "Barth Fructus 2017                                      Barth Fructus 2017  \n",
       "Stefan Bietighöfer Reserve                              Steinterrassen Reh  \n",
       "Julian Haart Mosel Riesling           Barth Hattenheimer Schützenhaus 2016  \n",
       "No Sex                                          Stefan Bietighöfer Reserve  \n",
       "Steinterrassen Reh                             Julian Haart Mosel Riesling  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "riesling_challenge = pd.read_csv('data/riesling_challenge.csv', index_col=0, header=[0, 1])\n",
    "\n",
    "pd.set_option('display.max_columns', None)\n",
    "riesling_challenge"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(\n",
    "    np.unique(riesling_challenge.xs(\"Guess\", level=1, axis=1).values.flatten())\n",
    ") == len(riesling_challenge.index), \"Typos!\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/germannp/anaconda/lib/python3.6/site-packages/pandas/core/generic.py:2530: PerformanceWarning: dropping on a non-lexsorted multi-index without a level parameter may impact performance.\n",
      "  obj = obj._drop_axis(labels, axis, level=level, errors=errors)\n"
     ]
    }
   ],
   "source": [
    "order = riesling_challenge[\"Order\"]\n",
    "ratings = riesling_challenge.drop(\"Order\", axis=1).xs(\n",
    "    \"Rating\", level=1, axis=1\n",
    ")\n",
    "guesses = riesling_challenge.drop([\"Order\", \"Geri*\"], axis=1).xs(\n",
    "    \"Guess\", level=1, axis=1\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Lüku      1\n",
       "Lilä      1\n",
       "Marc      1\n",
       "Fränzi    1\n",
       "Cheese    1\n",
       "Michu     0\n",
       "Nesa      0\n",
       "Ste       0\n",
       "Stef      0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "guesses.apply(\n",
    "    lambda guess: sum(guess == guess.index)\n",
    ").sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To assess how significantly bad we were, we could ask \"how likely is one to guess at least X wines correctly at random?\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prob. of 1 or more correct at random: 0.63\n",
      "Prob. of 2 or more correct at random: 0.262\n",
      "Prob. of 3 or more correct at random: 0.076\n",
      "Prob. of 4 or more correct at random: 0.022\n",
      "Prob. of 5 or more correct at random: 0.001\n"
     ]
    }
   ],
   "source": [
    "n_samples = 10000\n",
    "n_friends = len(guesses.columns)\n",
    "n_wines = len(guesses.index)\n",
    "samples = np.array(\n",
    "    [\n",
    "        [\n",
    "            np.random.choice(\n",
    "                np.arange(n_wines), n_wines, replace=False\n",
    "            )\n",
    "            == np.arange(n_wines)\n",
    "            for _ in range(n_friends)\n",
    "        ]\n",
    "        for _ in range(n_samples)\n",
    "    ]\n",
    ")\n",
    "\n",
    "for i in range(1, 6):\n",
    "    better = sum(samples.sum(axis=2).flatten() >= i)\n",
    "    prob = better / n_samples / n_friends\n",
    "    print(f\"Prob. of {i} or more correct at random: {np.round(prob, decimals=3)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we did not make one friend drink all the wine alone, but rather took the challenge as a group, the correct question to ask would be \"how likely is one in a group of nine to guess at least one wine correctly at random?\" However, this happens essentially in every sample, so the more interesting question is \"how likely is a group of nine to guess at least five wines correctly?\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.95\n"
     ]
    }
   ],
   "source": [
    "five_or_more = samples.sum(axis=2).sum(axis=1) >= 5\n",
    "prob = sum(five_or_more) / n_samples\n",
    "print(np.round(prob, decimals=2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Given that we were incredibly bad as a group it is interesting to check, whether one description was particularly misleading."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>most guesses</th>\n",
       "      <th># of guesses</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Barth Hattenheimer Schützenhaus 2016</th>\n",
       "      <td>Steinterrassen Reh</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Barth Fructus 2017</th>\n",
       "      <td>Julian Haart Mosel Riesling</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Stefan Bietighöfer Reserve</th>\n",
       "      <td>Barth Fructus 2017</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Julian Haart Mosel Riesling</th>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No Sex</th>\n",
       "      <td>No Sex</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Steinterrassen Reh</th>\n",
       "      <td>Stefan Bietighöfer Reserve</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                     most guesses  \\\n",
       "Barth Hattenheimer Schützenhaus 2016           Steinterrassen Reh   \n",
       "Barth Fructus 2017                    Julian Haart Mosel Riesling   \n",
       "Stefan Bietighöfer Reserve                     Barth Fructus 2017   \n",
       "Julian Haart Mosel Riesling            Stefan Bietighöfer Reserve   \n",
       "No Sex                                                     No Sex   \n",
       "Steinterrassen Reh                     Stefan Bietighöfer Reserve   \n",
       "\n",
       "                                      # of guesses  \n",
       "Barth Hattenheimer Schützenhaus 2016             5  \n",
       "Barth Fructus 2017                               5  \n",
       "Stefan Bietighöfer Reserve                       5  \n",
       "Julian Haart Mosel Riesling                      4  \n",
       "No Sex                                           3  \n",
       "Steinterrassen Reh                               3  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "guesses.apply(\n",
    "    lambda wine: pd.Series(\n",
    "        {\n",
    "            \"most guesses\": wine.value_counts().index[0],\n",
    "            \"# of guesses\": wine.value_counts().iloc[0],\n",
    "        }\n",
    "    ),\n",
    "    axis=1,\n",
    ")[[\"most guesses\", \"# of guesses\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prob. of 2 or more same at random: 1.0\n",
      "Prob. of 3 or more same at random: 1.0\n",
      "Prob. of 4 or more same at random: 0.847\n",
      "Prob. of 5 or more same at random: 0.277\n",
      "Prob. of 6 or more same at random: 0.039\n",
      "Prob. of 7 or more same at random: 0.003\n"
     ]
    }
   ],
   "source": [
    "max_same_guesses = np.array(\n",
    "    [\n",
    "        max(\n",
    "            [\n",
    "                np.unique(col, return_counts=True)[1].max()\n",
    "                for col in np.array(\n",
    "                    [\n",
    "                        np.random.choice(np.arange(n_wines), n_wines, replace=False)\n",
    "                        for _ in range(n_friends)\n",
    "                    ]\n",
    "                ).T\n",
    "            ]\n",
    "        )\n",
    "        for _ in range(n_samples)\n",
    "    ]\n",
    ").flatten()\n",
    "for i in range(2, 8):\n",
    "    same = sum(max_same_guesses >= i)\n",
    "    prob = same / n_samples\n",
    "    print(f\"Prob. of {i} or more same at random: {np.round(prob, decimals=3)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "No Sex                                  3\n",
       "Stefan Bietighöfer Reserve              1\n",
       "Barth Fructus 2017                      1\n",
       "Steinterrassen Reh                      0\n",
       "Julian Haart Mosel Riesling             0\n",
       "Barth Hattenheimer Schützenhaus 2016    0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "guesses.apply(\n",
    "    lambda wine: sum(wine == wine.name), axis=1\n",
    ").sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Similarly, we can compute how likely it is, to guess a wine at least three times correctly as a group at random:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.66\n"
     ]
    }
   ],
   "source": [
    "three_or_better = (samples.sum(axis=1) >= 3).sum(axis=1) >= 1\n",
    "prob = sum(three_or_better) / n_samples\n",
    "print(np.round(prob, decimals=2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>mean</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Stefan Bietighöfer Reserve</th>\n",
       "      <td>3.8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Barth Fructus 2017</th>\n",
       "      <td>3.6</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Barth Hattenheimer Schützenhaus 2016</th>\n",
       "      <td>3.2</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Steinterrassen Reh</th>\n",
       "      <td>3.2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Julian Haart Mosel Riesling</th>\n",
       "      <td>2.7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>No Sex</th>\n",
       "      <td>2.7</td>\n",
       "      <td>1.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      mean  min  max\n",
       "Stefan Bietighöfer Reserve             3.8  1.0  5.0\n",
       "Barth Fructus 2017                     3.6  3.0  4.0\n",
       "Barth Hattenheimer Schützenhaus 2016   3.2  3.0  4.0\n",
       "Steinterrassen Reh                     3.2  2.0  5.0\n",
       "Julian Haart Mosel Riesling            2.7  2.0  3.0\n",
       "No Sex                                 2.7  1.0  4.0"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ratings.apply(\n",
    "    lambda wine: pd.Series(\n",
    "        {\"mean\": np.round(wine.mean(), 1), \"min\": wine.min(), \"max\": wine.max()}\n",
    "    ),\n",
    "    axis=1,\n",
    ").sort_values(\"mean\", ascending=False)[[\"mean\", \"min\", \"max\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x108eabe10>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.stripplot(\n",
    "    y=\"level_0\",\n",
    "    x=0,\n",
    "    hue=\"level_1\",\n",
    "    dodge=True,\n",
    "    order=order.sort_values(\"Order\").index,\n",
    "    data=ratings.stack().reset_index(),\n",
    ")\n",
    "sns.despine()\n",
    "plt.ylabel(\"\")\n",
    "plt.xlabel(\"\")\n",
    "plt.xticks(np.arange(5)+1, [\"★\" * i for i in range(1, 6)])\n",
    "plt.legend([], frameon=False)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [default]",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}