{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "9e43247d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0ab32394",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(200, 9)\n"
     ]
    },
    {
     "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>config_id</th>\n",
       "      <th>persona_id</th>\n",
       "      <th>persona</th>\n",
       "      <th>filters</th>\n",
       "      <th>context</th>\n",
       "      <th>city</th>\n",
       "      <th>llama_query_v</th>\n",
       "      <th>llama_query_p0</th>\n",
       "      <th>llama_query_p1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_p_0_pop_low_easy</td>\n",
       "      <td>p_0</td>\n",
       "      <td>A top-scoring player in the local league who i...</td>\n",
       "      <td>{'popularity': 'low', 'month': 'February'}</td>\n",
       "      <td>Adana has low popularity. Adana has low season...</td>\n",
       "      <td>['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...</td>\n",
       "      <td>\"Less crowded European cities to visit in Febr...</td>\n",
       "      <td>\"European cities with ice hockey facilities, l...</td>\n",
       "      <td>Which European cities have ice hockey arenas a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_p_1_pop_medium_medium</td>\n",
       "      <td>p_1</td>\n",
       "      <td>A former DJ at WSUM who is now working as a mu...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'medium', '...</td>\n",
       "      <td>Coimbra has medium popularity and medium budge...</td>\n",
       "      <td>['Coimbra', 'Brno', 'Braga']</td>\n",
       "      <td>'medium budget European city breaks with parks...</td>\n",
       "      <td>\"Medium budget European cities with parks and ...</td>\n",
       "      <td>Which European cities offer a mix of music fes...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_p_2_pop_high_hard</td>\n",
       "      <td>p_2</td>\n",
       "      <td>A fellow agent-turned-author who shares the sa...</td>\n",
       "      <td>{'popularity': 'high', 'budget': 'low', 'inter...</td>\n",
       "      <td>Zagreb has high popularity and low budget. Zag...</td>\n",
       "      <td>['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...</td>\n",
       "      <td>\"Looking for a popular and affordable European...</td>\n",
       "      <td>\"Low-budget European cities with museums and n...</td>\n",
       "      <td>Which European cities offer a mix of historica...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_p_3_pop_low_sustainable</td>\n",
       "      <td>p_3</td>\n",
       "      <td>a film critic who dislikes storylines involvin...</td>\n",
       "      <td>{'popularity': 'low', 'interests': 'Outdoors &amp;...</td>\n",
       "      <td>Van has low popularity. Van has low season in ...</td>\n",
       "      <td>['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...</td>\n",
       "      <td>\"European cities with low popularity, monaster...</td>\n",
       "      <td>\"off the beaten path European city breaks in l...</td>\n",
       "      <td>Which European cities offer a culturally rich ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_p_4_pop_medium_easy</td>\n",
       "      <td>p_4</td>\n",
       "      <td>A biology major conducting research on equine ...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'high'}</td>\n",
       "      <td>Aalborg has medium popularity and high budget....</td>\n",
       "      <td>['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...</td>\n",
       "      <td>\"European cities for a luxurious trip.\"</td>\n",
       "      <td>\"European cities with horse riding trails and ...</td>\n",
       "      <td>Which European cities have horse riding school...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   config_id persona_id  \\\n",
       "0         c_p_0_pop_low_easy        p_0   \n",
       "1    c_p_1_pop_medium_medium        p_1   \n",
       "2        c_p_2_pop_high_hard        p_2   \n",
       "3  c_p_3_pop_low_sustainable        p_3   \n",
       "4      c_p_4_pop_medium_easy        p_4   \n",
       "\n",
       "                                             persona  \\\n",
       "0  A top-scoring player in the local league who i...   \n",
       "1  A former DJ at WSUM who is now working as a mu...   \n",
       "2  A fellow agent-turned-author who shares the sa...   \n",
       "3  a film critic who dislikes storylines involvin...   \n",
       "4  A biology major conducting research on equine ...   \n",
       "\n",
       "                                             filters  \\\n",
       "0         {'popularity': 'low', 'month': 'February'}   \n",
       "1  {'popularity': 'medium', 'budget': 'medium', '...   \n",
       "2  {'popularity': 'high', 'budget': 'low', 'inter...   \n",
       "3  {'popularity': 'low', 'interests': 'Outdoors &...   \n",
       "4         {'popularity': 'medium', 'budget': 'high'}   \n",
       "\n",
       "                                             context  \\\n",
       "0  Adana has low popularity. Adana has low season...   \n",
       "1  Coimbra has medium popularity and medium budge...   \n",
       "2  Zagreb has high popularity and low budget. Zag...   \n",
       "3  Van has low popularity. Van has low season in ...   \n",
       "4  Aalborg has medium popularity and high budget....   \n",
       "\n",
       "                                                city  \\\n",
       "0  ['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...   \n",
       "1                       ['Coimbra', 'Brno', 'Braga']   \n",
       "2  ['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...   \n",
       "3  ['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...   \n",
       "4  ['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...   \n",
       "\n",
       "                                       llama_query_v  \\\n",
       "0  \"Less crowded European cities to visit in Febr...   \n",
       "1  'medium budget European city breaks with parks...   \n",
       "2  \"Looking for a popular and affordable European...   \n",
       "3  \"European cities with low popularity, monaster...   \n",
       "4            \"European cities for a luxurious trip.\"   \n",
       "\n",
       "                                      llama_query_p0  \\\n",
       "0  \"European cities with ice hockey facilities, l...   \n",
       "1  \"Medium budget European cities with parks and ...   \n",
       "2  \"Low-budget European cities with museums and n...   \n",
       "3  \"off the beaten path European city breaks in l...   \n",
       "4  \"European cities with horse riding trails and ...   \n",
       "\n",
       "                                      llama_query_p1  \n",
       "0  Which European cities have ice hockey arenas a...  \n",
       "1  Which European cities offer a mix of music fes...  \n",
       "2  Which European cities offer a mix of historica...  \n",
       "3  Which European cities offer a culturally rich ...  \n",
       "4  Which European cities have horse riding school...  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "llama = pd.read_csv(\"../../data/llama_results_subset.csv\")\n",
    "# llama[\"model\"] = \"llama-3.2-90b\"\n",
    "\n",
    "llama.rename(columns={'query_v': \"llama_query_v\", 'query_p0': \"llama_query_p0\", \"query_p1\": \"llama_query_p1\"}, inplace = True)\n",
    "\n",
    "print(llama.shape)\n",
    "llama.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "aed0fba7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(200, 9)\n"
     ]
    },
    {
     "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>config_id</th>\n",
       "      <th>persona_id</th>\n",
       "      <th>persona</th>\n",
       "      <th>filters</th>\n",
       "      <th>context</th>\n",
       "      <th>city</th>\n",
       "      <th>gemini_query_v</th>\n",
       "      <th>gemini_query_p0</th>\n",
       "      <th>gemini_query_p1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_p_0_pop_low_easy</td>\n",
       "      <td>p_0</td>\n",
       "      <td>A top-scoring player in the local league who i...</td>\n",
       "      <td>{'popularity': 'low', 'month': 'February'}</td>\n",
       "      <td>Adana has low popularity. Adana has low season...</td>\n",
       "      <td>['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...</td>\n",
       "      <td>Cheap European city break in February.\\n</td>\n",
       "      <td>European city break in February, less crowded ...</td>\n",
       "      <td>Best European cities for intense physical trai...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_p_1_pop_medium_medium</td>\n",
       "      <td>p_1</td>\n",
       "      <td>A former DJ at WSUM who is now working as a mu...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'medium', '...</td>\n",
       "      <td>Coimbra has medium popularity and medium budge...</td>\n",
       "      <td>['Coimbra', 'Brno', 'Braga']</td>\n",
       "      <td>Medium budget European city break with parks a...</td>\n",
       "      <td>European city break, medium budget, good parks...</td>\n",
       "      <td>Best European cities for live music, especiall...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_p_2_pop_high_hard</td>\n",
       "      <td>p_2</td>\n",
       "      <td>A fellow agent-turned-author who shares the sa...</td>\n",
       "      <td>{'popularity': 'high', 'budget': 'low', 'inter...</td>\n",
       "      <td>Zagreb has high popularity and low budget. Zag...</td>\n",
       "      <td>['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...</td>\n",
       "      <td>Cheap January city break in Europe with museum...</td>\n",
       "      <td>European city break in January. Budget friendl...</td>\n",
       "      <td>Where can I find inspiring European cities wit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_p_3_pop_low_sustainable</td>\n",
       "      <td>p_3</td>\n",
       "      <td>a film critic who dislikes storylines involvin...</td>\n",
       "      <td>{'popularity': 'low', 'interests': 'Outdoors &amp;...</td>\n",
       "      <td>Van has low popularity. Van has low season in ...</td>\n",
       "      <td>['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...</td>\n",
       "      <td>Cheap European city break in January with inte...</td>\n",
       "      <td>European city break in January. Low-budget des...</td>\n",
       "      <td>Best European cities for unique, artistic expe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_p_4_pop_medium_easy</td>\n",
       "      <td>p_4</td>\n",
       "      <td>A biology major conducting research on equine ...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'high'}</td>\n",
       "      <td>Aalborg has medium popularity and high budget....</td>\n",
       "      <td>['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...</td>\n",
       "      <td>Suggest some moderately popular European citie...</td>\n",
       "      <td>High-budget European city with equestrian attr...</td>\n",
       "      <td>European cities with renowned veterinary or ag...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   config_id persona_id  \\\n",
       "0         c_p_0_pop_low_easy        p_0   \n",
       "1    c_p_1_pop_medium_medium        p_1   \n",
       "2        c_p_2_pop_high_hard        p_2   \n",
       "3  c_p_3_pop_low_sustainable        p_3   \n",
       "4      c_p_4_pop_medium_easy        p_4   \n",
       "\n",
       "                                             persona  \\\n",
       "0  A top-scoring player in the local league who i...   \n",
       "1  A former DJ at WSUM who is now working as a mu...   \n",
       "2  A fellow agent-turned-author who shares the sa...   \n",
       "3  a film critic who dislikes storylines involvin...   \n",
       "4  A biology major conducting research on equine ...   \n",
       "\n",
       "                                             filters  \\\n",
       "0         {'popularity': 'low', 'month': 'February'}   \n",
       "1  {'popularity': 'medium', 'budget': 'medium', '...   \n",
       "2  {'popularity': 'high', 'budget': 'low', 'inter...   \n",
       "3  {'popularity': 'low', 'interests': 'Outdoors &...   \n",
       "4         {'popularity': 'medium', 'budget': 'high'}   \n",
       "\n",
       "                                             context  \\\n",
       "0  Adana has low popularity. Adana has low season...   \n",
       "1  Coimbra has medium popularity and medium budge...   \n",
       "2  Zagreb has high popularity and low budget. Zag...   \n",
       "3  Van has low popularity. Van has low season in ...   \n",
       "4  Aalborg has medium popularity and high budget....   \n",
       "\n",
       "                                                city  \\\n",
       "0  ['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...   \n",
       "1                       ['Coimbra', 'Brno', 'Braga']   \n",
       "2  ['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...   \n",
       "3  ['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...   \n",
       "4  ['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...   \n",
       "\n",
       "                                      gemini_query_v  \\\n",
       "0           Cheap European city break in February.\\n   \n",
       "1  Medium budget European city break with parks a...   \n",
       "2  Cheap January city break in Europe with museum...   \n",
       "3  Cheap European city break in January with inte...   \n",
       "4  Suggest some moderately popular European citie...   \n",
       "\n",
       "                                     gemini_query_p0  \\\n",
       "0  European city break in February, less crowded ...   \n",
       "1  European city break, medium budget, good parks...   \n",
       "2  European city break in January. Budget friendl...   \n",
       "3  European city break in January. Low-budget des...   \n",
       "4  High-budget European city with equestrian attr...   \n",
       "\n",
       "                                     gemini_query_p1  \n",
       "0  Best European cities for intense physical trai...  \n",
       "1  Best European cities for live music, especiall...  \n",
       "2  Where can I find inspiring European cities wit...  \n",
       "3  Best European cities for unique, artistic expe...  \n",
       "4  European cities with renowned veterinary or ag...  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gemini = pd.read_csv(\"../../data/gemini_results_subset.csv\")\n",
    "# gemini[\"model\"] = \"gemini-1.5-pro-002\"\n",
    "gemini.rename(columns={'query_v': \"gemini_query_v\", 'query_p0': \"gemini_query_p0\", \"query_p1\": \"gemini_query_p1\"}, inplace = True)\n",
    "\n",
    "print(gemini.shape)\n",
    "gemini.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "7a0bbd7e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['config_id', 'persona_id', 'filters', 'persona', 'city', 'context']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "common_cols = list(set(list(gemini)).intersection(set(list(llama))))\n",
    "print(len(common_cols))\n",
    "common_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ca123dd4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(200, 12)\n"
     ]
    },
    {
     "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>config_id</th>\n",
       "      <th>persona_id</th>\n",
       "      <th>persona</th>\n",
       "      <th>filters</th>\n",
       "      <th>context</th>\n",
       "      <th>city</th>\n",
       "      <th>gemini_query_v</th>\n",
       "      <th>gemini_query_p0</th>\n",
       "      <th>gemini_query_p1</th>\n",
       "      <th>llama_query_v</th>\n",
       "      <th>llama_query_p0</th>\n",
       "      <th>llama_query_p1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>c_p_0_pop_low_easy</td>\n",
       "      <td>p_0</td>\n",
       "      <td>A top-scoring player in the local league who i...</td>\n",
       "      <td>{'popularity': 'low', 'month': 'February'}</td>\n",
       "      <td>Adana has low popularity. Adana has low season...</td>\n",
       "      <td>['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...</td>\n",
       "      <td>Cheap European city break in February.\\n</td>\n",
       "      <td>European city break in February, less crowded ...</td>\n",
       "      <td>Best European cities for intense physical trai...</td>\n",
       "      <td>\"Less crowded European cities to visit in Febr...</td>\n",
       "      <td>\"European cities with ice hockey facilities, l...</td>\n",
       "      <td>Which European cities have ice hockey arenas a...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>c_p_1_pop_medium_medium</td>\n",
       "      <td>p_1</td>\n",
       "      <td>A former DJ at WSUM who is now working as a mu...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'medium', '...</td>\n",
       "      <td>Coimbra has medium popularity and medium budge...</td>\n",
       "      <td>['Coimbra', 'Brno', 'Braga']</td>\n",
       "      <td>Medium budget European city break with parks a...</td>\n",
       "      <td>European city break, medium budget, good parks...</td>\n",
       "      <td>Best European cities for live music, especiall...</td>\n",
       "      <td>'medium budget European city breaks with parks...</td>\n",
       "      <td>\"Medium budget European cities with parks and ...</td>\n",
       "      <td>Which European cities offer a mix of music fes...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c_p_2_pop_high_hard</td>\n",
       "      <td>p_2</td>\n",
       "      <td>A fellow agent-turned-author who shares the sa...</td>\n",
       "      <td>{'popularity': 'high', 'budget': 'low', 'inter...</td>\n",
       "      <td>Zagreb has high popularity and low budget. Zag...</td>\n",
       "      <td>['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...</td>\n",
       "      <td>Cheap January city break in Europe with museum...</td>\n",
       "      <td>European city break in January. Budget friendl...</td>\n",
       "      <td>Where can I find inspiring European cities wit...</td>\n",
       "      <td>\"Looking for a popular and affordable European...</td>\n",
       "      <td>\"Low-budget European cities with museums and n...</td>\n",
       "      <td>Which European cities offer a mix of historica...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>c_p_3_pop_low_sustainable</td>\n",
       "      <td>p_3</td>\n",
       "      <td>a film critic who dislikes storylines involvin...</td>\n",
       "      <td>{'popularity': 'low', 'interests': 'Outdoors &amp;...</td>\n",
       "      <td>Van has low popularity. Van has low season in ...</td>\n",
       "      <td>['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...</td>\n",
       "      <td>Cheap European city break in January with inte...</td>\n",
       "      <td>European city break in January. Low-budget des...</td>\n",
       "      <td>Best European cities for unique, artistic expe...</td>\n",
       "      <td>\"European cities with low popularity, monaster...</td>\n",
       "      <td>\"off the beaten path European city breaks in l...</td>\n",
       "      <td>Which European cities offer a culturally rich ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>c_p_4_pop_medium_easy</td>\n",
       "      <td>p_4</td>\n",
       "      <td>A biology major conducting research on equine ...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'high'}</td>\n",
       "      <td>Aalborg has medium popularity and high budget....</td>\n",
       "      <td>['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...</td>\n",
       "      <td>Suggest some moderately popular European citie...</td>\n",
       "      <td>High-budget European city with equestrian attr...</td>\n",
       "      <td>European cities with renowned veterinary or ag...</td>\n",
       "      <td>\"European cities for a luxurious trip.\"</td>\n",
       "      <td>\"European cities with horse riding trails and ...</td>\n",
       "      <td>Which European cities have horse riding school...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195</th>\n",
       "      <td>c_p_8_pop_medium_easy</td>\n",
       "      <td>p_8</td>\n",
       "      <td>A junior Flash developer seeking guidance on c...</td>\n",
       "      <td>{'popularity': 'medium', 'interests': 'Outdoor...</td>\n",
       "      <td>Zaragoza has medium popularity. In Zaragoza yo...</td>\n",
       "      <td>['Zaragoza', 'Varna', 'Turku', 'Tampere', 'Szc...</td>\n",
       "      <td>Good places to swim outdoors in Europe\\n</td>\n",
       "      <td>Good places for a Flash developer to visit in ...</td>\n",
       "      <td>Best European cities for tech meetups &amp; cowork...</td>\n",
       "      <td>\"European cities with medium popularity for ou...</td>\n",
       "      <td>A junior Flash developer seeking guidance on c...</td>\n",
       "      <td>Which European cities offer a mix of cultural ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>c_p_9_pop_high_medium</td>\n",
       "      <td>p_9</td>\n",
       "      <td>An avid gamer who has played Sins of a Solar E...</td>\n",
       "      <td>{'popularity': 'high', 'month': 'April', 'budg...</td>\n",
       "      <td>Sarajevo has high popularity and medium budget...</td>\n",
       "      <td>['Sarajevo', 'Tallinn', 'Vilnius', 'Belgrade',...</td>\n",
       "      <td>Medium budget European city break in April wit...</td>\n",
       "      <td>Budget-friendly European city break in April w...</td>\n",
       "      <td>European cities with a grand, futuristic feel ...</td>\n",
       "      <td>'medium budget trip in April to a popular Euro...</td>\n",
       "      <td>\"Medium budget trip to a popular European city...</td>\n",
       "      <td>Which European cities have a rich history of e...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>197</th>\n",
       "      <td>c_p_10_pop_low_hard</td>\n",
       "      <td>p_10</td>\n",
       "      <td>An atheist, philosophy lecturer who encourages...</td>\n",
       "      <td>{'popularity': 'low', 'interests': 'Outdoors &amp;...</td>\n",
       "      <td>Malatya has low popularity and high budget. Ma...</td>\n",
       "      <td>['Malatya', 'Ioannina']</td>\n",
       "      <td>High-budget European city trip in February wit...</td>\n",
       "      <td>European city break in February. Low season &amp; ...</td>\n",
       "      <td>Where can I find European cities rich in histo...</td>\n",
       "      <td>\"Less crowded European destinations with water...</td>\n",
       "      <td>\"Less crowded European destinations for a pric...</td>\n",
       "      <td>European cities with rich history of philosoph...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>c_p_11_pop_medium_sustainable</td>\n",
       "      <td>p_11</td>\n",
       "      <td>A young apprentice fascinated by the technolog...</td>\n",
       "      <td>{'popularity': 'medium', 'budget': 'low', 'mon...</td>\n",
       "      <td>Chelyabinsk has medium popularity , low budget...</td>\n",
       "      <td>['Chelyabinsk', 'Kirov', 'Podgorica', 'Rijeka'...</td>\n",
       "      <td>Suggest a low-budget, walkable European city b...</td>\n",
       "      <td>Budget-friendly European city break in April, ...</td>\n",
       "      <td>Which European cities offer glimpses into the ...</td>\n",
       "      <td>\"European cities with great walkability and lo...</td>\n",
       "      <td>\"European cities with industrial heritage site...</td>\n",
       "      <td>Which European cities showcase historic indust...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>199</th>\n",
       "      <td>c_p_12_pop_high_easy</td>\n",
       "      <td>p_12</td>\n",
       "      <td>A high school guidance counselor in Winnebago ...</td>\n",
       "      <td>{'popularity': 'high', 'budget': 'medium'}</td>\n",
       "      <td>Belgrade has high popularity and medium budget...</td>\n",
       "      <td>['Belgrade', 'Bratislava', 'Budapest', 'Nicosi...</td>\n",
       "      <td>Popular and affordable European city break des...</td>\n",
       "      <td>Affordable, popular European capitals for a we...</td>\n",
       "      <td>Affordable, safe European cities with historic...</td>\n",
       "      <td>'medium budget european city breaks in popular...</td>\n",
       "      <td>\"Medium budget city breaks in Europe for a sch...</td>\n",
       "      <td>Which European cities offer a mix of cultural ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>200 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                         config_id persona_id  \\\n",
       "0               c_p_0_pop_low_easy        p_0   \n",
       "1          c_p_1_pop_medium_medium        p_1   \n",
       "2              c_p_2_pop_high_hard        p_2   \n",
       "3        c_p_3_pop_low_sustainable        p_3   \n",
       "4            c_p_4_pop_medium_easy        p_4   \n",
       "..                             ...        ...   \n",
       "195          c_p_8_pop_medium_easy        p_8   \n",
       "196          c_p_9_pop_high_medium        p_9   \n",
       "197            c_p_10_pop_low_hard       p_10   \n",
       "198  c_p_11_pop_medium_sustainable       p_11   \n",
       "199           c_p_12_pop_high_easy       p_12   \n",
       "\n",
       "                                               persona  \\\n",
       "0    A top-scoring player in the local league who i...   \n",
       "1    A former DJ at WSUM who is now working as a mu...   \n",
       "2    A fellow agent-turned-author who shares the sa...   \n",
       "3    a film critic who dislikes storylines involvin...   \n",
       "4    A biology major conducting research on equine ...   \n",
       "..                                                 ...   \n",
       "195  A junior Flash developer seeking guidance on c...   \n",
       "196  An avid gamer who has played Sins of a Solar E...   \n",
       "197  An atheist, philosophy lecturer who encourages...   \n",
       "198  A young apprentice fascinated by the technolog...   \n",
       "199  A high school guidance counselor in Winnebago ...   \n",
       "\n",
       "                                               filters  \\\n",
       "0           {'popularity': 'low', 'month': 'February'}   \n",
       "1    {'popularity': 'medium', 'budget': 'medium', '...   \n",
       "2    {'popularity': 'high', 'budget': 'low', 'inter...   \n",
       "3    {'popularity': 'low', 'interests': 'Outdoors &...   \n",
       "4           {'popularity': 'medium', 'budget': 'high'}   \n",
       "..                                                 ...   \n",
       "195  {'popularity': 'medium', 'interests': 'Outdoor...   \n",
       "196  {'popularity': 'high', 'month': 'April', 'budg...   \n",
       "197  {'popularity': 'low', 'interests': 'Outdoors &...   \n",
       "198  {'popularity': 'medium', 'budget': 'low', 'mon...   \n",
       "199         {'popularity': 'high', 'budget': 'medium'}   \n",
       "\n",
       "                                               context  \\\n",
       "0    Adana has low popularity. Adana has low season...   \n",
       "1    Coimbra has medium popularity and medium budge...   \n",
       "2    Zagreb has high popularity and low budget. Zag...   \n",
       "3    Van has low popularity. Van has low season in ...   \n",
       "4    Aalborg has medium popularity and high budget....   \n",
       "..                                                 ...   \n",
       "195  Zaragoza has medium popularity. In Zaragoza yo...   \n",
       "196  Sarajevo has high popularity and medium budget...   \n",
       "197  Malatya has low popularity and high budget. Ma...   \n",
       "198  Chelyabinsk has medium popularity , low budget...   \n",
       "199  Belgrade has high popularity and medium budget...   \n",
       "\n",
       "                                                  city  \\\n",
       "0    ['Adana', 'Adiyaman', 'Agri', 'Arad', 'Arkhang...   \n",
       "1                         ['Coimbra', 'Brno', 'Braga']   \n",
       "2    ['Zagreb', 'Volgograd', 'Tirana', 'Tbilisi', '...   \n",
       "3    ['Van', 'Uzhhorod', 'Trabzon', 'Thessaloniki',...   \n",
       "4    ['Aalborg', 'Astrakhan', 'Bari', 'Bremen', 'Ch...   \n",
       "..                                                 ...   \n",
       "195  ['Zaragoza', 'Varna', 'Turku', 'Tampere', 'Szc...   \n",
       "196  ['Sarajevo', 'Tallinn', 'Vilnius', 'Belgrade',...   \n",
       "197                            ['Malatya', 'Ioannina']   \n",
       "198  ['Chelyabinsk', 'Kirov', 'Podgorica', 'Rijeka'...   \n",
       "199  ['Belgrade', 'Bratislava', 'Budapest', 'Nicosi...   \n",
       "\n",
       "                                        gemini_query_v  \\\n",
       "0             Cheap European city break in February.\\n   \n",
       "1    Medium budget European city break with parks a...   \n",
       "2    Cheap January city break in Europe with museum...   \n",
       "3    Cheap European city break in January with inte...   \n",
       "4    Suggest some moderately popular European citie...   \n",
       "..                                                 ...   \n",
       "195           Good places to swim outdoors in Europe\\n   \n",
       "196  Medium budget European city break in April wit...   \n",
       "197  High-budget European city trip in February wit...   \n",
       "198  Suggest a low-budget, walkable European city b...   \n",
       "199  Popular and affordable European city break des...   \n",
       "\n",
       "                                       gemini_query_p0  \\\n",
       "0    European city break in February, less crowded ...   \n",
       "1    European city break, medium budget, good parks...   \n",
       "2    European city break in January. Budget friendl...   \n",
       "3    European city break in January. Low-budget des...   \n",
       "4    High-budget European city with equestrian attr...   \n",
       "..                                                 ...   \n",
       "195  Good places for a Flash developer to visit in ...   \n",
       "196  Budget-friendly European city break in April w...   \n",
       "197  European city break in February. Low season & ...   \n",
       "198  Budget-friendly European city break in April, ...   \n",
       "199  Affordable, popular European capitals for a we...   \n",
       "\n",
       "                                       gemini_query_p1  \\\n",
       "0    Best European cities for intense physical trai...   \n",
       "1    Best European cities for live music, especiall...   \n",
       "2    Where can I find inspiring European cities wit...   \n",
       "3    Best European cities for unique, artistic expe...   \n",
       "4    European cities with renowned veterinary or ag...   \n",
       "..                                                 ...   \n",
       "195  Best European cities for tech meetups & cowork...   \n",
       "196  European cities with a grand, futuristic feel ...   \n",
       "197  Where can I find European cities rich in histo...   \n",
       "198  Which European cities offer glimpses into the ...   \n",
       "199  Affordable, safe European cities with historic...   \n",
       "\n",
       "                                         llama_query_v  \\\n",
       "0    \"Less crowded European cities to visit in Febr...   \n",
       "1    'medium budget European city breaks with parks...   \n",
       "2    \"Looking for a popular and affordable European...   \n",
       "3    \"European cities with low popularity, monaster...   \n",
       "4              \"European cities for a luxurious trip.\"   \n",
       "..                                                 ...   \n",
       "195  \"European cities with medium popularity for ou...   \n",
       "196  'medium budget trip in April to a popular Euro...   \n",
       "197  \"Less crowded European destinations with water...   \n",
       "198  \"European cities with great walkability and lo...   \n",
       "199  'medium budget european city breaks in popular...   \n",
       "\n",
       "                                        llama_query_p0  \\\n",
       "0    \"European cities with ice hockey facilities, l...   \n",
       "1    \"Medium budget European cities with parks and ...   \n",
       "2    \"Low-budget European cities with museums and n...   \n",
       "3    \"off the beaten path European city breaks in l...   \n",
       "4    \"European cities with horse riding trails and ...   \n",
       "..                                                 ...   \n",
       "195  A junior Flash developer seeking guidance on c...   \n",
       "196  \"Medium budget trip to a popular European city...   \n",
       "197  \"Less crowded European destinations for a pric...   \n",
       "198  \"European cities with industrial heritage site...   \n",
       "199  \"Medium budget city breaks in Europe for a sch...   \n",
       "\n",
       "                                        llama_query_p1  \n",
       "0    Which European cities have ice hockey arenas a...  \n",
       "1    Which European cities offer a mix of music fes...  \n",
       "2    Which European cities offer a mix of historica...  \n",
       "3    Which European cities offer a culturally rich ...  \n",
       "4    Which European cities have horse riding school...  \n",
       "..                                                 ...  \n",
       "195  Which European cities offer a mix of cultural ...  \n",
       "196  Which European cities have a rich history of e...  \n",
       "197  European cities with rich history of philosoph...  \n",
       "198  Which European cities showcase historic indust...  \n",
       "199  Which European cities offer a mix of cultural ...  \n",
       "\n",
       "[200 rows x 12 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged = pd.merge(gemini, llama, on=common_cols, how=\"right\")\n",
    "print(merged.shape)\n",
    "merged"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "53314432",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "config_id          0\n",
       "persona_id         0\n",
       "persona            0\n",
       "filters            0\n",
       "context            0\n",
       "city               0\n",
       "gemini_query_v     0\n",
       "gemini_query_p0    0\n",
       "gemini_query_p1    0\n",
       "llama_query_v      0\n",
       "llama_query_p0     0\n",
       "llama_query_p1     0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "merged.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "b4e05bdc",
   "metadata": {},
   "outputs": [],
   "source": [
    "merged.to_csv(\"../../data/user-evaluation/merged.csv\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7abcac32",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".crs-venv",
   "language": "python",
   "name": ".crs-venv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}