2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Читаем данные:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Intitializing Scala interpreter ..."
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "Spark Web UI available at http://e14c60f0b525:4040\n",
       "SparkContext available as 'sc' (version = 3.0.1, master = local[*], app id = local-1604873783815)\n",
       "SparkSession available as 'spark'\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "import org.apache.spark.sql._\n",
       "import org.apache.spark.sql.functions._\n",
       "import org.apache.spark.sql.expressions._\n",
       "spark: org.apache.spark.sql.SparkSession = org.apache.spark.sql.SparkSession@3ef137e2\n",
       "recipes: org.apache.spark.sql.DataFrame = [name: string, id: int ... 10 more fields]\n",
       "interactions: org.apache.spark.sql.DataFrame = [user_id: int, recipe_id: int ... 3 more fields]\n"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import org.apache.spark.sql._\n",
    "import org.apache.spark.sql.functions._\n",
    "import org.apache.spark.sql.expressions._\n",
    "\n",
    "val spark = org.apache.spark.sql.SparkSession.builder\n",
    "        .master(\"local\")\n",
    "        .appName(\"Spark CSV Reader\")\n",
    "        .getOrCreate;\n",
    "\n",
    " val recipes = spark.read\n",
    "        .format(\"csv\")\n",
    "        .option(\"header\", \"true\")\n",
    "        .option(\"inferSchema\", \"true\")\n",
    "        .load(\"recipes.csv\")\n",
    "\n",
    " val interactions = spark.read\n",
    "        .format(\"csv\")\n",
    "        .option(\"header\", \"true\")\n",
    "        .option(\"inferSchema\", \"true\")\n",
    "        .load(\"interactions.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------+\n",
      "|count(DISTINCT id)|\n",
      "+------------------+\n",
      "|            231637|\n",
      "+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "recipes.selectExpr(\"count(distinct id)\").show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "0. Посмотрите на данные"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "recipes.printSchema; interactions.printSchema"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Итак `recipes` - состоит из 12 колонок, `interactions` -  из 5. Кажется данные взяты с какого-то кулираного сайта, где пользователи смотрят рецепты и комментируют"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. Найдите пользователя с максимальным количеством отзывов"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "interactions\n",
    "    .groupBy(\"user_id\")\n",
    "    .agg(count(\"*\") as \"feedbacks\")\n",
    "    .orderBy(-$\"feedbacks\")\n",
    "    .take(1)\n",
    "\n",
    "// ответ: user_id 424680"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**ANS: 424680**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "2. Найдите пользователя, который потратил больше всех времени на готовку (придумайте как это примерно посчитать с имеющимися данными)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-----------+-----------+\n",
      "|user_id|cnt_recipie|minutes_sum|\n",
      "+-------+-----------+-----------+\n",
      "| 424680|       7671| 2148367661|\n",
      "| 222564|       2073| 2148016493|\n",
      "| 482376|       1719| 2147619603|\n",
      "| 573325|       1207| 2147569146|\n",
      "| 237123|        462| 2147556014|\n",
      "| 315565|       1015| 2147551999|\n",
      "| 213139|        857| 2147550635|\n",
      "| 228458|        622| 2147524546|\n",
      "| 407007|        453| 2147511110|\n",
      "| 385678|        336| 2147501802|\n",
      "+-------+-----------+-----------+\n",
      "only showing top 10 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "interactions\n",
    "    .select(\"user_id\", \"recipe_id\")\n",
    "    .join(recipes, interactions(\"recipe_id\") === recipes(\"id\"), \"inner\")\n",
    "    .groupBy(\"user_id\")\n",
    "    .agg(\n",
    "        count(\"*\") as \"cnt_recipie\",\n",
    "        sum(\"minutes\") as \"minutes_sum\"\n",
    "    )\n",
    "    .orderBy(-$\"minutes_sum\")\n",
    "    .show(10)\n",
    "\n",
    "//Ans: 424680"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "res180: Long = 7671\n"
      ]
     },
     "execution_count": 213,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "interactions\n",
    "    .join(recipes, interactions(\"recipe_id\") === recipes(\"id\"), \"inner\")\n",
    "    .filter($\"user_id\" === 424680)\n",
    "    .select(\"user_id\", \"minutes\", \"recipe_id\")\n",
    "    .count"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**ANS: 424680**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "3. Найдите пользователя, который дольше всего пользуется сайтом (придумайте как это примерно посчитать с имеющимися данными)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+\n",
      "|      date|\n",
      "+----------+\n",
      "|2003-02-17|\n",
      "|2011-12-21|\n",
      "|2002-12-01|\n",
      "|2010-02-27|\n",
      "|2011-10-01|\n",
      "|2005-05-21|\n",
      "|2011-08-06|\n",
      "|2015-05-10|\n",
      "|2005-09-02|\n",
      "|2005-12-22|\n",
      "+----------+\n",
      "only showing top 10 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "interactions\n",
    "    .select($\"date\")\n",
    "    .show(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "firstDate: Array[org.apache.spark.sql.Row] = Array([2000-01-25])\n"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val firstDate = interactions\n",
    "                    .selectExpr(\"min(date)\")\n",
    "                    .take(2)\n",
    "\n",
    "val firstDate_str = firstDate(0).getString(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+\n",
      "|user_id|\n",
      "+-------+\n",
      "|   2008|\n",
      "|   2008|\n",
      "+-------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "interactions\n",
    "    .filter($\"date\" === firstDate_str)\n",
    "    .select(\"user_id\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "// check\n",
    "\n",
    "interactions\n",
    "    .filter($\"user_id\" === 2008)\n",
    "    .show"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Ответ: 2008**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "4. Найдите пользователя, который берёт только новые рецепты (придумайте как это примерно посчитать с имеющимися данными)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+--------+------------+\n",
      "|   user_id|cnt_meal|sum_per_user|\n",
      "+----------+--------+------------+\n",
      "|     67344|       4|           0|\n",
      "|    543644|       4|           0|\n",
      "|    238395|       3|           0|\n",
      "|1803072009|       2|           0|\n",
      "|   2734292|       2|           0|\n",
      "|   2198435|       2|           0|\n",
      "|   1684663|       2|           0|\n",
      "|   1271572|       2|           0|\n",
      "|   2193344|       2|           0|\n",
      "|   2691191|       2|           0|\n",
      "|   2199258|       2|           0|\n",
      "|   1131759|       2|           0|\n",
      "|1803066303|       2|           0|\n",
      "|   2719902|       2|           0|\n",
      "|   2733922|       2|           0|\n",
      "|   1640593|       2|           0|\n",
      "|    560807|       1|           0|\n",
      "|    338428|       1|           0|\n",
      "|    579241|       1|           0|\n",
      "|   2688526|       1|           0|\n",
      "+----------+--------+------------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "// не зыбть вместо отрицательный чисел поставить 0\n",
    "\n",
    "interactions\n",
    "    .join(recipes, interactions(\"recipe_id\") === recipes(\"id\"), \"inner\")\n",
    "    .select(\"user_id\", \"id\", \"date\", \"submitted\")\n",
    "    .withColumn(\"datediff_days\", datediff($\"date\", $\"submitted\"))\n",
    "    .where(\"datediff_days > -1\")\n",
    "    .groupBy(\"user_id\")\n",
    "    .agg(\n",
    "        count(\"*\") as \"cnt_meal\",\n",
    "        sum(\"datediff_days\") as \"sum_per_user\"\n",
    "    )\n",
    "    .orderBy(struct($\"sum_per_user\", -$\"cnt_meal\"))\n",
    "    .show\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Проверка"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+---------+----------+------+--------------------+\n",
      "|user_id|recipe_id|      date|rating|              review|\n",
      "+-------+---------+----------+------+--------------------+\n",
      "|  67344|    50456|2003-01-08|     5|Unique and delicous!|\n",
      "|  67344|    50430|2003-01-08|     5|Light but satisfy...|\n",
      "|  67344|    50436|2003-01-08|     5|this was to die f...|\n",
      "|  67344|    50453|2003-01-08|     5|Light and quick d...|\n",
      "+-------+---------+----------+------+--------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "interactions\n",
    "    .filter($\"user_id\" === 67344)\n",
    "    .show"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----+----------+--------------------+-------+\n",
      "|   id| submitted|                name|minutes|\n",
      "+-----+----------+--------------------+-------+\n",
      "|50456|2003-01-08|mom s famous grah...|    100|\n",
      "|50430|2003-01-08|teresa s 10 minut...|     13|\n",
      "|50436|2003-01-08|teresa s chicken ...|     25|\n",
      "|50453|2003-01-08|teresa s veal  or...|     15|\n",
      "+-----+----------+--------------------+-------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "recipes\n",
    "    .filter($\"id\" === 50456 || $\"id\" === 50430 || $\"id\" === 50436 || $\"id\" === 50453)\n",
    "    .select(\"id\", \"submitted\", \"name\", \"minutes\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Вывод:** Рецепты появились на сайте и были приготовлены в один день"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Ответ:** 67344"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "5. Найдите пользователя, который выбирает самые сложные рецепты. Метрика: 0.2 * (медианное количество шагов в выбранном рецепте) + 0.3 * (среднее количество ингредиентов) + (длина описания, нормированная на максимальную длину описания)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "val max_descr = recipes\n",
    "                    .selectExpr(\"max(description) as m\")\n",
    "                    .take(1)\n",
    "\n",
    "val max_descr_str_length = max_descr(0).getString(0).length.toFloat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "val metrics = recipes\n",
    "    .select(\n",
    "        $\"id\" as \"recipe_id\",\n",
    "        $\"n_steps\",\n",
    "        $\"n_ingredients\",\n",
    "        length($\"description\") as \"desc_len\"\n",
    "    )\n",
    "    .withColumn(\"n_ingredients\", $\"n_ingredients\")\n",
    "    .filter($\"n_ingredients\" isNotNull)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+---------------+---------+-------------------+----------------+\n",
      "|user_id|ingredients_cnt|steps_cnt|     desc_normalize|          metric|\n",
      "+-------+---------------+---------+-------------------+----------------+\n",
      "| 171610|         2006.0|        3|0.03699551569506727|602.436995515695|\n",
      "+-------+---------------+---------+-------------------+----------------+\n",
      "only showing top 1 row\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "warning: there was one feature warning; for details, enable `:setting -feature' or `:replay -feature'\n",
       "recipes_cast: org.apache.spark.sql.Dataset[org.apache.spark.sql.Row] = [name: string, id: int ... 10 more fields]\n"
      ]
     },
     "execution_count": 350,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val recipes_cast = recipes\n",
    "    .withColumn(\"n_ingredients\", $\"n_ingredients\" cast(\"Int\"))\n",
    "    .withColumn(\"n_steps\", $\"n_steps\" cast(\"Int\"))\n",
    "    .filter($\"n_ingredients\" isNotNull)\n",
    "\n",
    "interactions\n",
    "    .select(\"user_id\", \"recipe_id\")\n",
    "    .join(recipes_cast, $\"user_id\" === $\"id\")\n",
    "    .groupBy(\"user_id\")\n",
    "    .agg(\n",
    "        mean(\"n_ingredients\") as \"ingredients_cnt\",\n",
    "        callUDF(\"percentile_approx\", col(\"n_steps\"), lit(0.5)) as \"steps_cnt\", //медиана\n",
    "        mean(length($\"description\")) / max_descr_str_length as \"desc_normalize\"\n",
    "    )\n",
    "    .withColumn(\"metric\", $\"steps_cnt\" * 0.2 + $\"ingredients_cnt\" * 0.3 + $\"desc_normalize\")\n",
    "    .orderBy(-$\"metric\")\n",
    "    .show(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Ответ: 171610**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "6. Найдите количество пользователей, которые выбирают наиболее разнообразную пищу. 2 вариации:\n",
    "    1. На основании тэгов\n",
    "    2. На основании лемматизированных описаний и TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "uniqueTags: org.apache.spark.sql.Dataset[org.apache.spark.sql.Row] = [user_id: int, unique_tags_count: bigint]\n"
      ]
     },
     "execution_count": 298,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val cnt_recipe_per_user = recipes\n",
    "    .join(interactions, interactions(\"recipe_id\") === recipes(\"id\"), \"inner\")\n",
    "    .groupBy( \"user_id\")\n",
    "    .agg(count(\"*\").cast(\"float\") as \"cnt\")\n",
    "\n",
    "val uniqueTags = recipes\n",
    "    .select($\"id\" as \"recipe_id\", explode(split(expr(\"substring(tags, 2, length(tags) - 1)\"), \",\")) as \"tags\")\n",
    "    .join(interactions.select(\"user_id\", \"recipe_id\"), Seq(\"recipe_id\"))\n",
    "    .groupBy(\"user_id\")\n",
    "    .agg(countDistinct(\"tags\") as \"unique_tags_cnt\")\n",
    "    .orderBy(-$\"unique_tags_cnt\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-----------------+-------+------+-------------------+\n",
      "|user_id|unique_tags_count|user_id|   cnt|           avg_tags|\n",
      "+-------+-----------------+-------+------+-------------------+\n",
      "| 424680|              643| 424680|7671.0|0.08382218745926216|\n",
      "|  37449|              606|  37449|5603.0|0.10815634481527753|\n",
      "| 169430|              580| 169430|4076.0| 0.1422963689892051|\n",
      "| 128473|              578| 128473|3917.0| 0.1475619096247128|\n",
      "|  58104|              571|  58104|3288.0|  0.173661800486618|\n",
      "| 133174|              559| 133174|3107.0| 0.1799163179916318|\n",
      "| 107135|              557| 107135|2020.0|0.27574257425742577|\n",
      "| 383346|              551| 383346|4628.0|0.11905790838375108|\n",
      "| 226863|              551| 226863|2874.0|0.19171885873347252|\n",
      "|  80353|              548|  80353|2220.0|0.24684684684684685|\n",
      "| 498271|              542| 498271|2644.0|0.20499243570347958|\n",
      "| 131126|              541| 131126|2138.0| 0.2530402245088868|\n",
      "|  89831|              534|  89831|3353.0|0.15926036385326572|\n",
      "| 107583|              528| 107583|2252.0| 0.2344582593250444|\n",
      "|  39835|              523|  39835|2260.0|0.23141592920353982|\n",
      "| 486725|              522| 486725|1776.0| 0.2939189189189189|\n",
      "|1072593|              522|1072593|2350.0| 0.2221276595744681|\n",
      "| 126440|              519| 126440|1950.0|0.26615384615384613|\n",
      "| 305531|              519| 305531|2902.0|  0.178842177808408|\n",
      "| 204024|              514| 204024|1595.0|0.32225705329153603|\n",
      "+-------+-----------------+-------+------+-------------------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "// первая ячейка\n",
    "\n",
    "uniqueTags\n",
    "    .join(cnt_recipe_per_user, cnt_recipe_per_user(\"user_id\") === uniqueTags(\"user_id\"))\n",
    "    .withColumn(\"avg_tags\", $\"unique_tags_cnt\"/$\"cnt\") \n",
    "    .orderBy(-$\"unique_tags_cnt\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+-----------------+----------+---+--------+\n",
      "|   user_id|unique_tags_count|   user_id|cnt|avg_tags|\n",
      "+----------+-----------------+----------+---+--------+\n",
      "|1803050464|               68|1803050464|1.0|    68.0|\n",
      "|1802640046|               68|1802640046|1.0|    68.0|\n",
      "|1800225281|               68|1800225281|1.0|    68.0|\n",
      "|2001286124|               68|2001286124|1.0|    68.0|\n",
      "|   1434863|               68|   1434863|1.0|    68.0|\n",
      "|    852602|               68|    852602|1.0|    68.0|\n",
      "|   1112186|               68|   1112186|1.0|    68.0|\n",
      "|    246492|               68|    246492|1.0|    68.0|\n",
      "|2001985095|               68|2001985095|1.0|    68.0|\n",
      "|    191658|               68|    191658|1.0|    68.0|\n",
      "|     99971|               68|     99971|1.0|    68.0|\n",
      "|   1543399|               68|   1543399|1.0|    68.0|\n",
      "|   1376859|               68|   1376859|1.0|    68.0|\n",
      "|    412777|               68|    412777|1.0|    68.0|\n",
      "|    108682|               68|    108682|1.0|    68.0|\n",
      "|   1387886|               68|   1387886|1.0|    68.0|\n",
      "|    857412|               68|    857412|1.0|    68.0|\n",
      "|   2281963|               68|   2281963|1.0|    68.0|\n",
      "|   1245091|               68|   1245091|1.0|    68.0|\n",
      "|   2106013|               68|   2106013|1.0|    68.0|\n",
      "+----------+-----------------+----------+---+--------+\n",
      "only showing top 20 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "// вторая ячейка\n",
    "\n",
    "uniqueTags\n",
    "    .join(cnt_recipe_per_user, cnt_recipe_per_user(\"user_id\") === uniqueTags(\"user_id\"), \"left\")\n",
    "    .withColumn(\"avg_tags\", $\"unique_tags_count\"/$\"cnt\") \n",
    "    .orderBy(-$\"avg_tags\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Ответ:**\n",
    "\n",
    "Условие можно понимать по-разнному, если интересует:\n",
    "\n",
    "    - *наиболее* разнообразный человек, то это 424680 (первая ячейка)\n",
    "    - максимальное соотношение числа тегов к количеству рецептов, то это список выше (вторая ячейка)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "7. \n",
    "\n",
    "\n",
    "а) Найдите медианное значение калорийности блюд, приготовленных пользователями с более чем 5-ю отзывами\n",
    "\n",
    "б) В качестве веса в медиане используйте количество отзывов на блюдо. Взвешенную медиану необходимо реализовать через оконные функции"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Для нахождения медианы используем `callUDF(\"percentile_approx\", $\"comment_count\", lit(0.5))`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "// TODO: сначала понять"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "8. Найдите слова в отзыве, которые наиболее коррелируют с хорошим рейтингом (4-5). Отфильтруйте слишком редкие слова"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----------+------------------+----------+\n",
      "|      word|       correlation|word_count|\n",
      "+----------+------------------+----------+\n",
      "| honkitten|               1.0|       348|\n",
      "|  gailanng|               1.0|       222|\n",
      "| honkitten|               1.0|       348|\n",
      "|  gailanng|               1.0|       222|\n",
      "|  adoptees| 0.998158379373849|      1086|\n",
      "|  adoptees| 0.998158379373849|      1086|\n",
      "|aussiekiwi|0.9960474308300395|       253|\n",
      "|aussiekiwi|0.9960474308300395|       253|\n",
      "| againmade|0.9957264957264957|       234|\n",
      "| againmade|0.9957264957264957|       234|\n",
      "|    boomie|0.9956709956709957|       231|\n",
      "|    boomie|0.9956709956709957|       231|\n",
      "| travelers|0.9955555555555555|       225|\n",
      "| travelers|0.9955555555555555|       225|\n",
      "| recipenap|0.9952380952380953|       210|\n",
      "| recipenap|0.9952380952380953|       210|\n",
      "|      pets|0.9951923076923077|       208|\n",
      "|      pets|0.9951923076923077|       208|\n",
      "|      kitz| 0.995049504950495|       202|\n",
      "|      kitz| 0.995049504950495|       202|\n",
      "+----------+------------------+----------+\n",
      "only showing top 20 rows\n",
      "\n",
      "Time: 50.71823024749756 seconds.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "// рабочее\n",
    "interactions\n",
    "    .select($\"rating\", explode(split(lower(regexp_replace($\"review\",\"[^a-zA-Z ]\", \"\")), \" \")) as \"word\")\n",
    "    .filter($\"word\" !== \"\")\n",
    "    .withColumn(\"word_count\", count(\"*\") over Window.partitionBy($\"word\")) // общее количество вхождения слова\n",
    "    .filter($\"word_count\" >= 200 && $\"rating\" > 3)\n",
    "    .withColumn(\"correlation\", count(\"*\") over Window.partitionBy($\"word\"))\n",
    "    .distinct\n",
    "    .select($\"word\", $\"correlation\" / $\"word_count\" as \"correlation\", $\"word_count\")\n",
    "    .orderBy(-$\"correlation\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "9. Найдите теги, блюда с которыми имеют в среднем худшие отзывы"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------------+------------------+\n",
      "|                 tag|       mean_rating|\n",
      "+--------------------+------------------+\n",
      "|      beef-crock-pot|               0.0|\n",
      "|lamb-sheep-main-dish|               0.0|\n",
      "|   main-dish-seafood|               0.0|\n",
      "|    shrimp-main-dish|               0.0|\n",
      "|           pot-roast|               0.0|\n",
      "|           bean-soup|               3.0|\n",
      "|     black-bean-soup|               3.0|\n",
      "|      main-dish-beef|               3.0|\n",
      "|       stews-poultry|3.3333333333333335|\n",
      "|                bear|3.6666666666666665|\n",
      "|    pressure-canning|3.6742424242424243|\n",
      "|       sugar-cookies| 3.707070707070707|\n",
      "|            honduran|3.7982456140350878|\n",
      "| unprocessed-freezer| 3.824915824915825|\n",
      "|fillings-and-fros...|3.9347826086956523|\n",
      "|            birthday| 3.944741532976827|\n",
      "|             jellies|3.9685714285714284|\n",
      "|          water-bath|3.9867256637168142|\n",
      "|    beef-barley-soup|               4.0|\n",
      "|high-in-something...|               4.0|\n",
      "+--------------------+------------------+\n",
      "only showing top 20 rows\n",
      "\n",
      "Time: 21.372616052627563 seconds.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "recipes\n",
    "    .select($\"id\", explode(split(expr(\"substring(tags, 2, length(tags)-1)\"), \", \")) as \"tag\")\n",
    "    .withColumn(\"tag\", trim($\"tag\", \"' []\"))\n",
    "    .join(interactions, $\"id\" === $\"recipe_id\")\n",
    "    .select(\"recipe_id\", \"rating\", \"tag\")\n",
    "    .groupBy(\"tag\")\n",
    "    .agg(mean(\"rating\") as \"mean_rating\")\n",
    "    .orderBy(\"mean_rating\")\n",
    "    .show"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "10. Найдите contributor'а, который прислал:\n",
    "    1. Наиболее используемый рецепт\n",
    "    2. Наиболее высоко оцененный рецепт"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------------+-------+----------+\n",
      "|contributor_id|max_cnt|max_rating|\n",
      "+--------------+-------+----------+\n",
      "|          1762|   1613|       5.0|\n",
      "+--------------+-------+----------+\n",
      "only showing top 1 row\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "recipeUserMetrics: Unit = ()\n"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val recipeUserMetrics = \n",
    "interactions\n",
    "    .groupBy(\"recipe_id\")\n",
    "    .agg(count(\"*\") as \"cnt\", mean(\"rating\") as \"mean_rating\")\n",
    "    .join(recipes, $\"id\" === $\"recipe_id\")\n",
    "    .groupBy(\"contributor_id\")\n",
    "    .agg(max(\"cnt\") as \"max_cnt\", max(\"mean_rating\") as \"max_rating\")\n",
    "    .filter($\"max_rating\" === 5.0)\n",
    "    .orderBy(-$\"max_cnt\")\n",
    "    .show(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "11. Опишите процесс работы broadcast join"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`broadcast join` используются при join маленького датафрейма с большим, для оптимизации времени исполнения запроса.\n",
    "Spark - разделяет данные на разных узлах кластера, чтобы впоследствии обрабатывать запросы паралелльно. Понятно, что мы хотим минимизировать количество операций перебора которые происходят при join, поэтому просто закинем маленькую таблицу во все узлы кластера сразу(на все рабочие машины).  Как раз для этого и пользуеемся **broadcast**: `.join(broadcast(small_df_name), ...)`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "12. Опишите процесс работы cache и причины его использования"
   ]
  },
  {
   "attachments": {
    "1*PlhXq8tHCT7SndMc1G8W6A.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Как работает?**\n",
    "\n",
    "На этапе планирование запроса spark делает следующее: \n",
    "\n",
    "![1*PlhXq8tHCT7SndMc1G8W6A.png](attachment:1*PlhXq8tHCT7SndMc1G8W6A.png)\n",
    "\n",
    "> \"When you run a query with an action, the query plan will be processed and transformed. In the step of the Cache Manager (just before the optimizer) Spark will check for each subtree of the analyzed plan if it is stored in the cachedData sequence. If it finds a match it means that the same plan (the same computation) has already been cached (perhaps in some previous query) and so Spark can use that and thus it adds that information to the query plan using the InMemoryRelation operator which will carry information about this cached plan.\"\n",
    "\n",
    "\n",
    "**Почему стоит использовать:**\n",
    "\n",
    "RDD.cache по-прежнему ленивая операция, которая \"ничего не делает\". Но если мы вызовем df.count 2 раза, то во второй раз spark достанет значение из кеша, а не будет его пересчитывать. Соответственно, если вы часто работаете с какими-то данными, стоит использовать .cache"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "spylon-kernel",
   "language": "scala",
   "name": "spylon-kernel"
  },
  "language_info": {
   "codemirror_mode": "text/x-scala",
   "file_extension": ".scala",
   "help_links": [
    {
     "text": "MetaKernel Magics",
     "url": "https://metakernel.readthedocs.io/en/latest/source/README.html"
    }
   ],
   "mimetype": "text/x-scala",
   "name": "scala",
   "pygments_lexer": "scala",
   "version": "0.4.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}