Final_Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2021-12-05T18:23:18.592252Z",
     "iopub.status.busy": "2021-12-05T18:23:18.591547Z",
     "iopub.status.idle": "2021-12-05T18:23:18.627294Z",
     "shell.execute_reply": "2021-12-05T18:23:18.626693Z",
     "shell.execute_reply.started": "2021-12-05T18:23:18.592118Z"
    }
   },
   "source": [
    "# Data Openning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:23:18.763985Z",
     "iopub.status.busy": "2021-12-05T18:23:18.763738Z",
     "iopub.status.idle": "2021-12-05T18:24:20.435074Z",
     "shell.execute_reply": "2021-12-05T18:24:20.434136Z",
     "shell.execute_reply.started": "2021-12-05T18:23:18.763959Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pyspark in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (2.4.0)\n",
      "Requirement already satisfied: py4j==0.10.7 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pyspark) (0.10.7)\n",
      "\u001b[33mWARNING: You are using pip version 21.2.4; however, version 21.3.1 is available.\n",
      "You should consider upgrading via the '/home/ec2-user/anaconda3/envs/python3/bin/python -m pip install --upgrade pip' command.\u001b[0m\n",
      "Requirement already satisfied: pyspark_dist_explore in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (0.1.8)\n",
      "Requirement already satisfied: matplotlib in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pyspark_dist_explore) (3.3.4)\n",
      "Requirement already satisfied: pandas in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pyspark_dist_explore) (1.1.5)\n",
      "Requirement already satisfied: numpy in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pyspark_dist_explore) (1.19.5)\n",
      "Requirement already satisfied: scipy in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pyspark_dist_explore) (1.5.3)\n",
      "Requirement already satisfied: python-dateutil>=2.1 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from matplotlib->pyspark_dist_explore) (2.8.2)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from matplotlib->pyspark_dist_explore) (0.10.0)\n",
      "Requirement already satisfied: pillow>=6.2.0 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from matplotlib->pyspark_dist_explore) (8.3.2)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from matplotlib->pyspark_dist_explore) (1.3.1)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.3 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from matplotlib->pyspark_dist_explore) (2.4.7)\n",
      "Requirement already satisfied: six in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from cycler>=0.10->matplotlib->pyspark_dist_explore) (1.16.0)\n",
      "Requirement already satisfied: pytz>=2017.2 in /home/ec2-user/anaconda3/envs/python3/lib/python3.6/site-packages (from pandas->pyspark_dist_explore) (2021.3)\n",
      "\u001b[33mWARNING: You are using pip version 21.2.4; however, version 21.3.1 is available.\n",
      "You should consider upgrading via the '/home/ec2-user/anaconda3/envs/python3/bin/python -m pip install --upgrade pip' command.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!pip install pyspark\n",
    "! pip install pyspark_dist_explore"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Notebook 1 - Data Wrangling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:29.285423Z",
     "iopub.status.busy": "2021-12-05T18:24:29.285178Z",
     "iopub.status.idle": "2021-12-05T18:24:29.655352Z",
     "shell.execute_reply": "2021-12-05T18:24:29.654488Z",
     "shell.execute_reply.started": "2021-12-05T18:24:29.285394Z"
    }
   },
   "outputs": [],
   "source": [
    "# Spark\n",
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql.functions import when\n",
    "from pyspark.sql.types import StructType\n",
    "from pyspark.sql.types import *\n",
    "from pyspark.sql.functions import isnan, when, count, col\n",
    "\n",
    "\n",
    "# EDA\n",
    "from pyspark_dist_explore import hist\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# ML\n",
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.feature import OneHotEncoder, StringIndexer, VectorAssembler, StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:29.658104Z",
     "iopub.status.busy": "2021-12-05T18:24:29.657616Z",
     "iopub.status.idle": "2021-12-05T18:24:35.231162Z",
     "shell.execute_reply": "2021-12-05T18:24:35.230154Z",
     "shell.execute_reply.started": "2021-12-05T18:24:29.658058Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "            <div>\n",
       "                <p><b>SparkSession - in-memory</b></p>\n",
       "                \n",
       "        <div>\n",
       "            <p><b>SparkContext</b></p>\n",
       "\n",
       "            <p><a href=\"http://ip-172-16-120-185.ec2.internal:4041\">Spark UI</a></p>\n",
       "\n",
       "            <dl>\n",
       "              <dt>Version</dt>\n",
       "                <dd><code>v2.4.0</code></dd>\n",
       "              <dt>Master</dt>\n",
       "                <dd><code>local[*]</code></dd>\n",
       "              <dt>AppName</dt>\n",
       "                <dd><code>Data wrangling with Spark SQL</code></dd>\n",
       "            </dl>\n",
       "        </div>\n",
       "        \n",
       "            </div>\n",
       "        "
      ],
      "text/plain": [
       "<pyspark.sql.session.SparkSession at 0x7f90c5199240>"
      ]
     },
     "execution_count": 230,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spark = SparkSession \\\n",
    "    .builder \\\n",
    "    .appName(\"Data wrangling with Spark SQL\") \\\n",
    "    .getOrCreate()\n",
    "\n",
    "spark"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:35.233050Z",
     "iopub.status.busy": "2021-12-05T18:24:35.232751Z",
     "iopub.status.idle": "2021-12-05T18:24:40.147156Z",
     "shell.execute_reply": "2021-12-05T18:24:40.146153Z",
     "shell.execute_reply.started": "2021-12-05T18:24:35.233006Z"
    }
   },
   "outputs": [],
   "source": [
    "# local do arquivo\n",
    "path = \"bank-additional-full.csv\"\n",
    "\n",
    "# Lendo o arquivo em um objeto Spark\n",
    "#df = spark.read.option(\"delimiter\", \";\").option(\"header\", \"true\").option(\"nullValue\", \"unknown\").option(\"nullValue\", \"null\").csv(path)\n",
    "df = spark.read.option(\"delimiter\", \";\").option(\"header\", \"true\").option(\"inferSchema\" , \"true\").csv(path)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.withColumnRenamed(\"emp.var.rate\",\"emp_var_rate\") \\\n",
    "                        .withColumnRenamed(\"cons.price.idx\",\"cons_price_idx\")\\\n",
    "                        .withColumnRenamed(\"cons.conf.idx\",\"cons_conf_idx\")\\\n",
    "                        .withColumnRenamed(\"nr.employed\",\"nr_employed\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:40.152626Z",
     "iopub.status.busy": "2021-12-05T18:24:40.152362Z",
     "iopub.status.idle": "2021-12-05T18:24:40.216454Z",
     "shell.execute_reply": "2021-12-05T18:24:40.215585Z",
     "shell.execute_reply.started": "2021-12-05T18:24:40.152589Z"
    }
   },
   "outputs": [],
   "source": [
    "df.createOrReplaceTempView(\"df_raw\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Show dataframe**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:40.218087Z",
     "iopub.status.busy": "2021-12-05T18:24:40.217806Z",
     "iopub.status.idle": "2021-12-05T18:24:40.645411Z",
     "shell.execute_reply": "2021-12-05T18:24:40.644539Z",
     "shell.execute_reply.started": "2021-12-05T18:24:40.218048Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "|age|  job|marital|education|default|housing|loan|contact|month|day_of_week|duration|campaign|pdays|previous|poutcome|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|nr_employed|  y|\n",
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "| 56|ho...|  ma...|    ba...|     no|     no|  no|  te...|  may|        mon|     261|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...| no|\n",
      "| 57|se...|  ma...|    hi...|  un...|     no|  no|  te...|  may|        mon|     149|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...| no|\n",
      "| 37|se...|  ma...|    hi...|     no|    yes|  no|  te...|  may|        mon|     226|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...| no|\n",
      "| 40|ad...|  ma...|    ba...|     no|     no|  no|  te...|  may|        mon|     151|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...| no|\n",
      "| 56|se...|  ma...|    hi...|     no|     no| yes|  te...|  may|        mon|     307|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...| no|\n",
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.show(5,truncate=5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Print the dataframe Schema**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:40.646974Z",
     "iopub.status.busy": "2021-12-05T18:24:40.646683Z",
     "iopub.status.idle": "2021-12-05T18:24:40.663122Z",
     "shell.execute_reply": "2021-12-05T18:24:40.662310Z",
     "shell.execute_reply.started": "2021-12-05T18:24:40.646934Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- age: integer (nullable = true)\n",
      " |-- job: string (nullable = true)\n",
      " |-- marital: string (nullable = true)\n",
      " |-- education: string (nullable = true)\n",
      " |-- default: string (nullable = true)\n",
      " |-- housing: string (nullable = true)\n",
      " |-- loan: string (nullable = true)\n",
      " |-- contact: string (nullable = true)\n",
      " |-- month: string (nullable = true)\n",
      " |-- day_of_week: string (nullable = true)\n",
      " |-- duration: integer (nullable = true)\n",
      " |-- campaign: integer (nullable = true)\n",
      " |-- pdays: integer (nullable = true)\n",
      " |-- previous: integer (nullable = true)\n",
      " |-- poutcome: string (nullable = true)\n",
      " |-- emp_var_rate: double (nullable = true)\n",
      " |-- cons_price_idx: double (nullable = true)\n",
      " |-- cons_conf_idx: double (nullable = true)\n",
      " |-- euribor3m: double (nullable = true)\n",
      " |-- nr_employed: double (nullable = true)\n",
      " |-- y: string (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Plotando o Schema da tabela spark\n",
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:40.664812Z",
     "iopub.status.busy": "2021-12-05T18:24:40.664500Z",
     "iopub.status.idle": "2021-12-05T18:24:41.445686Z",
     "shell.execute_reply": "2021-12-05T18:24:41.444734Z",
     "shell.execute_reply.started": "2021-12-05T18:24:40.664773Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There is 41188 samples in this DataFrame\n"
     ]
    }
   ],
   "source": [
    "print(\"There is \" + str(df.count()) +\" samples in this DataFrame\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Describe statistics into \"pdays\" column.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:41.450417Z",
     "iopub.status.busy": "2021-12-05T18:24:41.449924Z",
     "iopub.status.idle": "2021-12-05T18:24:42.650154Z",
     "shell.execute_reply": "2021-12-05T18:24:42.649579Z",
     "shell.execute_reply.started": "2021-12-05T18:24:41.450378Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-----------------+\n",
      "|summary|            pdays|\n",
      "+-------+-----------------+\n",
      "|  count|            41188|\n",
      "|   mean|962.4754540157328|\n",
      "| stddev|186.9109073447411|\n",
      "|    min|                0|\n",
      "|    max|              999|\n",
      "+-------+-----------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.describe(\"pdays\").show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:42.651238Z",
     "iopub.status.busy": "2021-12-05T18:24:42.651055Z",
     "iopub.status.idle": "2021-12-05T18:24:43.318757Z",
     "shell.execute_reply": "2021-12-05T18:24:43.317866Z",
     "shell.execute_reply.started": "2021-12-05T18:24:42.651214Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-----+\n",
      "|summary|    y|\n",
      "+-------+-----+\n",
      "|  count|41188|\n",
      "|   mean| null|\n",
      "| stddev| null|\n",
      "|    min|   no|\n",
      "|    max|  yes|\n",
      "+-------+-----+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.describe(\"y\").show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cleanning Raw Data\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:43.320282Z",
     "iopub.status.busy": "2021-12-05T18:24:43.319952Z",
     "iopub.status.idle": "2021-12-05T18:24:43.803194Z",
     "shell.execute_reply": "2021-12-05T18:24:43.802297Z",
     "shell.execute_reply.started": "2021-12-05T18:24:43.320182Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "|age|  job|marital|education|default|housing|loan|contact|month|day_of_week|campaign|pdays|previous|poutcome|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|nr_employed|  y|\n",
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "| 56|ho...|  ma...|    ba...|     no|     no|  no|  te...|  may|        mon|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...|  0|\n",
      "| 57|se...|  ma...|    hi...|   null|     no|  no|  te...|  may|        mon|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...|  0|\n",
      "| 37|se...|  ma...|    hi...|     no|    yes|  no|  te...|  may|        mon|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...|  0|\n",
      "| 40|ad...|  ma...|    ba...|     no|     no|  no|  te...|  may|        mon|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...|  0|\n",
      "| 56|se...|  ma...|    hi...|     no|     no| yes|  te...|  may|        mon|       1|  999|       0|   no...|         1.1|         93...|        -36.4|    4.857|      51...|  0|\n",
      "+---+-----+-------+---------+-------+-------+----+-------+-----+-----------+--------+-----+--------+--------+------------+--------------+-------------+---------+-----------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Task 1 - Replace statements by Null Values\n",
    "\n",
    "df_1 = df.replace('null', None, subset=['job', 'marital', 'education', 'default', 'housing', 'loan'])\\\n",
    "         .replace('unknown', None, subset=['job', 'marital', 'education', 'default', 'housing', 'loan'])\\\n",
    "\n",
    "\n",
    "# Task 2 -> Replace target values to Boolean variable\n",
    "df_2 = df_1.withColumn('y', when(df_1['y'] == 'no', 0).otherwise(1))\n",
    "\n",
    "# Task 3 -> Drop column \"duration\"\n",
    "columns_to_drop = ['duration']\n",
    "df_transient = df_2.drop(*columns_to_drop)\n",
    "\n",
    "df_transient.show(5,truncate=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:43.804390Z",
     "iopub.status.busy": "2021-12-05T18:24:43.804187Z",
     "iopub.status.idle": "2021-12-05T18:24:44.188077Z",
     "shell.execute_reply": "2021-12-05T18:24:44.187031Z",
     "shell.execute_reply.started": "2021-12-05T18:24:43.804366Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- age: integer (nullable = true)\n",
      " |-- job: string (nullable = true)\n",
      " |-- marital: string (nullable = true)\n",
      " |-- education: string (nullable = true)\n",
      " |-- default: string (nullable = true)\n",
      " |-- housing: string (nullable = true)\n",
      " |-- loan: string (nullable = true)\n",
      " |-- contact: string (nullable = true)\n",
      " |-- month: string (nullable = true)\n",
      " |-- day_of_week: string (nullable = true)\n",
      " |-- campaign: integer (nullable = true)\n",
      " |-- pdays: integer (nullable = true)\n",
      " |-- previous: integer (nullable = true)\n",
      " |-- poutcome: string (nullable = true)\n",
      " |-- emp_var_rate: double (nullable = true)\n",
      " |-- cons_price_idx: double (nullable = true)\n",
      " |-- cons_conf_idx: double (nullable = true)\n",
      " |-- euribor3m: double (nullable = true)\n",
      " |-- nr_employed: double (nullable = true)\n",
      " |-- y: integer (nullable = false)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Final Schema\n",
    "df_transient.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:44.546050Z",
     "iopub.status.busy": "2021-12-05T18:24:44.545771Z",
     "iopub.status.idle": "2021-12-05T18:24:45.830937Z",
     "shell.execute_reply": "2021-12-05T18:24:45.829887Z",
     "shell.execute_reply.started": "2021-12-05T18:24:44.546015Z"
    }
   },
   "outputs": [],
   "source": [
    "df_transient.write.option(\"header\",True).mode(\"overwrite\").option(\"delimiter\",\";\").csv('data_transient.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Notebook 2 - Data Engineering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Spark\n",
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql.functions import when\n",
    "from pyspark.sql.types import StructType\n",
    "from pyspark.sql.types import *\n",
    "from pyspark.sql.functions import isnan, when, count, col\n",
    "\n",
    "\n",
    "# EDA\n",
    "from pyspark_dist_explore import hist\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# ML\n",
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.feature import OneHotEncoder, StringIndexer, VectorAssembler, StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:45.832469Z",
     "iopub.status.busy": "2021-12-05T18:24:45.832162Z",
     "iopub.status.idle": "2021-12-05T18:24:45.842784Z",
     "shell.execute_reply": "2021-12-05T18:24:45.841732Z",
     "shell.execute_reply.started": "2021-12-05T18:24:45.832429Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "            <div>\n",
       "                <p><b>SparkSession - in-memory</b></p>\n",
       "                \n",
       "        <div>\n",
       "            <p><b>SparkContext</b></p>\n",
       "\n",
       "            <p><a href=\"http://ip-172-16-120-185.ec2.internal:4041\">Spark UI</a></p>\n",
       "\n",
       "            <dl>\n",
       "              <dt>Version</dt>\n",
       "                <dd><code>v2.4.0</code></dd>\n",
       "              <dt>Master</dt>\n",
       "                <dd><code>local[*]</code></dd>\n",
       "              <dt>AppName</dt>\n",
       "                <dd><code>Data wrangling with Spark SQL</code></dd>\n",
       "            </dl>\n",
       "        </div>\n",
       "        \n",
       "            </div>\n",
       "        "
      ],
      "text/plain": [
       "<pyspark.sql.session.SparkSession at 0x7f90c5199240>"
      ]
     },
     "execution_count": 243,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spark_new = SparkSession \\\n",
    "    .builder \\\n",
    "    .appName(\"Data Pre Processing with Spark\") \\\n",
    "    .getOrCreate()\n",
    "spark_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:45.845040Z",
     "iopub.status.busy": "2021-12-05T18:24:45.844369Z",
     "iopub.status.idle": "2021-12-05T18:24:45.920928Z",
     "shell.execute_reply": "2021-12-05T18:24:45.919970Z",
     "shell.execute_reply.started": "2021-12-05T18:24:45.844995Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "|age|      job|marital|  education|default|housing|loan|  contact|month|day_of_week|campaign|pdays|previous|   poutcome|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|nr_employed|  y|\n",
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "| 56|housemaid|married|   basic.4y|     no|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 57| services|married|high.school|   null|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 37| services|married|high.school|     no|    yes|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 40|   admin.|married|   basic.6y|     no|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 56| services|married|high.school|     no|     no| yes|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# local do arquivo\n",
    "path = 'data_transient.csv'\n",
    "\n",
    "# Lendo o arquivo em um objeto Spark\n",
    "#df = spark.read.option(\"delimiter\", \";\").option(\"header\", \"true\").option(\"nullValue\", \"unknown\").option(\"nullValue\", \"null\").csv(path)\n",
    "\n",
    "# Define csv input schema\n",
    "schema = StructType([\n",
    "    StructField(\"age\", IntegerType()),\n",
    "    StructField(\"job\", StringType()),\n",
    "    StructField(\"marital\", StringType()),\n",
    "    StructField(\"education\", StringType()),\n",
    "    StructField(\"default\", StringType()),\n",
    "    StructField(\"housing\", StringType()),\n",
    "    StructField(\"loan\", StringType()),\n",
    "    StructField(\"contact\", StringType()),\n",
    "    StructField(\"month\", StringType()),\n",
    "    StructField(\"day_of_week\", StringType()),\n",
    "    StructField(\"campaign\", IntegerType()),\n",
    "    StructField(\"pdays\", IntegerType()),\n",
    "    StructField(\"previous\", IntegerType()),\n",
    "    StructField(\"poutcome\", StringType()),\n",
    "    StructField(\"emp_var_rate\", DoubleType()),\n",
    "    StructField(\"cons_price_idx\", DoubleType()),\n",
    "    StructField(\"cons_conf_idx\", DoubleType()),\n",
    "    StructField(\"euribor3m\", DoubleType()),\n",
    "    StructField(\"nr_employed\", DoubleType()),\n",
    "    StructField(\"y\", IntegerType()),\n",
    "       \n",
    "])\n",
    "\n",
    "\n",
    "df = spark_new.read.option(\"delimiter\", \";\").option(\"header\", \"true\").csv(path, schema=schema)\n",
    "df.show(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:45.922578Z",
     "iopub.status.busy": "2021-12-05T18:24:45.922271Z",
     "iopub.status.idle": "2021-12-05T18:24:45.936251Z",
     "shell.execute_reply": "2021-12-05T18:24:45.935592Z",
     "shell.execute_reply.started": "2021-12-05T18:24:45.922534Z"
    }
   },
   "outputs": [],
   "source": [
    "df.createOrReplaceTempView(\"df_transient\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:45.937281Z",
     "iopub.status.busy": "2021-12-05T18:24:45.937083Z",
     "iopub.status.idle": "2021-12-05T18:24:45.943804Z",
     "shell.execute_reply": "2021-12-05T18:24:45.942959Z",
     "shell.execute_reply.started": "2021-12-05T18:24:45.937257Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- age: integer (nullable = true)\n",
      " |-- job: string (nullable = true)\n",
      " |-- marital: string (nullable = true)\n",
      " |-- education: string (nullable = true)\n",
      " |-- default: string (nullable = true)\n",
      " |-- housing: string (nullable = true)\n",
      " |-- loan: string (nullable = true)\n",
      " |-- contact: string (nullable = true)\n",
      " |-- month: string (nullable = true)\n",
      " |-- day_of_week: string (nullable = true)\n",
      " |-- campaign: integer (nullable = true)\n",
      " |-- pdays: integer (nullable = true)\n",
      " |-- previous: integer (nullable = true)\n",
      " |-- poutcome: string (nullable = true)\n",
      " |-- emp_var_rate: double (nullable = true)\n",
      " |-- cons_price_idx: double (nullable = true)\n",
      " |-- cons_conf_idx: double (nullable = true)\n",
      " |-- euribor3m: double (nullable = true)\n",
      " |-- nr_employed: double (nullable = true)\n",
      " |-- y: integer (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:45.945863Z",
     "iopub.status.busy": "2021-12-05T18:24:45.945577Z",
     "iopub.status.idle": "2021-12-05T18:24:46.221997Z",
     "shell.execute_reply": "2021-12-05T18:24:46.220836Z",
     "shell.execute_reply.started": "2021-12-05T18:24:45.945825Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "|age|      job|marital|  education|default|housing|loan|  contact|month|day_of_week|campaign|pdays|previous|   poutcome|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|nr_employed|  y|\n",
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "| 56|housemaid|married|   basic.4y|     no|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 57| services|married|high.school|   null|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 37| services|married|high.school|     no|    yes|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 40|   admin.|married|   basic.6y|     no|     no|  no|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "| 56| services|married|high.school|     no|     no| yes|telephone|  may|        mon|       1|  999|       0|nonexistent|         1.1|        93.994|        -36.4|    4.857|     5191.0|  0|\n",
      "+---+---------+-------+-----------+-------+-------+----+---------+-----+-----------+--------+-----+--------+-----------+------------+--------------+-------------+---------+-----------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "spark_new.sql('''Select * FROM df_transient''').show(5,truncate=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Some Exploratory Analisys - Numerical Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 248,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:46.401034Z",
     "iopub.status.busy": "2021-12-05T18:24:46.400750Z",
     "iopub.status.idle": "2021-12-05T18:24:50.927262Z",
     "shell.execute_reply": "2021-12-05T18:24:50.926435Z",
     "shell.execute_reply.started": "2021-12-05T18:24:46.400997Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x576 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting histogram\n",
    "df_eda = df\n",
    "\n",
    "fig, axes = plt.subplots(nrows=1, ncols=3)\n",
    "fig.set_size_inches(14, 8)\n",
    "\n",
    "# age histogram\n",
    "hist(axes[0], df_eda.select('age'), bins = 50, color=['red']);\n",
    "axes[0].set_title('Histogram of age');\n",
    "\n",
    "# nr_employed histogram\n",
    "hist(axes[1], df_eda.select('campaign'), bins = 50, color=['red']);\n",
    "axes[1].set_title('Histogram of campaign');\n",
    "\n",
    "# euribor3m histogram\n",
    "hist(axes[2], df_eda.select('y'), bins = 50, color=['red']);\n",
    "axes[2].set_title('Histogram of target');\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:50.929236Z",
     "iopub.status.busy": "2021-12-05T18:24:50.928751Z",
     "iopub.status.idle": "2021-12-05T18:24:56.361483Z",
     "shell.execute_reply": "2021-12-05T18:24:56.360622Z",
     "shell.execute_reply.started": "2021-12-05T18:24:50.929193Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA68AAAJOCAYAAABY9zF6AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAABsLUlEQVR4nO3de5hlVX3n//cnjaAICobWAA2CCZKgkyhpkcTRUVGnaYlNZowD4wUIGcII3iYZRUxiMrmMSQzgbWBaQUQFhiES+DltlICGEEVpBBFsCA0i3XYLjcQLIREI398fe5dsTp9T96pzqur9ep7z1L6svfd3n1O1an/PXmvtVBWSJEmSJI2ynxh2AJIkSZIkTcTkVZIkSZI08kxeJUmSJEkjz+RVkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXpegJDcnefGw4ximJL+aZFOS+5M8d4b72i9JJdlhtuKTlhrrpdmtl0ZFknOT/NE8H/PFSTbP5zGl2WJdOP91YXucZ7TT815naWpMXheZJHcmeVnPsmOTXD02X1XPqqovTLCfxZ6QvRc4uap2qarrhx2MtJhZL02a9ZK0iFkXTtq81oXtce6Yi30n+bM2Ef9Bkm8leddcHGcpMXnVUIxAhft04OYhxyBphCymemkEzkXSAjUC9ce8XKPN5Xmm8RPA2cDPVtWTgF8G/nOS/zBXx10KTF6XoO43f0kOSbK+/Ubo7iSntcWuan9+r21O8UtJfiLJ77TfHN2T5LwkT+7s9w3tuu8m+d2e4/x+kouTfCLJD4Bj22N/Kcn3kmxN8sEkO3b2V0nemOS2JD9M8odJfrrd5gdJLuqW7znHvrEm2SnJ/cAy4GtJbh+wfSV5c5I7ktyb5M/bSogky5K8t11+B/DKnm2PS7KhjfmOJL/ZWXdTkl/pzD+u3c9zkjy+fX++274n1yZ52mQ/V2khs16adL10Ynvsf0zyoSRp1x2b5O+TnJ7kPuD3J3i/f72tp/4xyWeTPH0655i2iW6SU9u67M4krx3nuP8lycYk9yW5LMle7fIPJfmLnrL/X5K3ttN7JfnLJNuSfDPJmzvlnpCmqd8/JvkG8Lzxzl0aZdaFk6oLfzbJ5W09cmuS13TWfSHJb3TmH3Nnu437pCS3Abd1lv1M5xB7tPv/YZK/7akffznN9dn325+/3HPsP07y98ADwDOq6taq+qfOvh8BfqYtP3YH/bg0d2f/MU0d/7wkN7bv/Qf7vQdLWlX5WkQv4E7gZT3LjgWu7lcG+BLw+nZ6F+DQdno/oIAdOtv9OrAReEZb9lPAx9t1BwH3A/8W2JGmycdDneP8fjt/JM2XJk8AfhE4FNihPd4G4K2d4xVwGfAk4FnAj4Ar2uM/GfgGcMyA92FgrJ19/8w472MBnweeAuwL/APwG+26E4FbgH3a9Z/vvlc0yexPAwH+HU0FdnC77u3A/+kcZw3w9Xb6N4H/D9iZpuL+ReBJw/6d8uVrpi/rpYlj7ex7onrp08Bubb20DVjVeT8fBt7Uxv6EcfZzZBvHz7Vlfwf44nTOEXhxe9zTgJ1o6rx/Ag5s158L/FE7/VLgXuDgtuwHgKvadYcAW4CfaOf3oKk7n9Z+NtcBv9d+js8A7gD+fVv2PcDf0dTH+wA3AZuH/Xvvy1fvC+vCCWPt7LtvXQg8EdgEHNfGdnBbrzyrXf8F2uu1Ae9vAZe39cUTeo9HU2f9EHhRW0+9b2z7dpt/BF7fHvvodv4nO8e+q30/dgAe1y4/pX3/i6buWtHzOZ4FPB54BfAvwF8BTwX2Bu4B/t2wf3dH6TX0AHzN8gfaVHr3A9/rvB5gcMV4FfAHwB49+xn7g+pWjFcAb+zMH0hT2e1Ac1FxQWfdzsCDPLZivGqC2N8KXNKZL+AFnfnrgHd05v8COGPAvgbG2tn3RBeJqzrzbwSuaKevBE7srHtF73vVs6+/At7STu/VVopPaucvBt7eTv868EXg54f9e+TL12y+rJcmjrWz74nqpX/bmb8IOKWdPha4a5Kfx2eA4zvzP9F+Hk+f6jnyaPL6xJ64fredPpdHk9ezgT/rlNulPf/92vkNwMvb6ZOBde3083vPDXgn8NF2+g4eW1+fgMmrrxF8WRdOHGtn34OS1/8E/F3Psv8NvLud/gITJ68v7dm+N3m9sLNuF+Bfab4Yez3wlZ5tvwQc2zn2/xgQd4Dntp/nrj2f496dct8F/lNn/i/pfGngq2w2vEgdWVW7jb1oEq9BjgeeCdzSNn84YpyyewHf6sx/i6ZSfFq7btPYiqp6gOYPsGtTdybJM5N8Osl32mYqf0LzbXvX3Z3pf+4zv8s0Yp2sbrzfavc5tu/edT+W5PAk17TNWb4HrKY9r6raAvw98B+T7AYcDnyy3fTjwGeBC5NsSdPJ/3FTiFcaZdZLs1Mvfacz/UDPsTYxOU8H3tc2SfsecB/NhdXenTJTOcd/rMc2i+vWl12POf+qup/m8xg77seA17XTr6OpE8fi3Wss3jbmU3n0fRu3TpZGjHXhzOrCpwPP76kPXgv81CS2HTNRXdl9r+6nqSP36hM37Xy37uy772pcT/O+/EHP6um+j0uSyesSV1W3VdXRNM0T/hS4OMkTab4J6rWFptIYsy/NN+53A1uBFWMrkjwB+Mnew/XMn0nT/PaAajqyn0pzATUbxot1svbp2X5LO721zzoAkuxE8y3Ze4Gntf+Y1vHY8xq7QPs14EtV9W2Aqnqoqv6gqg6i6dR/BPCGKcQrLQrWS9PW7/3pZxPwm90L6Kp6QlV9cZrH3b39fMZ068uux5x/u81PAt9uF30CWJPkF2iaNP9VJ95v9sS7a1WtbtcPrJOlhcy6sK9NwN/21Ae7VNV/bdf/E82d5TH9ktqJ6sof1ydJdqFpLrylT9xjsX+7Mz/Rvneg6VqmaTJ5XeKSvC7J8qp6hKb5CjTNI7bRdCp/Rqf4BcDbkuzf/jH/CU3/zYdpmr/+StuRfUeab5UmquR2BX4A3J/kZ4H/OkH5qRgv1sn670l2T7IP8Bbg/7TLLwLenGRFkt1p+jKM2ZGmj8Q24OEkh9M0K+76K5o+Gm8BzhtbmOQlSf5NkmU078tDNJ+FtKRYL825s4B3JnkWQDtQyq/NcJ9/kGTHJC+k+eLt//Ypcz5wXJoB6naiOf8vV9WdAFW1GbiW5o7rX1bVP7fbfQX4QZJ3pBmcaVmSZycZG5jpovZ8dk+ygqbfr7TgWRf29WngmUlen2bQy8e1Axz9XLv+BuA/JNk5zSBMx08jvtVJ/m37Xv0hTT21ieZmxDOT/OckOyT5TzT9iT/dbydpBqb6zbZuSpJDgJNomk1rmkxetQq4Oc3obu8Djqqqf2mblPwx8Pdts4xDgXNoLiquAr5J06n8TQBVdXM7fSHNN3w/pOlk/qNxjv3bwH9uy36YR5PD2TAw1im4lKYPxw3A/6PprwVNrJ8FvgZ8lWagAQCq6ofAm2kupv6R5vwu6+60vSD7S2D/7rY03w5eTPPPYgPwtzR3IqSlxnppDlXVJTR3cS5smwPeRNOFYbq+Q1PfbaHpBnFiVd3S57hXAL9LU/9tpbn7cFRPsY8B/4ZHmwxTVf8K/ArwHJr37V7gIzSDwkBzIf6tdt3nuttKC5x1YY/2OusVNHXHFpr6509pbhwAnE7Tn/dumvrkk312M5HzgXfTNBf+RZpmyVTVd2m+nPstmmbXbweOqKp7x9nXrwK307yPn6AZqO4D04hJrVRNtpWRNHntN2nfo2lu8s0hhzNlSYom9o1ztP/fA55ZVa+bsLCkWbHQ66VRlOTFwCeqasUERSe7vxfRXODt195tkjTLrAu1kHnnVbMmya+0zTSeSNPn8+s0o+apI8lTaJqxrB12LNJiZ720cKQZoO4twEdMXKXZZV2oxcLkVbNpDY92aD+ApnmLt/Y7kvwXmsEGPlNVV01UXtKMLcl6KclZSe7v8zpr2LH10/ZX+x6wJ3DGUIORFqclWRdq8bHZsCRJkiRp5HnnVZIkSZI08nYYdgBTsccee9R+++037DAkjZjrrrvu3qpaPuw4Zot1naR+rOskLQXj1XULKnndb7/9WL9+/bDDkDRiknxr2DHMJus6Sf1Y10laCsar62w2LEmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJFn8ipJkiRJGnmTTl6TLEtyfZJPt/NPSXJ5ktvan7sP2G5VkluTbExySmf5uUle3dnX9UmOm+kJSZIkSZIWn6nceX0LsKEzfwpwRVUdAFzRzj9GkmXAh4DDgYOAo5Mc1FPmycBngbVV9dGphS9JkiRJWgomlbwmWQG8EvhIZ/Ea4GPt9MeAI/tsegiwsaruqKoHgQvb7cbsAnwGOL+qzpxa6JIkSZKkpWKyd17PAN4OPNJZ9rSq2grQ/nxqn+32BjZ15je3y8acBlxdVacPOnCSE5KsT7J+27ZtkwxXkiRJkrSYTJi8JjkCuKeqrpvG/tNnWXWmrwTWJOmX+DaFq9ZW1cqqWrl8+fJphCBJkiRJWugmc+f1BcCrktxJ0+z3pUk+AdydZE+A9uc9fbbdDOzTmV8BbOnMXwicCaxLsuvUw5e04CXbv4YSRs5Jck+Sm3qWv6kddO7mJH82YFsHppM0vhGp6yRpTs1xXTdh8lpV76yqFVW1H3AUcGVVvQ64DDimLXYMcGmfza8FDkiyf5Id2+0v69n/GTQDPl3SlpGkYTgXWNVdkOQlNP30f76qngW8t3cjB6aTJEmaHzN5zut7gJcnuQ14eTtPkr2SrAOoqoeBk2ku2jYAF1XVzb07qqp30PSN/XgSnz0rad5V1VXAfT2L/yvwnqr6UVumXwsTB6aTJEmaBztMpXBVfQH4Qjv9XeCwPmW2AKs78+uAdX3KHdszb1M6SaPmmcALk/wx8C/Ab1fVtT1l+g1M9/zO/GnAR8YbmA6awemAEwD23XffmcYtSZK06HiXU5IG2wHYHTgU+O/ARcl2nTdmPDAdODidJEnSRExeJWmwzcCnqvEVmseF7dGnjAPTSZIkzTGTV0ka7K+AlwIkeSawI3BvTxkHppMkSZoHJq+SBCS5APgScGCSzUmOB84BntE+PudC4JiqKgemkyRJmn9TGrBJkharqjp6wKrX9SnrwHSSFqwk5wBHAPdU1bN71v028OfA8qrqbWlCklXA+4BlNIPRjT1t4lzg01V1cZKn0LQ2eb+PB5M0m/z2X5IkaWk5l57nWgMk2Yfm8Yd39dvI51pLGjaTV0mSpCVkwHOtAU4H3s5jR0zv8rnWkobK5FWSJGmJS/Iq4NtV9bVxivV7rvXenfnTgKvHe651khOSrE+yftu2bTOKWdLSY/IqSZK0hCXZGXgX8HsTFe2zbErPtfaZ1pJmwuRVkiRpaftpYH/ga0nupHle9VeT/FRPOZ9rLWmoTF4lSZKWsKr6elU9tar2q6r9aJLUg6vqOz1Ffa61pKEyeZUkSVpCBjzXelBZn2staWT4nFdJkqQlZJznWo+t368z7XOtJY0Mvw2TJEmSJI08k1dJkiRJ0sgzeZUkSZIkjTyTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLIM3mVJEmSJI08k1dJkiRJ0sibMHlN8vgkX0nytSQ3J/mDdvnvJ/l2khva1+oB269KcmuSjUlO6Sw/N8mr2+mnJLk+iQ+0liRJkiRtZ4dJlPkR8NKquj/J44Crk3ymXXd6Vb130IZJlgEfAl4ObAauTXJZVX2jU+bJwGeBtVX10emeiCRJkiRp8Zrwzms17m9nH9e+apL7PwTYWFV3VNWDwIXAms76XYDPAOdX1ZmTD1uSJEmStJRMqs9rkmVJbgDuAS6vqi+3q05OcmOSc5Ls3mfTvYFNnfnN7bIxpwFXV9Xp4xz7hCTrk6zftm3bZMKVJEmSJC0yk0peq+pfq+o5wArgkCTPBs4Efhp4DrAV+Is+m6bf7jrTVwJrkjx1nGOvraqVVbVy+fLlkwlXkqas/RLuniQ39Vn320kqyR4DtrVvvyRJ0hyb0mjDVfU94AvAqqq6u01qHwE+TNNEuNdmYJ/O/ApgS2f+QpokeF2SXacSiyTNsnOBVb0Lk+xD02//rn4bdfr2Hw4cBByd5KCeMvbtlyRJmqHJjDa8PMlu7fQTgJcBtyTZs1PsV4Ht7lYA1wIHJNk/yY7AUcBl3QJVdQZwBXBJW0aS5l1VXQXc12fV6cDbGdzX3779kiRJ82Ayd173BD6f5EaaZPTyqvo08GdJvt4ufwnwNoAkeyVZB1BVDwMn09xx2ABcVFU39x6gqt5B0zf240l89qykkZDkVcC3q+pr4xSbcd/+9lj275ckSRrHhI/Kqaobgef2Wf76AeW3AKs78+uAdX3KHdszbz8wSSMjyc7Au4BXTFS0z7J+ffvfW1X3DNpJVa0F1gKsXLlysiO6S5IkLRne5ZSk/n4a2B/4WpI7afrsfzXJT/WUs2+/JEnSPDB5laQ+qurrVfXUqtqvqvajSVIPrqrv9BS1b78kSdI8MHmVJCDJBcCXgAOTbE5y/Dhl7dsvSZI0zybs8ypJS0FVHT3B+v060/btlyRJmmd++y9JkiRJGnkmr5IkSZKkkWfyKkmStIQkOSfJPUlu6iz78yS3JLkxySVJdhuw7aoktybZmOSUzvJzk7y6nX5KkuuT2FVC0qwyeZUkSVpazgVW9Sy7HHh2Vf088A/AO3s3SrIM+BBwOHAQcHSSg3rKPJlmALu1VfXR2Q9d0lJm8ipJkrSEVNVVwH09yz7Xjp4OcA3NM6t7HQJsrKo7qupBmudYr+ms3wX4DHB+VZ05+5FLWupMXiVJktT16zRJaK+9aR75NWZzu2zMacDVVXX6oB0nOSHJ+iTrt23bNivBSlo6TF4lSZIEQJJ3AQ8Dn+y3us+y6kxfCaxJ8tRB+6+qtVW1sqpWLl++fGbBSlpyfM6rJEmSSHIMcARwWFVVnyKbgX068yuALZ35C4GrgXVJXlJVP5yzYCUtSd55lSRJWuKSrALeAbyqqh4YUOxa4IAk+yfZETgKuKxboKrOAK4ALmnLSNKsMXmVJElaQpJcAHwJODDJ5iTHAx8EdgUuT3JDkrPasnslWQfQDuh0Ms1owhuAi6rq5t79V9U7aPrGfjyJ15qSZo3NhiVJkpaQqjq6z+KzB5TdAqzuzK8D1vUpd2zPvM94lTTr/DZMkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXiVJkiRJI8/kVZIkSZI08iZMXpM8PslXknwtyc1J/qBd/pQklye5rf25+4DtVyW5NcnGJKd0lp+b5NWdfV2fxJHpJEmSJEnbmcyd1x8BL62qXwCeA6xKcihwCnBFVR1A8zDqU3o3TLIM+BBwOHAQcHSSg3rKPJnmeWFrq+qjMzgXSZIkSdIiNWHyWo3729nHta8C1gAfa5d/DDiyz+aHABur6o6qehC4sN1uzC7AZ4Dzq+rMaZ2BJEmSJGnRm1Sf1yTLktwA3ANcXlVfBp5WVVsB2p9P7bPp3sCmzvzmdtmY04Crq+r0cY59QpL1SdZv27ZtMuFKkiRJkhaZSSWvVfWvVfUcYAVwSJJnT3L/6be7zvSVwJok/RLfsWOvraqVVbVy+fLlkzysJE1NknOS3JPkps6yP09yS5Ibk1ySZLcB29q3X5IkaY5NabThqvoe8AVgFXB3kj0B2p/39NlkM7BPZ34FsKUzfyFwJrAuya5TiUWSZtm5NHVb1+XAs6vq54F/AN7Zu5F9+yVJkubHZEYbXj52tyHJE4CXAbcAlwHHtMWOAS7ts/m1wAFJ9k+yI3BUu92PVdUZNAM+XdKWkaR5V1VXAff1LPtcVT3czl5D8wVcL/v2S5IkzYPJ3HndE/h8khtpktHLq+rTwHuAlye5DXh5O0+SvZKsA2gv+k6mueOwAbioqm7uPUBVvYOmb+zHk/jsWUmj6NdpktBeM+7bD/bvlyRJmsgOExWoqhuB5/ZZ/l3gsD7LtwCrO/PrgHV9yh3bM28/MEkjKcm7gIeBT/Zb3WdZv779762qft0rmg2q1gJrAVauXFmDykmSJC1VEyavkrSUJTkGOAI4rKr6JZWT6dt/NU3f/pdU1Q/nLFhJkqRFzCa6kjRAklXAO4BXVdUDA4rZt1+SJGkemLxKEpDkAuBLwIFJNic5HvggsCtweZIbkpzVlrVvvyRJ0jyz2bAkAVV1dJ/FZw8oa99+SZKkeea3/5IkSZKkkWfyKkmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJFn8ipJkrSEJDknyT1Jbuose0qSy5Pc1v7cfcC2q5LcmmRjklM6y89N8urOvq5P4gjrkmaVyaskSdLSci6wqmfZKcAVVXUAcEU7/xhJlgEfAg4HDgKOTnJQT5kn0zz3em1VfXT2Q5e0lJm8SpIkLSFVdRVwX8/iNcDH2umPAUf22fQQYGNV3VFVDwIXttuN2QX4DHB+VZ05q0FLEiavkiRJgqdV1VaA9udT+5TZG9jUmd/cLhtzGnB1VZ0+6CBJTkiyPsn6bdu2zULYkpYSk1dJkiRNRvosq870lcCaJP0S36Zw1dqqWllVK5cvXz7rAUpa3ExeJUmSdHeSPQHan/f0KbMZ2KczvwLY0pm/EDgTWJdk17kKVNLSZfIqSZKky4Bj2uljgEv7lLkWOCDJ/kl2BI5qt/uxqjqDZsCnS9oykjRrTF4lSZKWkCQXAF8CDkyyOcnxwHuAlye5DXh5O0+SvZKsA6iqh4GTaUYT3gBcVFU39+6/qt5B0zf240m81pQ0a3YYdgCSJEmaP1V19IBVh/UpuwVY3ZlfB6zrU+7Ynnmf8Spp1vltmCRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkTZi8JtknyeeTbEhyc5K3tMt/P8m3k9zQvlYP2H5VkluTbExySmf5uUle3U4/Jcn1SewfIUmSJEnazmQGbHoY+K2q+mr7zK7rklzerju9qt47aMMky4AP0Yxatxm4NsllVfWNTpkn04xat7aqPjrdE5EkSZIkLV4T3nmtqq1V9dV2+oc0Q6PvPcn9HwJsrKo7qupBmodXr+ms3wX4DHB+VZ05pcglSZIkSUvGlPq8JtkPeC7w5XbRyUluTHJOkt37bLI3zXO+xmzmsYnvacDVVXX6OMc8Icn6JOu3bds2lXAladLaeuyeJDd1lj0lyeVJbmt/9qvn7B4hSZI0DyadvCbZBfhL4K1V9QPgTOCngecAW4G/6LdZn2XVmb4SWJPkqYOOW1Vrq2plVa1cvnz5ZMOVpKk6F1jVs+wU4IqqOgC4op1/jE73iMOBg4CjkxzUU8buEZIkSTM0qeQ1yeNoEtdPVtWnAKrq7qr616p6BPgwTRPhXpuBfTrzK4AtnfkLaZLgdW1/Wkkaiqq6CrivZ/Ea4GPt9MeAI/tsavcISZKkeTCZ0YYDnA1sqKrTOsv37BT7VeCm3m2Ba4EDkuyfZEfgKOCyboGqOoPmjsYlbRlJGhVPq6qt0PT/B/q1Eplx9wiwi4QkSRpxyWNfQzCZO68vAF4PvLTnsTh/luTrSW4EXgK8DSDJXknWAVTVw8DJNM3lNgAXVdXNvQeoqnfQXPx9PInPnpW0kMy4ewTYRUKSJGkiEz4qp6qupv/F2boB5bcAqzvz6/qVrapje+YdxETSqLk7yZ5VtbVtbXJPnzKT6R5xNU33iJe0o7ZLkiRpirzLKUmDXQYc004fA1zap4zdIyRJkuaByaskAUkuAL4EHJhkc5LjgfcAL09yG/Dydt7uEZIkSUMwYbNhSVoKquroAasO61PW7hGSJEnzzG//JUmSJEkjz+RVkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXiVJkiRJI8/kVZIkSZI08kxeJUmSJEkjz+RVkiRJkjTyTF4lSZJEkrcluTnJTUkuSPL4nvVJ8v4kG5PcmOTgzrr7O9Ork9yWZN/5jF/S4mfyKkmStMQl2Rt4M7Cyqp4NLAOO6il2OHBA+zoBOLPPfg4DPgCsqqq75jRoSUuOyaskSZIAdgCekGQHYGdgS8/6NcB51bgG2C3JnmMrk7wQ+DDwyqq6fb6ClrR0mLxKkiQtcVX1beC9wF3AVuD7VfW5nmJ7A5s685vbZQA7AZcCR1bVLYOOk+SEJOuTrN+2bdusxS9paTB5lSRJWuKS7E5zZ3V/YC/giUle11usz6bV/nwI+CJw/HjHqaq1VbWyqlYuX758hlFLWmpMXiVJkvQy4JtVta2qHgI+BfxyT5nNwD6d+RU82rT4EeA1wPOSnDrXwUpamkxeJUmSdBdwaJKdkwQ4DNjQU+Yy4A3tqMOH0jQt3jq2sqoeAI4AXptk3DuwkjQdOww7AEmSJA1XVX05ycXAV4GHgeuBtUlObNefBawDVgMbgQeA4/rs574kq4CrktxbVZfO1zlIWvxMXiVJGlXp08Wwavtl0iyoqncD7+5ZfFZnfQEnDdh2l870Jpq+s5I0q2w2LEmSJEkaeRMmr0n2SfL5JBuS3JzkLe3ypyS5PMlt7c/dB2y/KsmtSTYmOaWz/Nwkr+7s6/ok2zU/kSRJkiRpMndeHwZ+q6p+DjgUOCnJQcApwBVVdQBwRTv/GEmWAR8CDgcOAo5ut+2WeTLwWWBtVX10JicjSZIkSVqcJkxeq2prVX21nf4hzchze9M8C+xjbbGPAUf22fwQYGNV3VFVDwIXttuN2QX4DHB+VZ053ZOQpLmU5G1ty5ObklyQ5PE965Pk/W0LkxuTHNxZd39nenXbWmXf+YxfkiRpMZhSn9ck+wHPBb4MPG1sePT251P7bLI3sKkzv7ldNuY04OqqOn2cY56QZH2S9du2bZtKuJI0Y0n2Bt4MrKyqZwPLgKN6ih0OHNC+TgC2+zIuyWHAB4BVVXXXnAYtSZK0CE06eU2yC/CXwFur6geT3azPsu4wiVcCa5L0S3ybwlVrq2plVa1cvnz5ZMOVpNm0A/CEJDsAOwNbetavAc6rxjXAbkn2HFuZ5IXAh4FXVtXt8xW0JEnSYjKp5DXJ42gS109W1afaxXePXZy1P+/ps+lmYJ/O/Aoee9F3Ic0dinVJdp1i7JI056rq28B7gbuArcD3q+pzPcXGa2WyE3ApcGRV3TLoOLYykSRJGt9kRhsOcDawoapO66y6DDimnT6G5uKs17XAAUn2T7IjTVO7y7oFquoMmgGfLmnLSNLIaEdSX0PzzMK9gCcmeV1vsT6bjrUyeQj4InD8eMexlYkkSdL4JnPn9QXA64GXJrmhfa0G3gO8PMltwMvbeZLslWQdQFU9DJxMM5rwBuCiqrq59wBV9Q6auxYfT+KzZyWNkpcB36yqbVX1EPAp4Jd7yozXyuQR4DXA85KcOtfBSpIkLVY7TFSgqq6m/10FgMP6lN8CrO7MrwPW9Sl3bM+8z3iVNIruAg5NsjPwzzT13vqeMpcBJye5EHg+TdPirWMrq+qBJEcAf5fk7qo6e55ilyRJWjQmTF4laSmrqi8nuRj4Ks1zr68H1iY5sV1/Fs0XdKuBjcADwHZfxlXVfUlWAVclubeq+nW1kDQX0vMdfFX/cpKkkWbyKkkTqKp3A+/uWXxWZ30BJw3YdpfO9CaavrOSJEmaIvuXSpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkgBIsluSi5PckmRDkl/qWZ8k70+yMcmNSQ7urLu/M706yW1J9p3P+CUtbjsMOwBJkiSNjPcBf11Vr06yI7Bzz/rDgQPa1/OBM9ufP5bkMOADwCuq6q65D1nSUuGdV0mSJJHkScCLgLMBqurBqvpeT7E1wHnVuAbYLcmenX28EPgw8Mqqun1+Ipe0VJi8SpIkCeAZwDbgo0muT/KRJE/sKbM3sKkzv7ldBrATcClwZFXd0u8ASU5Isj7J+m3bts1y+JIWO5NXSZIkQdOd7GDgzKp6LvBPwCk9ZdJnu2p/PgR8ETh+0AGqam1VrayqlcuXL5+FkCUtJSavkiRJguYu6uaq+nI7fzFNMttbZp/O/ApgSzv9CPAa4HlJTp3LQCUtTSavkiRJoqq+A2xKcmC76DDgGz3FLgPe0I46fCjw/ara2tnHA8ARwGuTDLwDK0nTMWHymuScJPckuamz7PeTfDvJDe1r9YBtVyW5tR1O/ZTO8nOTvLqdfkrbr+K42TghSZIkTdubgE8muRF4DvAnSU5McmK7fh1wB7CRZmCmN/buoKruA1YBv5NkzbxELWlJmMyjcs4FPgic17P89Kp676CNkiwDPgS8nKaJybVJLquqb3TKPBn4LLC2qj46xdglSZI0i6rqBmBlz+KzOusLOGnAtrt0pjcB+89BiJKWsAnvvFbVVcB909j3IcDGqrqjqh4ELqQZXn3MLsBngPOr6sxp7F+S5kWS3ZJcnOSWJBuS/FLP+iR5f9vK5MYkB3fW3d+ZXp3ktiT7zmf8kiRJi8FM+rye3F6knZNk9z7rxxtKHeA04OqqOn28gzikuqQR8D7gr6vqZ4FfADb0rD8cOKB9nQBs94VcksOADwCrququuQ1XkiRp8Zlu8nom8NM0fSG2An/Rp8x4Q6kDXAmsSfLU8Q7kkOqShinJk4AXAWcDVNWDVfW9nmJrgPOqcQ2wW5I9O/t4IU3fsFdW1e3zE7kkSdLiMq3ktarurqp/rapHaC7IDulTbLyh1KFpRnwmsC7JrtOJQ5LmwTOAbcBH28HlPpLkiT1lxmtpshNwKXBkVd0y6CC2MpEkSRrftJLX7h0F4FeBm/oUuxY4IMn+SXYEjqIZXv3HquoM4ArgkraMJI2aHWiec3hmVT0X+CfglJ4y47U0eQj4IjDuIyNsZSJJkjS+yTwq5wLgS8CBSTa3z+z6syRfb4dRfwnwtrbsXknWAVTVw8DJNKMJbwAuqqqbe/dfVe+guWPx8SQ+d1bSqNkMbK6qL7fzF9Mks71lBrU0eQR4DfC8JKfOZaCSJEmL2YSPyqmqo/ssPntA2S3A6s78OprngfWWO7Zn3me8ShpJVfWdJJuSHFhVtwKHAd/oKXYZzSB2FwLPB75fVVs7+3ggyRHA3yW5u6r61qGSJEkabDLPeZWkpe5NwCfb7g13AMclORGgqs6i+ZJuNbAReADY7gu5qrovySrgqiT3VtWl8xa9JEnSImDyKkkTqKobgJU9i8/qrC/gpAHb7tKZ3gTsPwchSpIkLXr2MZUkSZIkjTyTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLIc8AmSZIkabKSx85XDScOaQnyzqskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkCYAky5Jcn+TTfdYlyfuTbExyY5KDO+vu70yvTnJbkn3nK25JS4PJqyRJksa8BdgwYN3hwAHt6wTgzN4CSQ4DPgCsqqq75ipISUuTyaskSZJIsgJ4JfCRAUXWAOdV4xpgtyR7drZ/IfBh4JVVdfucByxpyTF5lSRJEsAZwNuBRwas3xvY1Jnf3C4D2Am4FDiyqm4ZdIAkJyRZn2T9tm3bZh6xpCXF5FWSJGmJS3IEcE9VXTdesT7Lqv35EPBF4PjxjlNVa6tqZVWtXL58+fSClbRkTZi8JjknyT1Jbuose0qSy9vO+Jcn2X3AtquS3Np27D+ls/zcJK/u7Ov6JMfNxglJkiRpyl4AvCrJncCFwEuTfKKnzGZgn878CmBLO/0I8BrgeUlOneNYJS1Rk7nzei6wqmfZKcAVVXUAcEU7/xhJlgEfouncfxBwdJKDeso8GfgssLaqPjrl6CVpnjgCp6TFrKreWVUrqmo/4Cjgyqp6XU+xy4A3tHXeocD3q2prZx8PAEcAr00y7h1YSZqOCZPXqroKuK9n8RrgY+30x4Aj+2x6CLCxqu6oqgdpvsVb01m/C/AZ4Pyq2m60OkkaMY7AKWnJSXJikhPb2XXAHcBGmoGZ3thbvqruo7np8TtJ1vSul6SZ2GGa2z1t7Ju2qtqa5Kl9yvTr1P/8zvxpwEeq6vTxDpTkBJqLQfbd15sVkuZfZwTOPwb+W58iPx6BE7gmyW5J9hyrJzsjcK52BE5Jo66qvgB8oZ0+q7O8gJMGbLNLZ3oTsP+cBilpSZrLAZvG69QPcCWwZkDi++gGduyXNHxn4AickiRJQzXd5PXused6tT/v6VNmvE790DQjPhNYl2TXacYhSXPKETglSZJGw3ST18uAY9rpY2juKvS6Fjggyf5JdqTp/H9Zt0BVnUEz4NMlbRlJGjWOwClJkjQCJvOonAuALwEHJtncjh73HuDlSW4DXt7Ok2SvJOsAquph4GSa0YQ3ABdV1c29+6+qd9A0t/t4Ep87K2mkOAKnJEnSaJhwwKaqOnrAqsP6lN0CrO7Mr6MZma633LE98z7jVdKCMjb6ZjuYyTqaum8j8ACwXZ1WVfclWQVcleTequrXYkWSJEkDTHe0YUlachyBU1ok0qebetX2yyRJI8VmupIkSZKkkWfyKkmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJFn8ipJkiRJGnkmr5IkSZKkkWfyKkmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJG3w7ADkCRJkhasZPtlVfMfh7QEeOdVkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXiVJkiRJI88BmyRJkiRpFDgA2Li88ypJkiSS7JPk80k2JLk5yVv6lEmS9yfZmOTGJAd31t3fmV6d5LYk+85X/JIWP++8SpIkCeBh4Leq6qtJdgWuS3J5VX2jU+Zw4ID29XzgzPbnjyU5DPgA8Iqqumt+Qpe0FHjnVZIkSVTV1qr6ajv9Q2ADsHdPsTXAedW4BtgtyZ5jK5O8EPgw8Mqqun2eQpe0RMwoeU1yZ5KvJ7khyfo+621aImlBsxmdpKUoyX7Ac4Ev96zaG9jUmd/MownuTsClwJFVdcuA/Z6QZH2S9du2bZvdoCUterNx5/UlVfWcqlrZZ123ackJNE1LHqPTtGSVTUskjaCxZnQ/BxwKnJTkoJ4y1nWSFo0kuwB/Cby1qn7Qu7rPJmOjyTwEfBE4ftC+q2ptVa2sqpXLly+flXglLR1z3WzYpiWSFjSb0UlaSpI8jiZx/WRVfapPkc3APp35FcCWdvoR4DXA85KcOqeBSlqSZpq8FvC5JNclOaHP+hk1LQGbl0gaHXPVjK7dt3WdpKFKEuBsYENVnTag2GXAG9ruEocC36+qrWMrq+oB4AjgtUkG3oGVpOmYafL6gqo6mKbJ3ElJXtSzfkZNS8DmJZJGw1w2owPrOkkj4QXA64GXtuOZ3ND21T8xyYltmXXAHcBGmhYlb+zdSVXdB6wCfifJmnmKXdISMKNH5VTVlvbnPUkuAQ4BruoUmUzTkr9JcmpV/clMYpGkuTJLzeis6ySNtKq6mv5fxnXLFHDSgHW7dKY3AfvPaoCSlrxp33lN8sT2GWAkeSLwCuCmnmI2LZG0oNmMTpIkaTTM5M7r04BLmus6dgDOr6q/HmtWUlVn0TQtWU3TtOQB4LjenVTVfUlWAVclubeqLp1BTJI028aa0X09yQ3tslOBfcG6TpIkab5MO3mtqjuAX+iz/KzOtE1LJC1oNqPTlKXPr0vV9sskSdKUzPWjciRJkiRJmrEZDdgkSZIkjRRbP0iLlsmrJEmSRp9JqbTk2WxYkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPLs8ypJkiT106+fraSh8c6rJEmSJGnkeedVkjRcjiAqSZImwTuvkiRJkqSR551XSZIkLUz2SZWWFO+8SpIkSZJGnndeJUnjs0+qJElzw9YDU2LyKkmSJEmjyi+Rf2xxJ69+0JKWglGt60Y1LkmStCDZ51WSJEmSNPIW951XSZIkab7Z8kSaEyavkiRJWlpMLqUFyWbDkiRJkqSR551XSZIkLW4+jkSafUP4uxrqndckq5LcmmRjklOGGYskzRXrOklLhfWdpLk0tOQ1yTLgQ8DhwEHA0UkOGlY8kjQXrOumKXnsS9LIs76TNNeGeef1EGBjVd1RVQ8CFwJrhhiPJM0F6zpJS4X13Xh6v5Qb9JI00DD7vO4NbOrMbwae31soyQnACe3s/UluncIx9gDu7dnh1KIcru3jX1gWcvwLOXZY6PEnU43/6XMVyiyYj7pue3Nd122//8l9ZtONa27OZ/7+TmYv/j1IFtrf9sKojx77Gc1PzFP/vRjlug4mUd/Nel03M4/9nEflGnHmcSyMv7mp87y6RuX3dXzNuc1iXTfM5LXfWWw3RnlVrQXWTusAyfqqWjmdbUeB8Q/PQo4djH/EzHldNwoW4mdmzPPDmJeUCeu7UarrFuvn7HktLIv1vGBuzm2YzYY3A/t05lcAW4YUiyTNFes6SUuF9Z2kOTXM5PVa4IAk+yfZETgKuGyI8UjSXLCuk7RUWN9JmlNDazZcVQ8nORn4LLAMOKeqbp7lw4xEs5QZMP7hWcixg/GPjHmq60bBQvzMjHl+GPMSsQDru8X6OXteC8tiPS+Yg3NL1XZdryRJkiRJGinDbDYsSZIkSdKkmLxKkiRJkkbeokxek6xKcmuSjUlOGXY8/STZJ8nnk2xIcnOSt7TLn5Lk8iS3tT9372zzzvacbk3y74cX/Y/jWZbk+iSfbucXTOwASXZLcnGSW9rP4ZcWyjkkeVv7e3NTkguSPH6UY09yTpJ7ktzUWTbleJP8YpKvt+venyyMh5wtVEnubN/vG5Ksb5eN9Oc2lZiT7Jfkn9uyNyQ5a4Ri/rX2b/yRJCt7yo/q+9w35lF5n8eJ+8/b/wM3JrkkyW6d8kN/rzX7+v1PWgwy4NpyoUtzjfOVJF9rz+sPhh3TbErP9fRi0K+unTVVtaheNAME3A48A9gR+Bpw0LDj6hPnnsDB7fSuwD8ABwF/BpzSLj8F+NN2+qD2XHYC9m/PcdmQz+G/AecDn27nF0zsbVwfA36jnd4R2G0hnAPNQ+C/CTyhnb8IOHaUYwdeBBwM3NRZNuV4ga8Av0TzLMHPAIcP+/doMb+AO4E9epaN9Oc2xZj36/5O9mwz7Jh/DjgQ+AKwsrN8lN/nQTGPxPs8TtyvAHZop/901H6nfc3J78F2/5MWw4sB15bDjmsWzivALu3044AvA4cOO65ZPL/HXE8vhle/una2XovxzushwMaquqOqHgQuBNYMOabtVNXWqvpqO/1DYANNUrKGJqmi/XlkO70GuLCqflRV3wQ20pzrUCRZAbwS+Ehn8YKIHSDJk2j+eZ0NUFUPVtX3WDjnsAPwhCQ7ADvTPEdvZGOvqquA+3oWTyneJHsCT6qqL1VTM57X2UbzZyF+boNi7msUYq6qDVV1a59VI/s+jxNzX6MQM0BVfa6qHm5nr6F5NimM8HutmRnwP2nBG+fackGrxv3t7OPa16IYcXbA9bTGsRiT172BTZ35zYz4H26S/YDn0nyT9LSq2gpNJQQ8tS02aud1BvB24JHOsoUSOzR35rcBH22banwkyRNZAOdQVd8G3gvcBWwFvl9Vn2MBxN5jqvHu3U73LtfcKeBzSa5LckK7bNQ/t6nEDLB/Wwf8bZIXtstGIeZBRvl9Hs8ovM8wcdy/TnMndSy+UXivpSnrubZc8NqmtTcA9wCXV9WiOC/6X08vBlP9HzFpQ3vO6xzq1+9kZL+dSbIL8JfAW6vqB+N0mxmZ80pyBHBPVV2X5MWT2aTPsmF/JjvQNBl6U1V9Ocn7aJoTDjIy55Cmv94ammZs3wP+b5LXjbdJn2XDfv/HMyjehXYei8ELqmpLkqcClye5ZZyyo/K5TSXmrcC+VfXdJL8I/FWSZzECMbd3hvoZ2fd5nJhH5X2GceJO8i7gYeCTbdlRea+lKem9thx2PLOhqv4VeE7bJ/2SJM+uqgXdZ3ka19MLyVT+R0zJYrzzuhnYpzO/gqZJ5chJ8jiayuWTVfWpdvHdbZOksSZV97TLR+m8XgC8KsmdNM2yX5rkEyyM2MdsBjZ3vrm7mCaZXQjn8DLgm1W1raoeAj4F/DILI/auqca7mUeb83WXa45U1Zb25z3AJTTNzUf6c5tKzG1z0O+209fR9Gl85ojEPMgov8+Dyo7E+9wev2/cSY4BjgBe2zYFhhF5r6WpGHBtuWi0Xby+AKwabiSzYtD19II3xf9rU7IYk9drgQOS7J9kR+Ao4LIhx7SddmTCs4ENVXVaZ9VlwDHt9DHApZ3lRyXZKcn+wAE0A0bMu6p6Z1WtqKr9aN7fK6vqdSyA2MdU1XeATUkObBcdBnyDhXEOdwGHJtm5/T06jKZfy0KIvWtK8bbNPX+Y5ND2vN/Q2UazLMkTk+w6Nk0zqM1NjPDnNtWYkyxPsqydfkYb8x0jEvMgo/w+Dyo/9Pd5vLiTrALeAbyqqh7obDL091qainGuLRe0tg7ZrZ1+As2X+OO1qlkQxrmeXtCm8X9tamoERqSa7RewmmaEtduBdw07ngEx/luaZkY3Aje0r9XATwJXALe1P5/S2eZd7TndyoiMbAi8mEdHG15osT8HWN9+Bn8F7L5QzgH4A5qK+ybg4zSjYY5s7MAFNE0HH6K5a3H8dOIFVrbnfDvwQSDD/j1arC+afuFfa183j9Wlo/y5TTVm4D+25b4GfBX4lRGK+Vfbv5UfAXcDn10A73PfmEfhfZ4g7o00fVtvaF9njcp77WvOfhe2+5807Jhm6bz6XlsOO65ZOK+fB65vz+sm4PeGHdMcnOOLWSSjDQ+qa2frlfYgkiRJkiSNrMXYbFiSJEmStMiYvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyeuISnJzkhcPO45hSvKrSTYluT/Jc4cdj6TZZ123tOu6JAcmuT7JD5O8eZxyZyX53XHWV5KfmZsopemzjlu8ddxE9dIM9/3aJJ8bZ/0XkvzGXBx71KWqhh3DkpPkTuA3qupvOsuObZf92ynsZz/gm8DjqurhWQ5z6JLcDvy3qrp02LEsJIv990ILh3Xd5Czlui7J2cAPquptM9xPAQdU1cbZiUyamHXc5CzlOm6uJPkC8Imq+siwY5lv3nnVQEl2GHIITwduHnIM8yLJsrkoK2li1nVDtZTPXZoX1nGzz2ux4TF5HVFJ7kzysnb6kCTrk/wgyd1JTmuLXdX+/F7bFOOXkvxEkt9J8q0k9yQ5L8mTO/t9Q7vuu0l+t+c4v5/k4iSfSPID4Nj22F9K8r0kW5N8MMmOnf1Vkjcmua1t9vWHSX663eYHSS7qlu85x76xJtkpyf3AMuBr7Td2/bb/2SSXJ7kvya1JXtNZd26S/5XkM+178/dJfirJGUn+Mckt3aYr7fvwziTfaNd/NMnjJ/iMNiQ5ojO/Q5J7kxzczv/fJN9J8v0kVyV5Vk98ZyZZl+SfgJeMc5ztyiZ5ZZqmdj9I0xTn9zubbPd70e7n19uY/zHJZ5M8fbzzk+aDdd2k6rpndeq6u5Oc2i7fqa3TtrSvM5Ls1K57cZLNSX6rPebWJMd19rm6re9+mOTbSX57Ep/VmiQ3tOd7e5JV7fK9klzWxrcxyX/pbPP77XtzXnusm5OsbNddSVP3fbD9XJ85zrHPTfJHnfn/3p7TliS/3lm+Yxvjm9r5ZWnq/9+b6PykuWAdtzDquM7+Tk1zLXdnktd21ve7FuutlwbVkU9OcnYb47eT/FEmSH6THJvk6s78y9Ncu34/yQeBdNadmeTizvyfJrkiSViMqsrXPL+AO4GX9Sw7Fri6XxngS8Dr2+ldgEPb6f2AAnbobPfrwEbgGW3ZTwEfb9cdBNwP/FtgR+C9wEOd4/x+O38kzRcbTwB+ETgU2KE93gbgrZ3jFXAZ8CTgWcCPgCva4z8Z+AZwzID3YWCsnX3/zIBtnwhsAo5rYzsYuBd4Vrv+3Hb+F4HHA1fSNMl5A00l+kfA53ve75uAfYCnAH8P/NEEn+PvAZ/szL8SuKXn/HYFdgLOAG7orDsX+D7wgva9fvw4x9muLPBi4N+08z8P3A0cOc7vxZHte/1z7fv1O8AXh/234Gtxv7CumzDWzr4H1XW7AluB32r/9ncFnt+u+x/ANcBTgeXAF4E/bNe9GHi4LfM4YDXwALB7u34r8MJ2enfg4Ak+y0No6qGXt+/Z3sDPtuv+FvhfbXzPAbYBh3Xe639pj78M+J/ANZ39foGmieVEv0vn0tbJwCqaOu/ZNP8Lzu++h+3yf6Sp797VvkfLhv334GvxvbCOmzDWzr5HvY4b299pNNdt/w74J+DAdv25bH8tdi6P1kvj1ZF/BfxvmvrqqcBXgN+cIJ5jaX+PgD2AHwCvbs/1bW2sv9Gu3xn4h3abF9Jc/64Y9t/HnP3dDTuApfiiqcjuB77XeT3A4MruKuAPgD169rMf21d2VwBv7MwfSFOB7UCTbF3QWbcz8CCPreyumiD2twKXdOYLeEFn/jrgHZ35vwDOGLCvgbF29j2osvtPwN/1LPvfwLvb6XOBD3fWvQnY0Jn/N8D3et7vEzvzq4HbJ3gvfgb4IbBzO/9J4PcGlN2tPZ8nd+I7b5K/LxOWpUmOTx/n9+IzwPGd+Z9of+eePuy/B1+L92VdN3GsnX0PquuOBq4fsO52YHVn/t8Dd7bTLwb+uec9u4dHL5bvAn4TeNIkP8v/PVbH9CzfB/hXYNfOsv8JnNt5r/+ms+4g4J87819g6snrOcB7Ouue2fse0lwI30KTxB4w7L8FX4vzZR03caydfY96HfdimoTwiZ1lFwG/206fS8+1WE+9NKiOfBrNFwFP6Dnnz08Qz7E8mry+gcd+6RdgM526kyZ5vg/4FnD0sP825vJls+HhObKqdht7AW8cp+zxNP+cb0lybTpNVfvYi+YXd8y3aCq6p7XrNo2tqKoHgO/2bL+pO5PkmUk+nab56w+AP6H5Bqjr7s70P/eZ32UasU7k6cDz2+Yv30vyPeC1wE/NIK7uuX+rjW+gagYG2QD8SpKdgVfR3AEYa6r2nrbZyA9o/nnBY9+7x7zXE+j9XJ6f5PNJtiX5PnAi238uXU8H3td5r+6jqfz2nkIM0nRY182srtuH5gJusvvt1lvfrccO/vJAJ8b/SPMl3beS/G3a7gXTiGMv4L6q+mFPHN265Ts9MTw+M+uD95jPl8e+B2M+RpMQrKuq22ZwLGki1nGLo44D+Meq+qdxjjfeddug83g6zd3SrZ1rsP9Ncwd2sno/7+qNpaq+AtxBc2130RT2veCYvC4AVXVbVR1N84v+p8DFSZ5I801Wry00fyhj9qX5JulumiYUK8ZWJHkC8JO9h+uZP5Pm2+sDqupJwKl02tnP0HixTmQT8LfdfxhVtUtV/dcZxLNPTyxbJrHNBTTfoK0BvlGPjnT5n9tlL6NpbrNfu7z73vX7/AbpLXs+TfOefarqycBZnX332+8mmiYq3ffrCVX1xSnEIM0p67q+NgE/PYX9Tqbeoqqurao1NO/1XzHxxc6gOLYAT0mya08c355MHNO0le3r617/C/g08O+TTHrUV2kuWcf1NSp1HMDu7ecx6HjjXbcNOo9NNHde9+hcfz2pqp7Vp+wgj6nz2r6s3TqQJCfRNHfeArx9CvtecExeF4Akr0uyvKoeoWmSAk0zrW3AIzR9DMZcALwtyf5JdqH5Zu3/tN9MXUxzl/CX03S6/wMmrrh2pWlnf3+SnwVmkhz2Gi/WiXwaeGaS1yd5XPt6XpKfm0E8JyVZkeQpNJX6/5nENhcCr6B5X87vLN+VprL6Lk1znj+ZQVz97Epzt+NfkhxCkyyP6fd7cRbwzrSDRrWDB/zaLMckzYh1XV+fBn4qyVvTDF6ya5Lnd/b7O0mWJ9mDpinhJybaYZpBjV6b5MlV9RDNef/rBJudDRyX5LA0g7PsneRnq2oTTT+0/5nk8Ul+nubu0icncW7TdRHNADQHta1e3t1dmeT1NP37jgXeDHysfd+lobKO62tU6rgxf9Bu/0LgCOD/TnK7QXXkVuBzwF8keVK77qeT/LtJ7hfg/wHPSvIf2lYrb6bT0jDNYHd/BLwOeD3w9iTPmcL+FxST14VhFXBzmhHb3gccVVX/0jYT+WPg79umCIfS9AX6OE2/im/SDJTxJoCqurmdvpDmW5wf0vQP+NE4x/5tmsToh8CHmVxCN1kDY51I20TtFcBRNN8yfYfmW8ydZhDP+TQVzB3t64/GLw5tpfQl4Jd57HtzHk1zk2/TDHJwzQzi6ueNwP9I8kOayvzH3yj2+72oqkto3p8L2+ZCNwGHz3JM0kxZ1/Vo67qXA79CU8/dxqOjk/8RsB64Efg68FUmUW+1Xg/c2dYHJ9Jc9IwXx1doBsg7nWZQkr/l0TsiR9O0LtkCXEIz9sDlk4xjyqrqMzT9/K+kGSTmyrF1SfZt172hqu6vqvNp3qPT5yoeaQqs43qMSh3X+g5NP/ktNF/AnVhVt0zyPMarI99AM7DWN9r9XwzsOcnzoKruBX4NeA/NTZEDaAYWpU1mPwH8aVV9re0mcSrw8bQjMy82aZpNaylqvx37Hk0Tkm8OOZyhSp8HjUtaHKzrJC1m1nEzl+TFwCeqasUERTVk3nldYpL8SpKd2zb976X5JuvO4UYlSbPLuk7SYmYdp6XK5HXpWUPTHGILTbODo8rb7wOleVj1/X1en5nl49w84DivnXhrSX1Y103RfNV34xzfelCaPOu4KRp2HdcnnrMGxHPWMOJZKGw2LEmSJEkaed55lSRJkiSNvJk8JHze7bHHHrXffvsNOwxJI+a66667t6qWDzuO2WJdJ6kf6zpJS8F4dd2CSl73228/1q9fP+wwJI2YJN8adgyzybpOUj9zXdcleRvwG0DRDAB0XFX9S2d9aB7xshp4ADi2qr7arru/qnZpp1e35Q6rqrsGHc+6TlI/49V1NhuWJEla4pLsDbwZWFlVzwaW0TxLvetwmsGBDgBOAM7ss5/DgA8Aq8ZLXCVpOkxeJUmSBE2LvCck2QHYmWYk2641wHnVuAbYLcmeYyuTvBD4MPDKqrp9voKWtHSYvEqSJC1xVfVtmueF3gVsBb5fVZ/rKbY3sKkzv7ldBrATcClwZFXdMug4SU5Isj7J+m3bts1a/JKWBpNXSZKkJS7J7jR3VvcH9gKemOR1vcX6bDr2zMWHgC8Cx493nKpaW1Urq2rl8uWLZuwpSfPE5FWSJEkvA75ZVduq6iHgU8Av95TZDOzTmV/Bo02LHwFeAzwvyalzHaykpcnkVZIkSXcBhybZuR1V+DBgQ0+Zy4A3pHEoTdPirWMrq+oB4AjgtUnGvQMrSdOxoB6VI0mSpNlXVV9OcjHwVeBh4HpgbZIT2/VnAetoHpOzkeZROcf12c99SVYBVyW5t6ouna9zkLT4mbxKkiSJqno38O6exWd11hdw0oBtd+lMb6LpOytJs8pmw5IkSZKkkeedV0nDlT6DV1Ztv0zS7PPvT5KmrrfutN6cNxPeeU1yTpJ7ktzUs/xNSW5NcnOSPxuw7aq2zMYkp3SWn5vk1e30U5Jcn2S7fhOSJEmSJMHkmg2fC6zqLkjyEppngf18VT2L5qHW9JRZBnwIOBw4CDg6yUE9ZZ4MfBZYW1Ufnc4JSJIkSZIWvwmT16q6CrivZ/F/Bd5TVT9qy9zTZ9NDgI1VdUdVPQhcSJPwjtkF+AxwflWdOZ3gJUmSJElLw3QHbHom8MIkX07yt0me16fM3sCmzvzmdtmY04Crq+r08Q6U5IQk65Os37Zt2zTDlSRJkiQtZNNNXncAdgcOBf47cFH7QOuuPqNA0O3NfCWwJslTxztQVa2tqpVVtXL58uXTDFeSJEmStJBNN3ndDHyqGl8BHgH26FNmn878CmBLZ/5C4ExgXZJdpxmHJEmSJGkJmG7y+lfASwGSPBPYEbi3p8y1wAFJ9k+yI3AUcFm3QFWdAVwBXNKWkSRJkiRpO5N5VM4FwJeAA5NsTnI8cA7wjPbxORcCx1RVJdkryTqAqnoYOJlmNOENwEVVdXPv/qvqHTR9Yz+eZLrJtCRJkiRpEdthogJVdfSAVa/rU3YLsLozvw5Y16fcsT3zPuNVkiRJkjSQdzolSZIkSSPP5FWSJEmSNPJMXiVJkiRJI8/kVZIkSZI08kxeJUmSJEkjz+RVkoAk5yS5p30E2NiyP09yS5Ibk1ySZLcB265KcmuSjUlO6Sw/N8mr2+mnJLk+iaOrS5IkTYPJqyQ1zgVW9Sy7HHh2Vf088A/AO3s3SrIM+BBwOHAQcHSSg3rKPJnmmddrq+qjsx+6JEnS4mfyKklAVV0F3Nez7HNV9XA7ew2wos+mhwAbq+qOqnoQuBBY01m/C/AZ4PyqOnP2I5ckSVoaTF4laXJ+nSYJ7bU3sKkzv7ldNuY04OqqOn28nSc5Icn6JOu3bds242AlSZIWG5NXSZpAkncBDwOf7Le6z7LqTF8JrEny1PGOUVVrq2plVa1cvnz59IOVJElapHYYdgCSNMqSHAMcARxWVdWnyGZgn878CmBLZ/5C4GpgXZKXVNUP5yxYSZKkRcw7r5I0QJJVwDuAV1XVAwOKXQsckGT/JDsCRwGXdQtU1RnAFcAlbRlJGjlJDkxyQ+f1gyRv7SmTJO9vR1e/McnBnXX3d6ZXJ7ktyb7zeAqSFjmTV0kCklwAfAk4MMnmJMcDHwR2BS5vL+TOasvulWQdQDug08k0owlvAC6qqpt7919V76DpG/vxJNa9kkZOVd1aVc+pqucAvwg8AFzSU+xw4ID2dQKw3UB0SQ4DPgCsqqq75jRoSUuKzYYlCaiqo/ssPntA2S3A6s78OmBdn3LH9sz7jFdJC8VhwO1V9a2e5WuA89puFNck2S3JnlW1FSDJC4EPA6ur6vb5DVnSYue3/5IkSep1FHBBn+XjjbC+E3ApcGRV3dJvp46sLmkmTF4lSZL0Y23f/FcB/7ff6j7Lxgazewj4InD8oH07srqkmTB5lSRJUtfhwFer6u4+68YbYf0R4DXA85KcOrchSlqKTF4lSZLUdTT9mwxDM5r6G9pRhw8Fvj/W3xWgHZn9COC17cB3kjRrHLBJkiRJACTZGXg58JudZScCVNVZNIPTrQY20oxGvN1AdFV1X/uosauS3FtVl85H7JIWP5NXSZIkAT++c/qTPcvO6kwXcNKAbXfpTG8C9p+jMCUtURM2G05yTpJ7ktzUZ91vJ6kkewzYdlWSW9sHWZ/SWX5ukle3009Jcn0SHyEhSZIkSeprMn1ezwVW9S5Msg9Ns5K+D59Osgz4EE2n/4OAo5Mc1FPmycBngbVV9dEpRS5JkiRJWjImTF6r6irgvj6rTgfezqPDo/c6BNhYVXdU1YPAhTQPth6zC/AZ4PyqOnNKUUuSJEmSlpRpjTac5FXAt6vqa+MUG+8h1gCnAVdX1ekTHMuHWUuSJEmaf8n2Lw3NlJPXdhS6dwG/N1HRPsu6d2mvBNYkeep4O/Fh1pIkSZKk6dx5/Wma0eO+luROmodTfzXJT/WUG+8h1tA0Iz4TWJdk12nEIUmSJElaIqacvFbV16vqqVW1X1XtR5OkHlxV3+kpei1wQJL9k+wIHEXzYOvuvs4ArgAuactIkiRJkrSdyTwq5wLgS8CBSTYnOX6csnslWQdQVQ8DJ9OMJrwBuKiqbu7dpqreQdM39uNJptUHV5IkSZK0uO0wUYGqOnqC9ft1prcAqzvz64B1fbY5tmfeZ7xKkiRJkgbyTqckSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZIkaeSZvEqSJEmSRp7JqyRJkiRp5Jm8SpIkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskAUnOSXJPkps6y56S5PIkt7U/dx+w7aoktybZmOSUzvJzk7y6s6/rkxw392cjSdOTZLckFye5JcmGJL/Usz5J3t/WdzcmObiz7v7O9Oq27tx3PuOXtLiZvEpS41xgVc+yU4ArquoA4Ip2/jGSLAM+BBwOHAQcneSgnjJPBj4LrK2qj85+6JI0a94H/HVV/SzwC8CGnvWHAwe0rxOAM3t3kOQw4APAqqq6a27DlbSUmLxKElBVVwH39SxeA3ysnf4YcGSfTQ8BNlbVHVX1IHBhu92YXYDPAOdX1XYXeZI0KpI8CXgRcDZAVT1YVd/rKbYGOK8a1wC7Jdmzs48XAh8GXllVt89P5JKWCpNXSRrsaVW1FaD9+dQ+ZfYGNnXmN7fLxpwGXF1Vp493oCQnJFmfZP22bdtmGLYkTcszgG3AR9tuDh9J8sSeMuPVeTsBlwJHVtUt/Q5gXSdpJkxeJWlm0mdZdaavBNYk6Zf4PrpB1dqqWllVK5cvXz6rAUrSJO0AHAycWVXPBf6J7btLjFfnPQR8ETh+0AGs6yTNhMmrJA1291hzuPbnPX3KbAb26cyvALZ05i+k6RO2LsmucxWoJM2CzcDmqvpyO38xTTLbW2ZQnfcI8BrgeUlOnctAJS1NJq+SNNhlwDHt9DE0zeF6XQsckGT/JDsCR7Xb/VhVnUEz4NMlbRlJGjlV9R1gU5ID20WHAd/oKXYZ8IZ21OFDge+Pda9o9/EAcATw2iQD78BK0nSYvEoSkOQC4EvAgUk2txdd7wFenuQ24OXtPEn2SrIOoKoeBk6mGU14A3BRVd3cu/+qegdNP7GPJ7HulTSq3gR8MsmNwHOAP0lyYpIT2/XrgDuAjTQDM72xdwdVdR/N6O2/k2RN73pJmq4dJiqQ5Byab9Duqapnt8v+HPgV4EHgduC4PqPRkWQVzZDry4CPVNXYhd+5wKer6uIkT6G5I/F+HyEhaViq6ugBqw7rU3YLsLozv47mgq633LE98z7jVdJIq6obgJU9i8/qrC/gpAHb7tKZ3gTsPwchSlrCJvPt/7ls/+zDy4FnV9XPA/8AvLN3I599KEmSJM1Qsv1LWqImTF77Pfuwqj7XNpUDuIams34vn30oSZIkSZoVs9Hv6tdpktBePvtQkiRJkjQrZpS8JnkX8DDwyX6r+yzz2YeSJEmSpCmbcMCmQZIcQzOQ02Ft5/1ek3n24dU0zz58SVX9cLqxSJIkSZIWt2ndeW1HEX4H8Kr2eV79+OxDSZIkSdKsmDB5HfDsww8CuwKXJ7khyVltWZ99KEmSJEmadRM2Gx7w7MOzB5T12YeSJEmSpFnnnU5JkiRJ0sgzeZUkSZIkjTyTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLIM3mVJEmSJI08k1dJkiRJ0sgzeZUkSZIkjTyTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLI22HYAUiSJElaQpLtl1XNfxxacLzzKkmSJEkaeSavkiRJAiDJnUm+nuSGJOv7rE+S9yfZmOTGJAd31t3fmV6d5LYk+85X7JIWP5sNS5IkqeslVXXvgHWHAwe0r+cDZ7Y/fyzJYcAHgFdU1V1zGaikpcU7r5IkSZqsNcB51bgG2C3JnmMrk7wQ+DDwyqq6fVhBSlqcTF4lSZI0poDPJbkuyQl91u8NbOrMb26XAewEXAocWVW39Nt5khOSrE+yftu2bbMZt6QlwORVkiRJY15QVQfTNA8+KcmLetb3GSaWsWFiHwK+CBw/aOdVtbaqVlbVyuXLl89KwJKWDpNXSZIkAVBVW9qf9wCXAIf0FNkM7NOZXwFsaacfAV4DPC/JqXMcqqQlyORVkiaQ5G1Jbk5yU5ILkjy+Z72jb0pa8JI8McmuY9PAK4CbeopdBryhrfcOBb5fVVvHVlbVA8ARwGuTDLwDK0nTMWHymuScJPckuamz7ClJLm8vwi5PsvuAbVclubW9oDuls/zcJK/u7Ov6JMfNxglJ0mxKsjfwZmBlVT0bWAYc1VOsO/rmCTSjb/buZ2z0zVWOvilpRD0NuDrJ14CvAP+vqv46yYlJTmzLrAPuADbSDMz0xt6dVNV9wCrgd5KsmZ/QJS0Fk3lUzrnAB4HzOstOAa6oqve0SekpwDu6GyVZBnwIeDlNE5Nrk1xWVd/olHky8FlgbVV9dCYnIklzaAfgCUkeAnbm0SZyY348+iZwTZLdkuw5djeiM/rmakfflDSqquoO4Bf6LD+rM13ASQO236UzvQnYfw7ClLSETXjntaquAu7rWbwG+Fg7/THgyD6bHgJsrKo7qupB4MJ2uzG7AJ8Bzq+q7e5SSNIoqKpvA+8F7gK20jSR+1xPsRmNvgmOwDmSku1fkiRpaKbb5/VpY3cU2p9P7VNmvIs5gNOAq6vq9PEO5AWdpGFqu0WsobmDsBfwxCSv6y3WZ9NJj74JjsApSZI0kbkcsGm8izmAK4E1Sfolvo9u4AWdpOF6GfDNqtpWVQ8BnwJ+uaeMo29KkiTNsekmr3cn2ROg/XlPnzLjXcxB04z4TGDd2Mh2kjSC7gIOTbJzkgCHARt6yjj6piRJ0hybbvJ6GXBMO30MTX+uXtcCByTZP8mONKNzXtYtUFVnAFcAl7RlJGmkVNWXgYuBrwJfp6k31zr6piRJ0vyacLThJBcALwb2SLIZeDfwHuCi9g7CXcCvtWX3Aj5SVaur6uEkJ9OMJrwMOKeqbu7df1W9I8lHgY8nObqqHpmlc5OkWVFV76ap+7ocfVOSJGkeTZi8VtXRA1Yd1qfsFmB1Z34dzR2J3nLH9sz7jFdJkiRJ0kBzOWCTJEmSJEmzwuRVkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXiVJkiRJI8/kVZIkSZI08iZ8VI6WkGT7ZVXzH4ckSZKk2bfAr/e98ypJkiRJGnkmr5IkSZKkkWfyKkmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJFn8ipJkiRJGnkmr5IkSZKkkWfyKkmSJACSLEtyfZJP91mXJO9PsjHJjUkO7qy7vzO9OsltSfadr7glLQ0mr5IkSRrzFmDDgHWHAwe0rxOAM3sLJDkM+ACwqqrumqsgJS1NJq+SpKUneexLEklWAK8EPjKgyBrgvGpcA+yWZM/O9i8EPgy8sqpun/OAJS05Jq+SJEkCOAN4O/DIgPV7A5s685vbZQA7AZcCR1bVLYMOkOSEJOuTrN+2bdvMI5a0pJi8SpIkLXFJjgDuqarrxivWZ1m1Px8CvggcP95xqmptVa2sqpXLly+fXrCSlqwZJa9J3pbk5iQ3JbkgyeN71tuxX5IkafS9AHhVkjuBC4GXJvlET5nNwD6d+RXAlnb6EeA1wPOSnDrHsUpaoqadvCbZG3gzsLKqng0sA47qKWbHfkmSpBFXVe+sqhVVtR/N9dyVVfW6nmKXAW9ob04cCny/qrZ29vEAcATw2iTj3oGVpOnYYRa2f0KSh4CdefTbtzE/7tgPXJNktyR7jlV0nY79q+3YL0mSNFqSnAhQVWcB64DVwEbgAeC43vJVdV+SVcBVSe6tqkvnM15Ji9u0k9eq+naS9wJ3Af8MfK6qPtdTbFDH/q082rH/xRN17Ke5a8u++9qqWJIkaS5V1ReAL7TTZ3WWF3DSgG126UxvAvaf0yAlLUkzaTa8O82d1f2BvYAnJultXmLHfkmSJEnSjM1kwKaXAd+sqm1V9RDwKeCXe8rYsV+SJEmSNGMzSV7vAg5NsnOSAIcBG3rK2LFfWiqSx74kSZKkWTTt5LWqvgxcDHwV+Hq7r7VJThzr3E/Tsf8Omo79Hwbe2Gc/9wGrgN9Jsma68UjSXGkHm7s4yS1JNiT5pZ71PhZMkiRpjs1otOGqejfw7p7FduyXtNi8D/jrqnp1kh1pRlfv6j4W7Pk0jwV7frdA57Fgr/CxYJIkSVM3k2bDkrToJXkS8CLgbICqerCqvtdT7MePBauqa4DdkuzZ2cfYY8Fe6WPBJEkjr7crkN2BNCJMXiVpfM8AtgEfTXJ9ko8keWJPmUGPBYNHHwt25ESPBUuyPsn6bdu2zWL4kiRJi4PJqySNbwfgYODMqnou8E/AKT1lfCyYJEnSHDN5laTxbQY2t4PUQTNQ3cF9yvhYMEmSpDlk8ipJ46iq7wCbkhzYLjoM+EZPMR8LNln2o5IkSdM0o9GGJWmJeBPwyXak4TuA48YeCVZVZ9E8Fmw1zWPBHgCO691BVd2XZBVwVZJ7q+rSeYtekiRpETB5laQJVNUNwMqexT4WTJIkaR7ZbFiSJEmSNPJMXiVJkiRJI8/kVZIkSZI08kxeJUmSJEkjz+RVkiRJkjTyTF4lSZIkSSPP5FWSJEmSNPJMXiVJkiRJI8/kVZIkSZI08kxeJUmSRJLHJ/lKkq8luTnJH/QpkyTvT7IxyY1JDu6su78zvTrJbUn2na/4JS1+Oww7AEmSJI2EHwEvrar7kzwOuDrJZ6rqmk6Zw4ED2tfzgTPbnz+W5DDgA8Arququ+Qld0lLgnVdJkiRRjbG7p49rX9VTbA1wXlv2GmC3JHuOrUzyQuDDwCur6vb5iFvS0mHyKkmSJACSLEtyA3APcHlVfbmnyN7Aps785nYZwE7ApcCRVXXLgP2fkGR9kvXbtm2b3eAlLXozSl6T7Jbk4iS3JNmQ5Jd61tsvQpIkaYGoqn+tqucAK4BDkjy7p0j6bdb+fAj4InD8OPtfW1Urq2rl8uXLZyNkSUvITO+8vg/466r6WeAXgA0967v9Ik6g6RfxGJ1+EavsFyFJkjR8VfU94AvAqp5Vm4F9OvMrgC3t9CPAa4DnJTl1jkOUtARNO3lN8iTgRcDZAFX1YFvRddkvQpIkaQFIsjzJbu30E4CXAb3Nfy8D3tC2rjsU+H5VbR1bWVUPAEcAr00y8A6sJE3HTEYbfgawDfhokl8ArgPeUlX/1CkzqF/EVh7tF/HiQf0ioOkbQXPXln33tVWxJEnSHNkT+FiSZTQ3OC6qqk8nORGgqs4C1gGrgY3AA8BxvTupqvuSrAKuSnJvVV06b2cgaVGbSfK6A3Aw8Kaq+nKS9wGnAL/bKTPZfhFvGXSQqloLrAVYuXJl74h3kiRJmgVVdSPw3D7Lz+pMF3DSgO136UxvAvafgzAlLWEz6fO6GdjcGYXuYppktreM/SIkSZIkSTMy7eS1qr4DbEpyYLvoMOAbPcXsFyFJkiRJmrGZNBsGeBPwySQ7AncAx9kvQpIkSZI022aUvFbVDcDKnsX2i5AkSZIkzaqZPudVkiRJkqQ5Z/IqSZIkSRp5Jq+SJEmSpJFn8ipJkiRJGnkmr5IkSZKkkWfyKkmTkGRZkuuTfLrPuiR5f5KNSW5McnBn3f2d6dVJbkuy73zFLUmStFiYvErS5LwF2DBg3eHAAe3rBODM3gJJDgM+AKyqqrvmKkhJkgZKtn9JC4jJqyRNIMkK4JXARwYUWQOcV41rgN2S7NnZ/oXAh4FXVtXtcx6wJEnSImTyKkkTOwN4O/DIgPV7A5s685vbZQA7AZcCR1bVLYMOkOSEJOuTrN+2bdvMI5YkSVpkTF4laRxJjgDuqarrxivWZ1m1Px8CvggcP95xqmptVa2sqpXLly+fXrCSJEmLmMmrJI3vBcCrktwJXAi8NMknespsBvbpzK8AtrTTjwCvAZ6X5NQ5jlWSJGnRMnmVpHFU1TurakVV7QccBVxZVa/rKXYZ8IZ21OFDge9X1dbOPh4AjgBem2TcO7CSJEnqb4dhByBJC1GSEwGq6ixgHbAa2Ag8ABzXW76q7kuyCrgqyb1Vdel8xitJkrTQmbxK0iRV1ReAL7TTZ3WWF3DSgG126UxvAvaf0yAlSZIWKZsNS5IkSZJGnsmrJEmSJGnkmbxKkiRJkkaeyaskSZJIsk+SzyfZkOTmJG/pUyZJ3p9kY5IbkxzcWXd/Z3p1ktuS7Dtf8Uta/BywSZIkSQAPA79VVV9NsitwXZLLq+obnTKHAwe0r+cDZ7Y/fyzJYcAHgFdU1V3zE7qkpcA7r5IkSaKqtlbVV9vpHwIbgL17iq0BzqvGNcBuSfYcW5nkhcCHgVdW1e3zFLqkJWLGyWuSZUmuT/LpPutsWiJJkrTAJNkPeC7w5Z5VewObOvObeTTB3Qm4FDiyqm4ZsN8TkqxPsn7btm2zG7SkRW827ry+heabuX66TUtOoGla8hidpiWrbFoiSZI0XEl2Af4SeGtV/aB3dZ9Nqv35EPBF4PhB+66qtVW1sqpWLl++fFbilbR0zCh5TbICeCXwkQFFbFoiSZK0QCR5HE3i+smq+lSfIpuBfTrzK4At7fQjwGuA5yU5dU4DlbQkzfTO6xnA22kqq35m1LQEbF4iSZI0H5IEOBvYUFWnDSh2GfCGtmvYocD3q2rr2MqqegA4AnhtkoF3YCVpOqadvCY5Arinqq4br1ifZZNuWgI2L5EkSZonLwBeD7w0yQ3ta3WSE5Oc2JZZB9wBbKRpPffG3p1U1X3AKuB3kqyZp9glLQEzeVTOC4BXJVkNPB54UpJPVNXrOmUm07Tkb5KcWlV/MoNYJEmSNANVdTX9bzx0yxRw0oB1u3SmNwH7z2qAkpa8ad95rap3VtWKqtoPOAq4sidxBZuWSJIkSZJmwUzuvPY11qykqs6iaVqymqZpyQPAcb3lq+q+JKuAq5LcW1WXznZM0lClz5fYVdsvkyRJkjTQrCSvVfUF4Avt9Fmd5TYtkSRJkiTN2Gw851WSJEmSpDll8ipJkiRJGnkmr5IkSZKkkWfyKkmSJEkaeSavkiRJkqSRZ/IqSZIkSRp5Jq+SJEmSpJE3K895lSSJ5LHzVcOJQ5IkLUreeZUkSZIkjTyTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLIM3mVpHEk2SfJ55NsSHJzkrf0KZMk70+yMcmNSQ7urLu/M706yW1J9p2v+CVJkhYLRxuWpPE9DPxWVX01ya7AdUkur6pvdMocDhzQvp4PnNn+/LEkhwEfAF5RVXfNT+iSJEmLh3deJWkcVbW1qr7aTv8Q2ADs3VNsDXBeNa4Bdkuy59jKJC8EPgy8sqpun6fQJUmSFhWTV0mapCT7Ac8Fvtyzam9gU2d+M48muDsBlwJHVtUt4+z7hCTrk6zftm3b7AUtSZK0SJi8StIkJNkF+EvgrVX1g97VfTap9udDwBeB48fbf1WtraqVVbVy+fLlM45XkiRpsTF5laQJJHkcTeL6yar6VJ8im4F9OvMrgC3t9CPAa4DnJTl1TgOVJElaxExeJWkcSQKcDWyoqtMGFLsMeEM76vChwPerauvYyqp6ADgCeG2Sce/AStKwJDknyT1Jbhqw3pHVJQ2Vow1L0vheALwe+HqSG9plpwL7AlTVWcA6YDWwEXgAOK53J1V1X5JVwFVJ7q2qS+chdkmainOBDwLnDVjvyOqShmrayWuSfWgqt5+iaRa3tqre11MmwPtoLuoeAI4dG7Uzyf1VtUs7vbotd5gVnaRRUlVX079Pa7dMAScNWLdLZ3oTsP+sBihJs6SqrmoHphvkxyOrA9ck2S3JnmMtTTojq692ZHVJc2EmzYbHnn34c8ChwElJDuop0/2G7gSab+geo/MN3SoTV0mSpJHlyOqShmrayavPPpQkSVpSHFld0lDNyoBNPvtQkiRp0XNkdUlDNePk1WcfSpIkLQmOrC5pqGY02vAsPfvwb5KcWlV/MpNYJEmSNH1JLgBeDOyRZDPwbuBx4MjqkkbDTEYbnuyzD09OciHNUOrbfUOX5Ajg75LcXVVnTzceSZIkTV9VHT3BekdWlzRUM7nz6rMPJUmSJEnzYtrJq88+BNLn9Ku2XyZJkqSZ8bpLWvJmZbRhSZIkSZLm0owGbJIkSZI0z7wLrSXKO6+SJEmSpJFn8ipJkiRJGnkmr5IkSZKkkWfyKkmSJEkaeSavkiRJkqSRt7hHG3YkNkmSJElaFLzzKkmSJEkaeYv7zqskSVq8bGGl6fJ3R1qQTF4lSZqJ3otgL4AlSZoTJq+SJGlp8a6bJC1IJq8an//gJUmSBvNaSZo3Jq+SJEmSNAr8MmRcJq8aPf7RSpKkparfdZAWLz/vKTF5lSRJU+dAVVPjF7PDZYIwP/w91xwzeZUkaTZ58SZJC9N0v+SY7Hb+L5gxk1dJkhY6E2Zp5vw7kh41on8PJq+StBiN6D8dSVoSrINHi5/HomHyKklLhX0Uh2c2m5TZd0+SNKrm+IuCn5i1PU1DklVJbk2yMckpw4xlQsn2L0mahDmv66yfNKom+7vp7/CisaCu7SQtOENLXpMsAz4EHA4cBByd5KBhxSNJc8G6TtJSYX2nBcUvzRakYd55PQTYWFV3VNWDwIXAmiHGI0lzwbpO0lJhfTcekyVpxobZ53VvYFNnfjPw/N5CSU4ATmhn709y64yOOrOKYg/g3jnc/3yZ+DzGM4xz3P6YMzuHYXv0fBb2eTxq+/OYye/J1Ld9+vQPNueGU9c9Vv/fs7n+W57s/icfx9z9vcx+rOMZfB6z+ZnM9SMfes9jtt/DuY9/THMeo/G/bSKjXNfBJOq7Ebuum8geJHP7P3ru/uYnX1+Oyt9kU24hXxfNXZ4wKr/n8xf/wLpumMlrv7PYrjdvVa0F1s59OBNLsr6qVg47jplaDOexGM4BPI8lYuh13WL5fDyP0eJ5qI8J67tRuq6byEL+3ViosS/UuGHhxr7Q4h5ms+HNwD6d+RXAliHFIklzxbpO0lJhfSdpTg0zeb0WOCDJ/kl2BI4CLhtiPJI0F6zrJC0V1neS5tTQmg1X1cNJTgY+CywDzqmqm4cVzyQtiGYuk7AYzmMxnAN4HoveiNR1i+Xz8TxGi+ehxxiR+m42LeTfjYUa+0KNGxZu7Asq7pQPqZckSZIkjbhhNhuWJEmSJGlSTF4lSZIkSSPP5HWKkvxhkhuT3JDkc0n2GnZMU5Xkz5Pc0p7HJUl2G3ZM05Hk15LcnOSRJAtmiO8xSVYluTXJxiSnDDue6UhyTpJ7ktw07Fi0vSS/n+TbbX11Q5LVnXXvbH/3bk3y74cZ52Ql+e0klWSPdn6/JP/cOb+zhh3jZPSeR7tsQXweg/4HLrTPYrz/5Qvls9D0JXlLkpvaa4i3tssmdW2U5M4kX29/d9bPZ9zt8fvFPqlr02Ffd8ww9qG97/3i7qzbrj7vWT9y73ln3USxD/V3faCq8jWFF/CkzvSbgbOGHdM0zuEVwA7t9J8CfzrsmKZ5Hj8HHAh8AVg57HimGPsy4HbgGcCOwNeAg4Yd1zTO40XAwcBNw47FV9/P5/eB3+6z/KD2d24nYP/2d3HZsOOd4Fz2oRkE5lvAHu2y/Rba796A81gwn8eg/4EL7bMY5zwWzGfha9qf/bOBm4CdaQYu/RvggMleGwF3jv3tjlDsE16bDvu6YyaxD/N9HxR3u267+nwhvOeTiX2Y7/lEL++8TlFV/aAz+0R6Hr69EFTV56rq4Xb2GprnsC04VbWhqm4ddhzTdAiwsaruqKoHgQuBNUOOacqq6irgvmHHoSlbA1xYVT+qqm8CG2l+J0fZ6cDbWYB1bo9+57FgPo/F8D8Qxj2PBfNZaNp+Drimqh5or4X+FvjVBXJtNCj2yfxdDvu6YyaxD1PfuNt1E/1fGsn3vF23YP+nmrxOQ5I/TrIJeC3we8OOZ4Z+HfjMsINYgvYGNnXmN7fLpNl2ctsk65wku7fLFtTvX5JXAd+uqq/1Wb1/kuuT/G2SF853bFMxznkstM9j0P/ABfNZwMDzWFCfhablJuBFSX4yyc7Aapq7UF3jXRsV8Lkk1yU5YQ7j7Gdg7JO4Nh327/ZMYofhve99457g/9KYkXzPJxk7DPd3faChPed1lCX5G+Cn+qx6V1VdWlXvAt6V5J3AycC75zXASZjoHNoy7wIeBj45n7FNxWTOY4FKn2UL7tsvDd94fyPAmcAf0vxu/SHwFzQXZSP3+zfBeZxK06Sv11Zg36r6bpJfBP4qybN6vsmfV9M8j5H6PKb5P3BBfRbjnMdIfRaafVW1IcmfApcD99M05Ry74zqZa6MXVNWWJE8FLk9yS9sKac6NF/skrk2H+rs9w9hhSO/7OHG/i/71edeovueTiR2G+Ls+HpPXPqrqZZMsej7w/xjB5HWic0hyDHAEcFi1DdtH0RQ+i4VmM4/9pncFsGVIsWgBm+zfSJIPA59uZ0fu92/QeST5NzR9D7+WBJpYv5rkkKr6DvCjdvvrktwOPBMY2sAS0zkPRuzzmM7/wKr6EQvks+ij+798pD4LzY2qOhs4GyDJn9B87pO6NqqqLe3Pe5JcQtM0dN4u6AfF3jHo2nTov9sziH2o73ufuO+muUs86P/SmFF8zycb+9B/1wex2fAUJTmgM/sq4JZhxTJdSVYB7wBeVVUPDDueJepa4IAk+yfZETgKuGzIMWmRSbJnZ/ZXaZoQQfO7dlSSnZLsTzNoxlfmO77JqKqvV9VTq2q/qtqP5mLg4Kr6TpLlSZYBJHkGzXncMcRwBxrvPFhAn8eg/4EL6bOAcf+XL5jPQtPX3kkiyb7AfwAumMy1UZInJtl1bJrm7tW8jrY/IPbJXJsO/bpjurEP+33vE/d549TnXaP4nk8q9mG/5+PxzuvUvSfJgcAjNCN0nTjkeKbjgzQjKV7efutyTVUtuPNI8qvAB4DlwP9LckNVLYjHGlTVw0lOphnpbRlwTlXdPOSwpizJBcCLgT2SbKa5A3P2cKNSx58leQ5NM6U7gd8EqKqbk1wEfIOmCdFJVfWvwwpyBl4E/I8kDwP/CpxYVQtuALEF9nkM+h+40D6LvuexwD4LTd9fJvlJ4CGaz/gfk/S9Nkrz6JaPVNVq4GnAJe36HYDzq+qvRyD2j/T7fe7GPiLXHdOKneG/79vFPajgQnjPBxUcsfd8oIxwi1FJkiRJkgCbDUuSJEmSFgCTV0mSJEnSyDN5lSRJkiSNPJNXSZIkSdLIM3mVJEmSJI08k1dJkiRJ0sgzeZUkSZIkjbz/H7Q545P2TQBXAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1152x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting histogram - Economic data\n",
    "\n",
    "fig, axes = plt.subplots(nrows=2, ncols=3)\n",
    "fig.set_size_inches(16, 10)\n",
    "\n",
    "# pdays histogram\n",
    "hist(axes[0,0], df_eda.select('pdays'), bins = 50, color=['red']);\n",
    "axes[0,0].set_title('Histogram of pdays');\n",
    "\n",
    "# nr_employed histogram\n",
    "hist(axes[0,1], df_eda.select('nr_employed'), bins = 50, color=['red']);\n",
    "axes[0,1].set_title('Histogram of nr_employed');\n",
    "\n",
    "# euribor3m histogram\n",
    "hist(axes[0,2], df_eda.select('euribor3m'), bins = 50, color=['red']);\n",
    "axes[0,2].set_title('Histogram of euribor3m');\n",
    "\n",
    "# emp_var_rate histogram\n",
    "hist(axes[1,0], df_eda.select('emp_var_rate'), bins = 50, color=['red']);\n",
    "axes[1,0].set_title('Histogram of emp_var_rate');\n",
    "\n",
    "# cons_conf_idx histogram\n",
    "hist(axes[1,1], df_eda.select('cons_conf_idx'), bins = 50, color=['red']);\n",
    "axes[1,1].set_title('Histogram of cons_conf_idx');\n",
    "\n",
    "# cons_price_idx histogram\n",
    "hist(axes[1,2], df_eda.select('cons_price_idx'), bins = 50, color=['red']);\n",
    "axes[1,2].set_title('Histogram of cons_price_idx');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Some Exploratory Analisys - Categorical Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:56.363042Z",
     "iopub.status.busy": "2021-12-05T18:24:56.362819Z",
     "iopub.status.idle": "2021-12-05T18:24:57.608308Z",
     "shell.execute_reply": "2021-12-05T18:24:57.607408Z",
     "shell.execute_reply.started": "2021-12-05T18:24:56.363015Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------+-------+-------------+------------------+\n",
      "|CAMPAIGN|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+--------+-------+-------------+------------------+\n",
      "|       1|  17642|         2300|13.037070626913048|\n",
      "|       2|  10570|         1211|11.456953642384105|\n",
      "|       3|   5341|          574|10.747051114023591|\n",
      "|       4|   2651|          249|  9.39268200678989|\n",
      "|       6|    979|           75|7.6608784473953015|\n",
      "|       5|   1599|          120|   7.5046904315197|\n",
      "|      17|     58|            4| 6.896551724137931|\n",
      "|      11|    177|           12| 6.779661016949152|\n",
      "|      23|     16|            1|              6.25|\n",
      "|       7|    629|           38| 6.041335453100159|\n",
      "|       9|    283|           17| 6.007067137809187|\n",
      "|      10|    225|           12| 5.333333333333334|\n",
      "|      13|     92|            4|4.3478260869565215|\n",
      "|       8|    400|           17|              4.25|\n",
      "|      15|     51|            2|3.9215686274509802|\n",
      "|      12|    125|            3|               2.4|\n",
      "|      14|     69|            1|1.4492753623188406|\n",
      "|      31|      7|            0|               0.0|\n",
      "|      34|      3|            0|               0.0|\n",
      "|      28|      8|            0|               0.0|\n",
      "|      27|     11|            0|               0.0|\n",
      "|      26|      8|            0|               0.0|\n",
      "|      20|     30|            0|               0.0|\n",
      "|      22|     17|            0|               0.0|\n",
      "|      19|     26|            0|               0.0|\n",
      "|      43|      2|            0|               0.0|\n",
      "|      16|     51|            0|               0.0|\n",
      "|      35|      5|            0|               0.0|\n",
      "|      39|      1|            0|               0.0|\n",
      "|      40|      2|            0|               0.0|\n",
      "|      24|     15|            0|               0.0|\n",
      "|      29|     10|            0|               0.0|\n",
      "|      41|      1|            0|               0.0|\n",
      "|      33|      4|            0|               0.0|\n",
      "|      42|      2|            0|               0.0|\n",
      "|      37|      1|            0|               0.0|\n",
      "|      18|     33|            0|               0.0|\n",
      "|      25|      8|            0|               0.0|\n",
      "|      21|     24|            0|               0.0|\n",
      "|      32|      4|            0|               0.0|\n",
      "|      56|      1|            0|               0.0|\n",
      "|      30|      7|            0|               0.0|\n",
      "+--------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping Campaign sucess\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT CAMPAIGN,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      campaign,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY campaign, y\n",
    "              )\n",
    "            GROUP BY CAMPAIGN\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:57.609738Z",
     "iopub.status.busy": "2021-12-05T18:24:57.609447Z",
     "iopub.status.idle": "2021-12-05T18:24:58.272670Z",
     "shell.execute_reply": "2021-12-05T18:24:58.271731Z",
     "shell.execute_reply.started": "2021-12-05T18:24:57.609701Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------------+-------+-------------+------------------+\n",
      "|          job|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+-------------+-------+-------------+------------------+\n",
      "|      student|    875|          275|31.428571428571427|\n",
      "|      retired|   1720|          434|25.232558139534884|\n",
      "|   unemployed|   1014|          144|14.201183431952662|\n",
      "|       admin.|  10422|         1352|12.972558050278257|\n",
      "|   management|   2924|          328| 11.21751025991792|\n",
      "|         null|    330|           37|11.212121212121213|\n",
      "|   technician|   6743|          730|10.826041821147857|\n",
      "|self-employed|   1421|          149| 10.48557353976073|\n",
      "|    housemaid|   1060|          106|              10.0|\n",
      "| entrepreneur|   1456|          124| 8.516483516483516|\n",
      "|     services|   3969|          323| 8.138070042831949|\n",
      "|  blue-collar|   9254|          638| 6.894315971471795|\n",
      "+-------------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by job the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT job,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      job,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY job, y\n",
    "              )\n",
    "            GROUP BY job\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:58.274143Z",
     "iopub.status.busy": "2021-12-05T18:24:58.273846Z",
     "iopub.status.idle": "2021-12-05T18:24:58.797275Z",
     "shell.execute_reply": "2021-12-05T18:24:58.796421Z",
     "shell.execute_reply.started": "2021-12-05T18:24:58.274089Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------------------+-------+-------------+------------------+\n",
      "|          education|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+-------------------+-------+-------------+------------------+\n",
      "|         illiterate|     18|            4| 22.22222222222222|\n",
      "|               null|   1731|          251|14.500288850375506|\n",
      "|  university.degree|  12168|         1670|13.724523339907954|\n",
      "|professional.course|   5243|          595|11.348464619492656|\n",
      "|        high.school|   9515|         1031|10.835522858644246|\n",
      "|           basic.4y|   4176|          428| 10.24904214559387|\n",
      "|           basic.6y|   2292|          188| 8.202443280977311|\n",
      "|           basic.9y|   6045|          473|  7.82464846980976|\n",
      "+-------------------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by education the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT education,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      education,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY education, y\n",
    "              )\n",
    "            GROUP BY education\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:58.806086Z",
     "iopub.status.busy": "2021-12-05T18:24:58.803809Z",
     "iopub.status.idle": "2021-12-05T18:24:59.315145Z",
     "shell.execute_reply": "2021-12-05T18:24:59.314227Z",
     "shell.execute_reply.started": "2021-12-05T18:24:58.806036Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+--------+-------+-------------+------------------+\n",
      "| marital|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+--------+-------+-------------+------------------+\n",
      "|    null|     80|           12|              15.0|\n",
      "|  single|  11568|         1620|14.004149377593361|\n",
      "|divorced|   4612|          476|10.320901994796184|\n",
      "| married|  24928|         2532|10.157252888318355|\n",
      "+--------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by marital status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT marital,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      marital,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY marital, y\n",
    "              )\n",
    "            GROUP BY marital\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:59.316560Z",
     "iopub.status.busy": "2021-12-05T18:24:59.316256Z",
     "iopub.status.idle": "2021-12-05T18:24:59.734688Z",
     "shell.execute_reply": "2021-12-05T18:24:59.733783Z",
     "shell.execute_reply.started": "2021-12-05T18:24:59.316495Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---------+-------+-------------+------------------+\n",
      "|  contact|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+---------+-------+-------------+------------------+\n",
      "| cellular|  26144|         3853|14.737607099143208|\n",
      "|telephone|  15044|          787| 5.231321457059293|\n",
      "+---------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by contact status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT contact,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      contact,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY contact, y\n",
    "              )\n",
    "            GROUP BY contact\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:24:59.736175Z",
     "iopub.status.busy": "2021-12-05T18:24:59.735890Z",
     "iopub.status.idle": "2021-12-05T18:25:00.209444Z",
     "shell.execute_reply": "2021-12-05T18:25:00.208839Z",
     "shell.execute_reply.started": "2021-12-05T18:24:59.736136Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-------+-------------+------------------+\n",
      "|default|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+-------+-------+-------------+------------------+\n",
      "|     no|  32588|         4197|12.878973855406898|\n",
      "|   null|   8597|          443| 5.152960335000581|\n",
      "|    yes|      3|            0|               0.0|\n",
      "+-------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by default status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT default,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      default,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY default, y\n",
    "              )\n",
    "            GROUP BY default\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:00.210818Z",
     "iopub.status.busy": "2021-12-05T18:25:00.210552Z",
     "iopub.status.idle": "2021-12-05T18:25:00.616329Z",
     "shell.execute_reply": "2021-12-05T18:25:00.615451Z",
     "shell.execute_reply.started": "2021-12-05T18:25:00.210780Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-------+-------+-------------+------------------+\n",
      "|housing|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+-------+-------+-------------+------------------+\n",
      "|    yes|  21576|         2507|11.619391916944753|\n",
      "|     no|  18622|         2026|10.879604768553323|\n",
      "|   null|    990|          107|10.808080808080808|\n",
      "+-------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by housing status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT housing,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      housing,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY housing, y\n",
    "              )\n",
    "            GROUP BY housing\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:00.617776Z",
     "iopub.status.busy": "2021-12-05T18:25:00.617489Z",
     "iopub.status.idle": "2021-12-05T18:25:01.016599Z",
     "shell.execute_reply": "2021-12-05T18:25:01.015629Z",
     "shell.execute_reply.started": "2021-12-05T18:25:00.617740Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+----+-------+-------------+------------------+\n",
      "|loan|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+----+-------+-------------+------------------+\n",
      "|  no|  33950|         3850| 11.34020618556701|\n",
      "| yes|   6248|          683|10.931498079385403|\n",
      "|null|    990|          107|10.808080808080808|\n",
      "+----+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by loan status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT loan,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      loan,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY loan, y\n",
    "              )\n",
    "            GROUP BY loan\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:01.018104Z",
     "iopub.status.busy": "2021-12-05T18:25:01.017819Z",
     "iopub.status.idle": "2021-12-05T18:25:01.437971Z",
     "shell.execute_reply": "2021-12-05T18:25:01.437072Z",
     "shell.execute_reply.started": "2021-12-05T18:25:01.018062Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+-----------+-------+-------------+------------------+\n",
      "|day_of_week|QTD_CLI|QTD_CONTRATOS|    PERC_CONVERSAO|\n",
      "+-----------+-------+-------------+------------------+\n",
      "|        thu|   8623|         1045|12.118752174417256|\n",
      "|        tue|   8090|          953|11.779975278121137|\n",
      "|        wed|   8134|          949|11.667076469141874|\n",
      "|        fri|   7827|          846| 10.80873898045228|\n",
      "|        mon|   8514|          847| 9.948320413436692|\n",
      "+-----------+-------+-------------+------------------+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Grouping by day_of_week status the campaign conversion\n",
    "\n",
    "spark_new.sql('''\n",
    "            SELECT day_of_week,\n",
    "                SUM(QTD) AS QTD_CLI,\n",
    "                SUM(QTD*FLAG_CONTRATOU) AS QTD_CONTRATOS,\n",
    "                (SUM(QTD*FLAG_CONTRATOU)/SUM(QTD))*100 AS PERC_CONVERSAO\n",
    "            FROM (\n",
    "                  SELECT DISTINCT \n",
    "                      day_of_week,\n",
    "                      y as flag_contratou,\n",
    "                      COUNT(*) as qtd \n",
    "                  FROM df_transient \n",
    "                  GROUP BY day_of_week, y\n",
    "              )\n",
    "            GROUP BY day_of_week\n",
    "            ORDER BY PERC_CONVERSAO DESC\n",
    "              ''').show(100)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data Engeneering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:01.441410Z",
     "iopub.status.busy": "2021-12-05T18:25:01.439175Z",
     "iopub.status.idle": "2021-12-05T18:25:01.452360Z",
     "shell.execute_reply": "2021-12-05T18:25:01.451543Z",
     "shell.execute_reply.started": "2021-12-05T18:25:01.441359Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "root\n",
      " |-- age: integer (nullable = true)\n",
      " |-- job: string (nullable = true)\n",
      " |-- marital: string (nullable = true)\n",
      " |-- education: string (nullable = true)\n",
      " |-- default: string (nullable = true)\n",
      " |-- housing: string (nullable = true)\n",
      " |-- loan: string (nullable = true)\n",
      " |-- contact: string (nullable = true)\n",
      " |-- month: string (nullable = true)\n",
      " |-- day_of_week: string (nullable = true)\n",
      " |-- campaign: integer (nullable = true)\n",
      " |-- pdays: integer (nullable = true)\n",
      " |-- previous: integer (nullable = true)\n",
      " |-- poutcome: string (nullable = true)\n",
      " |-- emp_var_rate: double (nullable = true)\n",
      " |-- cons_price_idx: double (nullable = true)\n",
      " |-- cons_conf_idx: double (nullable = true)\n",
      " |-- euribor3m: double (nullable = true)\n",
      " |-- nr_employed: double (nullable = true)\n",
      " |-- y: integer (nullable = true)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:01.454744Z",
     "iopub.status.busy": "2021-12-05T18:25:01.454146Z",
     "iopub.status.idle": "2021-12-05T18:25:01.761402Z",
     "shell.execute_reply": "2021-12-05T18:25:01.760751Z",
     "shell.execute_reply.started": "2021-12-05T18:25:01.454700Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "|age|      job|marital|       education|housing|loan|contact|day_of_week|pdays|previous|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|  y|\n",
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "| 56|housemaid|      0|Basic incomplete|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 57| employed|      0|     High School|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 37| employed|      0|     High School|      1|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 40| employed|      0|Basic incomplete|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 56| employed|      0|     High School|      0|   1|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Task 1 - Replace statements  in education and job columns\n",
    "\n",
    "# Rules (education):\n",
    "# 'Basic incomplete' : 'illiterate' 'basic.4y', 'basic.6y'\n",
    "# 'Basic Degree': 'basic.9y'\n",
    "# 'High School ': 'high.school'\n",
    "# 'Higher Education Degree': 'university.degree', 'professional.course'\n",
    "\n",
    "# Rules (job):\n",
    "# 'retired' : 'retired' \n",
    "# 'student': 'student'\n",
    "# 'unemployed' : 'unemployed'\n",
    "# 'housemaid' : housemaid'\n",
    "# 'employed': 'admin.' , 'technician', 'services', 'management'\n",
    "# 'enterpreneur': 'self-employed', 'blue-collar' , 'entrepreneur'\n",
    "df_1_new = df.withColumn('education',when(df.education == 'illiterate','Basic incomplete')  \\\n",
    "           .when(df.education == 'basic.4y','Basic incomplete')\\\n",
    "           .when(df.education == 'basic.6y','Basic incomplete') \\\n",
    "           .when(df.education == 'basic.9y','Basic Degree') \\\n",
    "           .when(df.education == 'high.school', 'High School') \\\n",
    "           .when(df.education == 'university.degree', 'Higher Education Degree') \\\n",
    "           .when(df.education == 'professional.course','Higher Education Degree') \\\n",
    "           .otherwise(df.education))\\\n",
    "            .withColumn('job', when(df.job == 'blue-collar','enterpreneur')\\\n",
    "                        .when(df.job == 'self-employed', 'enterpreneur') \\\n",
    "                        .when(df.job == 'management','employed') \\\n",
    "                        .when(df.job == 'admin.','employed') \\\n",
    "                        .when(df.job == 'technician','employed')\\\n",
    "                        .when(df.job == 'services','employed').otherwise(df.job))\n",
    "\n",
    "\n",
    "# Task 2 - Replace statements by Boolean type\n",
    "# contact,  housing, loan, contact, marital\n",
    "df_2_new = df_1_new.withColumn('contact', when(df_1_new.contact == 'telephone', 0).otherwise(1))\\\n",
    "    .withColumn('marital', when(df_1_new.marital == 'single', 1).otherwise(0))\\\n",
    "    .withColumn('loan', when(df_1_new.loan == 'yes', 1).otherwise(0))\\\n",
    "    .withColumn('housing', when(df_1_new.housing == 'no', 0).otherwise(1))\n",
    "\n",
    "\n",
    "# Task 3 - Drop default, month and campaign\n",
    "# contact,  housing, loan, contact, marital\n",
    "columns_to_drop = ['default', 'month', 'campaign', 'poutcome', 'nr_employed']\n",
    "df_trusted = df_2_new.drop(*columns_to_drop)\n",
    "\n",
    "\n",
    "# Task 4 -> Fill na\n",
    "# columns-> all columns\n",
    "df_trusted = df_trusted.fillna( { 'job':'employed',\n",
    "                              'education':'Higher Education Degree',\n",
    "                              'pdays':0} )\n",
    "df_trusted.show(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Counting null values**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:02.555858Z",
     "iopub.status.busy": "2021-12-05T18:25:02.555582Z",
     "iopub.status.idle": "2021-12-05T18:25:03.771967Z",
     "shell.execute_reply": "2021-12-05T18:25:03.771066Z",
     "shell.execute_reply.started": "2021-12-05T18:25:02.555821Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---+-------+---------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "|age|job|marital|education|housing|loan|contact|day_of_week|pdays|previous|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|  y|\n",
      "+---+---+-------+---------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "|  0|  0|      0|        0|      0|   0|      0|          0|    0|       0|           0|             0|            0|        0|  0|\n",
      "+---+---+-------+---------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_trusted.select([count(when(isnan(c) | col(c).isNull(), c)).alias(c) for c in df_trusted.columns]).show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Saving Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:03.773417Z",
     "iopub.status.busy": "2021-12-05T18:25:03.773130Z",
     "iopub.status.idle": "2021-12-05T18:25:04.491307Z",
     "shell.execute_reply": "2021-12-05T18:25:04.490390Z",
     "shell.execute_reply.started": "2021-12-05T18:25:03.773375Z"
    }
   },
   "outputs": [],
   "source": [
    "df_trusted.write.option(\"header\",True).mode(\"overwrite\").option(\"delimiter\",\";\").csv('data_trusted.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:04.493580Z",
     "iopub.status.busy": "2021-12-05T18:25:04.493267Z",
     "iopub.status.idle": "2021-12-05T18:25:04.500458Z",
     "shell.execute_reply": "2021-12-05T18:25:04.499567Z",
     "shell.execute_reply.started": "2021-12-05T18:25:04.493533Z"
    }
   },
   "outputs": [],
   "source": [
    "from pyspark.sql import SparkSession\n",
    "from pyspark.sql.types import StructType\n",
    "\n",
    "# Pre Processing\n",
    "from pyspark.ml import Pipeline\n",
    "from pyspark.ml.feature import OneHotEncoder, StringIndexer, VectorAssembler, StandardScaler\n",
    "\n",
    "# Machine Learning\n",
    "from pyspark.ml.classification import RandomForestClassifier, LogisticRegression, DecisionTreeClassifier, GBTClassifier, MultilayerPerceptronClassifier, LinearSVC, NaiveBayes\n",
    "from pyspark.ml.tuning import CrossValidator, ParamGridBuilder\n",
    "from pyspark.ml.evaluation import BinaryClassificationEvaluator as BCE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [],
   "source": [
    "spark_model = SparkSession \\\n",
    "    .builder \\\n",
    "    .appName(\"Data Modeling with Spark\") \\\n",
    "    .getOrCreate()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 265,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "            <div>\n",
       "                <p><b>SparkSession - in-memory</b></p>\n",
       "                \n",
       "        <div>\n",
       "            <p><b>SparkContext</b></p>\n",
       "\n",
       "            <p><a href=\"http://ip-172-16-120-185.ec2.internal:4041\">Spark UI</a></p>\n",
       "\n",
       "            <dl>\n",
       "              <dt>Version</dt>\n",
       "                <dd><code>v2.4.0</code></dd>\n",
       "              <dt>Master</dt>\n",
       "                <dd><code>local[*]</code></dd>\n",
       "              <dt>AppName</dt>\n",
       "                <dd><code>Data wrangling with Spark SQL</code></dd>\n",
       "            </dl>\n",
       "        </div>\n",
       "        \n",
       "            </div>\n",
       "        "
      ],
      "text/plain": [
       "<pyspark.sql.session.SparkSession at 0x7f90c5199240>"
      ]
     },
     "execution_count": 265,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spark_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 266,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:04.502707Z",
     "iopub.status.busy": "2021-12-05T18:25:04.502163Z",
     "iopub.status.idle": "2021-12-05T18:25:04.668185Z",
     "shell.execute_reply": "2021-12-05T18:25:04.667574Z",
     "shell.execute_reply.started": "2021-12-05T18:25:04.502652Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "|age|      job|marital|       education|housing|loan|contact|day_of_week|pdays|previous|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|  y|\n",
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "| 56|housemaid|      0|Basic incomplete|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 57| employed|      0|     High School|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 37| employed|      0|     High School|      1|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 40| employed|      0|Basic incomplete|      0|   0|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "| 56| employed|      0|     High School|      0|   1|      0|        mon|  999|       0|         1.1|        93.994|        -36.4|    4.857|  0|\n",
      "+---+---------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# local do arquivo\n",
    "path = 'data_trusted.csv'\n",
    "\n",
    "# Lendo o arquivo em um objeto Spark\n",
    "#df = spark.read.option(\"delimiter\", \";\").option(\"header\", \"true\").option(\"nullValue\", \"unknown\").option(\"nullValue\", \"null\").csv(path)\n",
    "\n",
    "# Define csv input schema\n",
    "schema = StructType([\n",
    "    StructField(\"age\", IntegerType()),\n",
    "    StructField(\"job\", StringType()),\n",
    "    StructField(\"marital\", IntegerType()),\n",
    "    StructField(\"education\", StringType()),\n",
    "    StructField(\"housing\", IntegerType()),\n",
    "    StructField(\"loan\", IntegerType()),\n",
    "    StructField(\"contact\", IntegerType()),\n",
    "    StructField(\"day_of_week\", StringType()),\n",
    "    StructField(\"pdays\", IntegerType()),\n",
    "    StructField(\"previous\", IntegerType()),\n",
    "    StructField(\"emp_var_rate\", FloatType()),\n",
    "    StructField(\"cons_price_idx\", FloatType()),\n",
    "    StructField(\"cons_conf_idx\", FloatType()),\n",
    "    StructField(\"euribor3m\", FloatType()),\n",
    "    StructField(\"y\", IntegerType()),\n",
    "])\n",
    "\n",
    "df_model = spark_model.read.option(\"delimiter\", \";\").option(\"header\", \"true\").csv(path, schema=schema)\n",
    "\n",
    "df_model.show(5)\n",
    "#df_model.printSchema()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 267,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:04.670118Z",
     "iopub.status.busy": "2021-12-05T18:25:04.669255Z",
     "iopub.status.idle": "2021-12-05T18:25:05.796435Z",
     "shell.execute_reply": "2021-12-05T18:25:05.795563Z",
     "shell.execute_reply.started": "2021-12-05T18:25:04.670080Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Dataset Count: 33015\n",
      "Test Dataset Count: 8173\n"
     ]
    }
   ],
   "source": [
    "############################################ Split data\n",
    "df_train, df_test = df_model.randomSplit(weights = [0.80, 0.20], seed = 42)\n",
    "\n",
    "print(\"Training Dataset Count: \" + str(df_train.count()))\n",
    "print(\"Test Dataset Count: \" + str(df_test.count()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:05.797895Z",
     "iopub.status.busy": "2021-12-05T18:25:05.797618Z",
     "iopub.status.idle": "2021-12-05T18:25:05.806101Z",
     "shell.execute_reply": "2021-12-05T18:25:05.805547Z",
     "shell.execute_reply.started": "2021-12-05T18:25:05.797857Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Result Categorical:  ['job', 'education', 'day_of_week']\n",
      "Result Numerical:  ['age', 'marital', 'housing', 'loan', 'contact', 'pdays', 'previous', 'emp_var_rate', 'cons_price_idx', 'cons_conf_idx', 'euribor3m', 'y']\n"
     ]
    }
   ],
   "source": [
    "############################################ Prepara data -> get dtypes\n",
    "dt = df_train.dtypes\n",
    "  \n",
    "# get columns by type\n",
    "categoricalList = [item[0] for item in dt if item[1].startswith('string') or item[1].startswith('bigint')]\n",
    "numList = [item[0] for item in dt if item[1].startswith('int') or item[1].startswith('float')]\n",
    "\n",
    "print(\"Result Categorical: \", categoricalList)\n",
    "print(\"Result Numerical: \", numList)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pré Processing Pipeline**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:05.807553Z",
     "iopub.status.busy": "2021-12-05T18:25:05.807333Z",
     "iopub.status.idle": "2021-12-05T18:25:05.871267Z",
     "shell.execute_reply": "2021-12-05T18:25:05.870476Z",
     "shell.execute_reply.started": "2021-12-05T18:25:05.807503Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[StringIndexer_d5fb125dcf84,\n",
       " StringIndexer_72c1ee7d82ed,\n",
       " StringIndexer_982b81c40676,\n",
       " OneHotEncoder_b322ded1d0b1,\n",
       " OneHotEncoder_622c3ae90b34,\n",
       " OneHotEncoder_ec87ce279f95,\n",
       " VectorAssembler_a1deea27eb21,\n",
       " StandardScaler_6649839feb88]"
      ]
     },
     "execution_count": 269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "###############################################################################\n",
    "# Task 1 -> String Indexer\n",
    "# columns-> ['job', 'education','day_of_week']\n",
    "# output columns-> ['job_StringIndexer', 'education_StringIndexer','day_of_week_StringIndexer']\n",
    "stage_string = [StringIndexer(inputCol= c, outputCol= c+\"_StringIndexer\", handleInvalid=\"keep\") for c in categoricalList]\n",
    "\n",
    "\n",
    "###############################################################################\n",
    "# Task 2 -> One Hot Encoder\n",
    "# columns-> ['job_new', 'education','day_of_week']\n",
    "# output columns-> ['job_OneHotEncoder', 'education_OneHotEncoder','day_of_week_OneHotEncoder']\n",
    "stage_one_hot = [OneHotEncoder(inputCol= f\"{c}_StringIndexer\", outputCol= f\"{c}_OneHotEncoder\") for c in categoricalList]\n",
    "\n",
    "\n",
    "###############################################################################\n",
    "# Task 3 -> Vector Assembler\n",
    "# allow us to use ML methods and standard scaler\n",
    "# columns-> all columns\n",
    "colunas = numList + ['job_OneHotEncoder', 'education_OneHotEncoder','day_of_week_OneHotEncoder']\n",
    "stage_vector_assembler = [VectorAssembler(inputCols=colunas,   outputCol='features') ]\n",
    "\n",
    "\n",
    "###############################################################################\n",
    "# Task 4 -> Standard Scaler\n",
    "# columns-> all columns\n",
    "stage_standard_scaler = StandardScaler().setInputCol(\"features\")\n",
    "\n",
    "###############################################################################\n",
    "# Creta an list os sequential steps of your pipeline\n",
    "stages_pre_processing = stage_string+ stage_one_hot+ stage_vector_assembler+ [stage_standard_scaler]\n",
    "stages_pre_processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:05.874996Z",
     "iopub.status.busy": "2021-12-05T18:25:05.874766Z",
     "iopub.status.idle": "2021-12-05T18:25:10.274484Z",
     "shell.execute_reply": "2021-12-05T18:25:10.273587Z",
     "shell.execute_reply.started": "2021-12-05T18:25:05.874970Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+---+-------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+-----------------+-----------------------+-------------------------+-----------------+-----------------------+-------------------------+--------------------+-----------------------------------+\n",
      "|age|    job|marital|       education|housing|loan|contact|day_of_week|pdays|previous|emp_var_rate|cons_price_idx|cons_conf_idx|euribor3m|  y|job_StringIndexer|education_StringIndexer|day_of_week_StringIndexer|job_OneHotEncoder|education_OneHotEncoder|day_of_week_OneHotEncoder|            features|StandardScaler_6649839feb88__output|\n",
      "+---+-------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+-----------------+-----------------------+-------------------------+-----------------+-----------------------+-------------------------+--------------------+-----------------------------------+\n",
      "| 18|student|      1|Basic incomplete|      0|   0|      1|        thu|  999|       0|        -1.8|        93.075|        -47.1|    1.365|  0|              6.0|                    2.0|                      0.0|    (7,[6],[1.0])|          (4,[2],[1.0])|            (5,[0],[1.0])|(28,[0,1,4,5,7,8,...|               (28,[0,1,4,5,7,8,...|\n",
      "| 19|student|      1|    Basic Degree|      0|   0|      1|        thu|  999|       0|        -1.8|        93.075|        -47.1|     1.41|  0|              6.0|                    3.0|                      0.0|    (7,[6],[1.0])|          (4,[3],[1.0])|            (5,[0],[1.0])|(28,[0,1,4,5,7,8,...|               (28,[0,1,4,5,7,8,...|\n",
      "+---+-------+-------+----------------+-------+----+-------+-----------+-----+--------+------------+--------------+-------------+---------+---+-----------------+-----------------------+-------------------------+-----------------+-----------------------+-------------------------+--------------------+-----------------------------------+\n",
      "only showing top 2 rows\n",
      "\n",
      "CPU times: user 113 ms, sys: 0 ns, total: 113 ms\n",
      "Wall time: 1.97 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Make Pipeline \n",
    "\n",
    "# create object\n",
    "ppl = Pipeline().setStages(stages_pre_processing)\n",
    "\n",
    "# Trainning pipeline\n",
    "# allow us save pipeline\n",
    "pipelineModel = ppl.fit(df_train)\n",
    "\n",
    "# Applying changes\n",
    "df_train_clean  = pipelineModel.transform(df_train)\n",
    "df_test_clean  = pipelineModel.transform(df_test)\n",
    "\n",
    "df_train_clean.show(2,truncate=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Saving Pipeline**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:10.275993Z",
     "iopub.status.busy": "2021-12-05T18:25:10.275693Z",
     "iopub.status.idle": "2021-12-05T18:25:12.833567Z",
     "shell.execute_reply": "2021-12-05T18:25:12.832565Z",
     "shell.execute_reply.started": "2021-12-05T18:25:10.275958Z"
    }
   },
   "outputs": [],
   "source": [
    "# Save pipeline\n",
    "\n",
    "pipelineModel.save(\"pipeline_pre_processing_1\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Checking final dataframe**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:12.835047Z",
     "iopub.status.busy": "2021-12-05T18:25:12.834775Z",
     "iopub.status.idle": "2021-12-05T18:25:14.072167Z",
     "shell.execute_reply": "2021-12-05T18:25:14.071208Z",
     "shell.execute_reply.started": "2021-12-05T18:25:12.835010Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "+------------------------------------------------------------------------------------------------------------------------------------------------+\n",
      "|features                                                                                                                                        |\n",
      "+------------------------------------------------------------------------------------------------------------------------------------------------+\n",
      "|(28,[0,1,4,5,7,8,9,10,18,21,23],[18.0,1.0,1.0,999.0,-1.7999999523162842,93.07499694824219,-47.099998474121094,1.3650000095367432,1.0,1.0,1.0])  |\n",
      "|(28,[0,1,4,5,7,8,9,10,18,22,23],[19.0,1.0,1.0,999.0,-1.7999999523162842,93.07499694824219,-47.099998474121094,1.409999966621399,1.0,1.0,1.0])   |\n",
      "|(28,[0,1,4,5,7,8,9,10,18,22,26],[19.0,1.0,1.0,999.0,-1.7999999523162842,93.07499694824219,-47.099998474121094,1.4529999494552612,1.0,1.0,1.0])  |\n",
      "|(28,[0,1,2,4,5,7,8,9,10,11,18,22,27],[19.0,1.0,1.0,1.0,999.0,-1.7999999523162842,92.84300231933594,-50.0,1.6399999856948853,1.0,1.0,1.0,1.0])   |\n",
      "|(28,[0,1,2,4,5,7,8,9,10,18,22,24],[19.0,1.0,1.0,1.0,999.0,1.399999976158142,93.91799926757812,-42.70000076293945,4.960000038146973,1.0,1.0,1.0])|\n",
      "+------------------------------------------------------------------------------------------------------------------------------------------------+\n",
      "only showing top 5 rows\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Up the data to cache to models learn faster\n",
    "df_train_clean.cache()\n",
    "df_test_clean.cache()\n",
    "\n",
    "df_train_clean.select(['features']).show(5, truncate=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Models\n",
    "\n",
    "* Logistic Regression\n",
    "* Decision Tree\n",
    "* Random Forest\n",
    "* Gradient-boosted tree\n",
    "* Linear Support Vector Machine\n",
    "* Naive Bayes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### **Metrics**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:14.073780Z",
     "iopub.status.busy": "2021-12-05T18:25:14.073453Z",
     "iopub.status.idle": "2021-12-05T18:25:14.681594Z",
     "shell.execute_reply": "2021-12-05T18:25:14.680864Z",
     "shell.execute_reply.started": "2021-12-05T18:25:14.073738Z"
    }
   },
   "outputs": [],
   "source": [
    "from pyspark.mllib.evaluation import MulticlassMetrics\n",
    "from pyspark.ml.evaluation import BinaryClassificationEvaluator, MulticlassClassificationEvaluator\n",
    "\n",
    "from sklearn.metrics import accuracy_score, f1_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import numpy as np\n",
    "\n",
    "main_metric = BCE(labelCol = \"y\",\\\n",
    "                    rawPredictionCol = \"prediction\",\\\n",
    "                    metricName = \"areaUnderROC\")\n",
    "\n",
    "def valuation_metrics(df, model):\n",
    "    \n",
    "    # Create both evaluators\n",
    "    evaluatorMulti = MulticlassClassificationEvaluator(labelCol=\"y\", predictionCol=\"prediction\")\n",
    "\n",
    "    # Make predicitons\n",
    "    predictionAndTarget = model.transform(df).select(\"y\", \"prediction\")\n",
    "    \n",
    "    data_y =  np.array(predictionAndTarget.select(\"y\").collect())\n",
    "    data_pred =  np.array(predictionAndTarget.select(\"prediction\").collect())\n",
    "\n",
    "\n",
    "    # Get metrics\n",
    "    acc = evaluatorMulti.evaluate(predictionAndTarget, {evaluatorMulti.metricName: \"accuracy\"})\n",
    "    f1 = evaluatorMulti.evaluate(predictionAndTarget, {evaluatorMulti.metricName: \"f1\"})\n",
    "    weightedPrecision = evaluatorMulti.evaluate(predictionAndTarget, {evaluatorMulti.metricName: \"weightedPrecision\"})\n",
    "    weightedRecall = evaluatorMulti.evaluate(predictionAndTarget, {evaluatorMulti.metricName: \"weightedRecall\"})\n",
    "    auc = main_metric.evaluate(predictionAndTarget)\n",
    "\n",
    "    \n",
    "    print(\"Summary Stats\")\n",
    "    print('Accuracy:', acc)\n",
    "    print('F1-Score:', f1)\n",
    "    print('weightedPrecision:', weightedPrecision)\n",
    "    print('AUC:', auc)\n",
    "    \n",
    "    logreg_matrix = confusion_matrix(data_y, data_pred)\n",
    "\n",
    "    print(\"Confusion Matrix\", logreg_matrix)\n",
    "    print('acuracy', accuracy_score(data_y,data_pred))\n",
    "    print('f1_score',f1_score(data_y,data_pred) )\n",
    "    print('End Metrics \\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Logistic Regression**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-12-05T18:25:14.682907Z",
     "iopub.status.busy": "2021-12-05T18:25:14.682700Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.69 s, sys: 1.84 s, total: 7.52 s\n",
      "Wall time: 1min 48s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Create model object\n",
    "lr = LogisticRegression(labelCol ='y',featuresCol=\"features\")\n",
    "\n",
    "# fit parameters grid\n",
    "paramGrid = ParamGridBuilder() \\\n",
    "    .addGrid(lr.regParam, [ 0.1, 0.5, 1.0]) \\\n",
    "    .addGrid(lr.elasticNetParam, [0.0, 0.5, 1.0]) \\\n",
    "    .addGrid(lr.maxIter, [5, 10, 20]) \\\n",
    "    .build()\n",
    "\n",
    "\n",
    "# cross validation\n",
    "crossval = CrossValidator(estimator = lr,\n",
    "                            estimatorParamMaps=paramGrid,\n",
    "                            evaluator = main_metric,\n",
    "                            numFolds= 5)\n",
    "\n",
    "cv_model_lr = crossval.fit(df_train_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Metrics:\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[29288     0]\n",
      " [    0  3727]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "Test Metrics:\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[7260    0]\n",
      " [   0  913]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Train Metrics:\")\n",
    "valuation_metrics(df_train_clean,cv_model_lr)\n",
    "\n",
    "print(\"Test Metrics:\")\n",
    "valuation_metrics(df_test_clean,cv_model_lr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Decision Tree**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 9.93 s, sys: 3.42 s, total: 13.4 s\n",
      "Wall time: 2min 21s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Create model object\n",
    "dt = DecisionTreeClassifier(labelCol ='y',featuresCol=\"features\")\n",
    "\n",
    "# fit parameters grid\n",
    "\n",
    "paramGrid = ParamGridBuilder() \\\n",
    "    .addGrid(dt.maxDepth, [2, 5, 10, 20, 30]) \\\n",
    "    .addGrid(dt.maxBins, [10, 20, 40, 80, 100]) \\\n",
    "    .addGrid(dt.impurity, ['gini', 'entropy']) \\\n",
    "    .build()\n",
    "\n",
    "\n",
    "# cross validation\n",
    "crossval_dt = CrossValidator(estimator = dt,\n",
    "                            estimatorParamMaps=paramGrid,\n",
    "                            evaluator = main_metric,\n",
    "                            numFolds= 5)\n",
    "\n",
    "cv_model_dt = crossval_dt.fit(df_train_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[29288     0]\n",
      " [    0  3727]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "Test Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[7260    0]\n",
      " [   0  913]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "CPU times: user 1.42 s, sys: 0 ns, total: 1.42 s\n",
      "Wall time: 2.93 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "print(\"Train Metrics:\\n\")\n",
    "valuation_metrics(df_train_clean,cv_model_dt)\n",
    "\n",
    "print(\"Test Metrics:\\n\")\n",
    "valuation_metrics(df_test_clean,cv_model_dt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Random Forest**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.86 s, sys: 2.37 s, total: 11.2 s\n",
      "Wall time: 6min 18s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Create model object\n",
    "random_forest = RandomForestClassifier(labelCol = \"y\", featuresCol = \"features\")\n",
    "\n",
    "# fit parameters grid\n",
    "paramGrid = ParamGridBuilder() \\\n",
    "    .addGrid(dt.maxDepth, [2, 10, 30]) \\\n",
    "    .addGrid(dt.maxBins, [10, 40, 100]) \\\n",
    "    .addGrid(random_forest.numTrees, [20, 30, 50, 100]) \\\n",
    "    .build()\n",
    "\n",
    "\n",
    "# cross validation\n",
    "crossval = CrossValidator(estimator = random_forest,\n",
    "                            estimatorParamMaps=paramGrid,\n",
    "                            evaluator = main_metric,\n",
    "                            numFolds= 5)\n",
    "\n",
    "cv_model_rf = crossval.fit(df_train_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[29288     0]\n",
      " [    0  3727]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "Test Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[7260    0]\n",
      " [   0  913]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "CPU times: user 1.37 s, sys: 0 ns, total: 1.37 s\n",
      "Wall time: 4.85 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "print(\"Train Metrics:\\n\")\n",
    "valuation_metrics(df_train_clean,cv_model_rf)\n",
    "\n",
    "print(\"Test Metrics:\\n\")\n",
    "valuation_metrics(df_test_clean,cv_model_rf)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Gradient-boosted tree**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 20.3 s, sys: 4.88 s, total: 25.2 s\n",
      "Wall time: 30min 4s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Create model object\n",
    "gbt = GBTClassifier(labelCol ='y', featuresCol = \"features\")\n",
    "\n",
    "# fit parameters grid\n",
    "\n",
    "paramGrid = ParamGridBuilder() \\\n",
    "    .addGrid(gbt.maxDepth, [2, 5, 10])\\\n",
    "    .addGrid(gbt.maxBins, [10, 20, 40])\\\n",
    "    .addGrid(gbt.maxIter, [5, 10, 20])\\\n",
    "    .build()\n",
    "\n",
    "\n",
    "# cross validation\n",
    "crossval = CrossValidator(estimator = gbt,\n",
    "                            estimatorParamMaps=paramGrid,\n",
    "                            evaluator = main_metric,\n",
    "                            numFolds= 5)\n",
    "\n",
    "cv_model_gbt = crossval.fit(df_train_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[29288     0]\n",
      " [    0  3727]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "Test Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[7260    0]\n",
      " [   0  913]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "CPU times: user 4.11 s, sys: 0 ns, total: 4.11 s\n",
      "Wall time: 11.4 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Train Metrics:\\n\")\n",
    "valuation_metrics(df_train_clean,cv_model_gbt)\n",
    "\n",
    "print(\"Test Metrics:\\n\")\n",
    "valuation_metrics(df_test_clean,cv_model_gbt)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Linear Support Vector Machine**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.09 s, sys: 1.19 s, total: 6.28 s\n",
      "Wall time: 8min 27s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# Create model object\n",
    "lsvc = LinearSVC(labelCol ='y')\n",
    "\n",
    "# fit parameters grid\n",
    "\n",
    "#### falta configurar aqui\n",
    "\n",
    "paramGrid = ParamGridBuilder() \\\n",
    "    .addGrid(lsvc.maxIter, [5, 10, 50]) \\\n",
    "    .addGrid(lsvc.regParam, [0.001, 0.01, 1.0]) \\\n",
    "    .build()\n",
    "\n",
    "\n",
    "# cross validation\n",
    "crossval = CrossValidator(estimator = lsvc,\n",
    "                            estimatorParamMaps=paramGrid,\n",
    "                            evaluator = main_metric,\n",
    "                            numFolds= 5)\n",
    "\n",
    "cv_model_svm = crossval.fit(df_train_clean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[29288     0]\n",
      " [    0  3727]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "Test Metrics:\n",
      "\n",
      "Summary Stats\n",
      "Accuracy: 1.0\n",
      "F1-Score: 1.0\n",
      "weightedPrecision: 1.0\n",
      "AUC: 1.0\n",
      "Confusion Matrix [[7260    0]\n",
      " [   0  913]]\n",
      "acuracy 1.0\n",
      "f1_score 1.0\n",
      "End Metrics \n",
      "\n",
      "CPU times: user 1.34 s, sys: 0 ns, total: 1.34 s\n",
      "Wall time: 3.61 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "print(\"Train Metrics:\\n\")\n",
    "valuation_metrics(df_train_clean,cv_model_svm)\n",
    "\n",
    "print(\"Test Metrics:\\n\")\n",
    "valuation_metrics(df_test_clean,cv_model_svm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.5, 0.5, 0.5, 0.6, 0.6, 0.5, 1.0, 1.0, 1.0]"
      ]
     },
     "execution_count": 285,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv_model_lr.avgMetrics\n",
    "cv_model_dt.avgMetrics\n",
    "cv_model_rf.avgMetrics\n",
    "cv_model_gbt.avgMetrics\n",
    "cv_model_svm.avgMetrics"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Results\n",
    "\n",
    "\n",
    "| Models                        | Time     | Accuracy | AUC |\n",
    "|-------------------------------|----------|----------|-----|\n",
    "| Logistic Regression           | 1min 54s | 1.0      | 1.0 |\n",
    "| Decision Tree                 | 2min 32s | 1.0      | 1.0 |\n",
    "| Random Forest                 | 5min 36s | 1.0      | 1.0 |\n",
    "| Gradient-boosted tree         | 7min 12s | 1.0      | 1.0 |\n",
    "| Linear Support Vector Machine | 2min 14s | 1.0      | 1.0 |\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'cv_model' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-286-830f4ef9c0d7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# Save best model\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mcv_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msave\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"best_model\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'cv_model' is not defined"
     ]
    }
   ],
   "source": [
    "# Save best model\n",
    "\n",
    "cv_model.save(\"best_model\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "conda_python3",
   "language": "python",
   "name": "conda_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": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}