skill_classifier/01-eda.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "24305037",
   "metadata": {},
   "source": [
    "## 1. Exploratory Data Analysis\n",
    "\n",
    "## Table of contents\n",
    "\n",
    "1. [Get the data and read it to pandas](#getdata)\n",
    "2. [Info Summary](#summarydata)\n",
    "3. [Class Distribution](#classdist)\n",
    "4. [Check Frequent and Uncommon Words](#topwords)\n",
    "    1. [Top 50 Most Frequent Words](#top50)\n",
    "    1. [Uncommon](#uncommon)\n",
    "4. [Distribution of Tokens per Document](#distpertoken)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "25e4d90c",
   "metadata": {},
   "source": [
    "### 1. Get the data and read it to pandas <a name=\"getdata\"></a>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "33b55ed6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tech_soft_none.json\r\n"
     ]
    }
   ],
   "source": [
    "!ls ./data/raw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1e78801c",
   "metadata": {},
   "outputs": [],
   "source": [
    "!wget \"https://drive.google.com/uc?id=1bYFMJ6cz7xm1zDD5Avm5JMz8pp4geRHE&export=download\" -q -O ./data/raw/tech_soft_none.json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9ec82576",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tech_soft_none.json\r\n"
     ]
    }
   ],
   "source": [
    "!ls ./data/raw/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "5bd672e3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\r\n",
      "    \"data\": [\r\n",
      "        {\r\n",
      "            \"text\": \"Standort Trovarit AG München, Deutschland\",\r\n",
      "            \"label\": \"none\"\r\n",
      "        },\r\n",
      "        {\r\n",
      "            \"text\": \"Wir freuen uns auf Ihre Bewerbung unter Angabe Ihrer Gehaltsvorstellung und Ihres möglichen Eintrittstermins. Der ALTE LEIPZIGER – HALLESCHE Konzern begrüßt ausdrücklich die Bewerbung von Menschen mit Behinderung. Bei Fragen und für weitere Informationen wenden Sie sich gerne an Elke Lang unter 0711-6603-2411 . Ihre ausführlichen Bewerbungsunterlagen senden Sie bitte an bewerbung@hallesche.de oder per Post an: HALLESCHE Krankenversicherung a. G. Personal und Soziales Reinsburgstraße 10 70178 Stuttgart www.hallesche.de\",\r\n",
      "            \"label\": \"none\"\r\n",
      "        },\r\n"
     ]
    }
   ],
   "source": [
    "# check file format\n",
    "!head -n 10 ./data/raw/tech_soft_none.json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "72b8122d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd \n",
    "import json\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "daad74e4",
   "metadata": {},
   "source": [
    "*Notes*:\n",
    "\n",
    "Load the json data then flatten it with pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "99a71368",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_raw = \"./data/raw/tech_soft_none.json\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c62ef15c",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(data_raw, 'r') as f: \n",
    "    data = json.loads(f.read())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "fcf1e176",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Transform it to Pandas DataFrame\n",
    "skills_df = pd.json_normalize(data, record_path = [\"data\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ee7290d6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Standort Trovarit AG München, Deutschland</td>\n",
       "      <td>none</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Wir freuen uns auf Ihre Bewerbung unter Angabe...</td>\n",
       "      <td>none</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Qualifikation zur Heimleitung gemäß Heimperson...</td>\n",
       "      <td>tech</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Gute organisatorische und konzeptionelle Fähig...</td>\n",
       "      <td>soft</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Teamfähigkeit, hohe Flexibilität und Einsatzbe...</td>\n",
       "      <td>soft</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                text label\n",
       "0          Standort Trovarit AG München, Deutschland  none\n",
       "1  Wir freuen uns auf Ihre Bewerbung unter Angabe...  none\n",
       "2  Qualifikation zur Heimleitung gemäß Heimperson...  tech\n",
       "3  Gute organisatorische und konzeptionelle Fähig...  soft\n",
       "4  Teamfähigkeit, hohe Flexibilität und Einsatzbe...  soft"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "skills_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f3c9471",
   "metadata": {},
   "source": [
    "**Notes**: \n",
    "\n",
    "Great now that we have the data in a DataFrame, let's do some Data Analysis to understand better our data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ae06b799",
   "metadata": {},
   "source": [
    "## 1.2 Check info summary of data <a name=\"summarydata\"></a>\n",
    "\n",
    "Looking at the table below, we can observe: \n",
    "\n",
    "- We have 10.249 records, all of them are non-null which will make it easier for processing the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "2d87fe6e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10249 entries, 0 to 10248\n",
      "Data columns (total 2 columns):\n",
      " #   Column  Non-Null Count  Dtype \n",
      "---  ------  --------------  ----- \n",
      " 0   text    10249 non-null  object\n",
      " 1   label   10249 non-null  object\n",
      "dtypes: object(2)\n",
      "memory usage: 160.3+ KB\n"
     ]
    }
   ],
   "source": [
    "skills_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f03eb5f",
   "metadata": {},
   "source": [
    "## 1.3 Check Class Distribution <a name=\"classdist\"></a>\n",
    "\n",
    "Let's check how many classes are in the dataset and the distribution per class.\n",
    "\n",
    "**Observations**: \n",
    "- We have 3 classes (none, soft and tech) skills\n",
    "- Distribution of classes seems to be a bit unbalanced with regards to `tech` class where only 22% of the dataset represents the class\n",
    "    - `none`: 42.19 % \n",
    "    - `soft`: 35.46 %\n",
    "    - `tech`: 22.33 %\n",
    "    \n",
    "**Considerations on model creation**: \n",
    "- The minority class `tech` might suffer from the imbalance at prediction time, we might want to check options like downsampling/upsampling or weighing of the classes to help with this issue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d7814aeb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "none    0.421992\n",
       "soft    0.354669\n",
       "tech    0.223339\n",
       "Name: label, dtype: float64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "skills_df['label'].value_counts() / skills_df['label'].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5afbb3f1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "skills_df[\"label\"].value_counts().sort_index().plot.bar(rot=0, grid=True, color=\"blue\", alpha=0.5)\n",
    "plt.xlabel(\"Class\")\n",
    "plt.ylabel(\"Number of Samples\")\n",
    "plt.title(\"Distribution per class\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "beaa0a99",
   "metadata": {},
   "source": [
    "## 1.4 Checking Top-Frequent and Uncommon words  <a name=\"topwords\"></a>\n",
    "\n",
    "By doing this check we can analyse if it's necessary to add additional stop words to our model, as certain words might not contribute to our model. \n",
    "\n",
    "Yet another interesting analysis we can perform is to observe the uncommon words, as they might signal our model towards specific prediction of a particular class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e64af35d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import spacy\n",
    "from spacy.tokens import Doc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f5948d88",
   "metadata": {},
   "outputs": [],
   "source": [
    "de_model = \"de_core_news_lg\"\n",
    "nlp = spacy.load(de_model)\n",
    "de_stop_words = nlp.Defaults.stop_words"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d958f2bd",
   "metadata": {},
   "source": [
    "####  Let's incorporate to the pipe a lenght of the Doc that is being processed "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "693a10df",
   "metadata": {},
   "outputs": [],
   "source": [
    "def len_doc(doc): \n",
    "    doc_length = len(doc)\n",
    "    return doc_length\n",
    "\n",
    "def len_doc_only_alpha(doc): \n",
    "    doc_length = sum([1 if token.is_alpha else 0 for token in doc])\n",
    "    return doc_length\n",
    "\n",
    "# Set the extension on the doc so we can plot it later\n",
    "Doc.set_extension(\"doc_len\", getter=len_doc, force=True)\n",
    "Doc.set_extension(\"doc_len_alpha\", getter=len_doc_only_alpha, force=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45edada4",
   "metadata": {},
   "source": [
    "Let's convert the text to Doc using spaCy, so we can obtain frequencies more easily"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "b5c1d22d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tokens:['Standort', 'Trovarit', 'AG', 'München', ',', 'Deutschland'], \n",
      "Doc length:6, \n",
      "Doc length (only alpha):5\n",
      "------------------------------------\n",
      "Tokens:['Wir', 'freuen', 'uns', 'auf', 'Ihre', 'Bewerbung', 'unter', 'Angabe', 'Ihrer', 'Gehaltsvorstellung', 'und', 'Ihres', 'möglichen', 'Eintrittstermins', '.', 'Der', 'ALTE', 'LEIPZIGER', '–', 'HALLESCHE', 'Konzern', 'begrüßt', 'ausdrücklich', 'die', 'Bewerbung', 'von', 'Menschen', 'mit', 'Behinderung', '.', 'Bei', 'Fragen', 'und', 'für', 'weitere', 'Informationen', 'wenden', 'Sie', 'sich', 'gerne', 'an', 'Elke', 'Lang', 'unter', '0711', '-', '6603', '-', '2411', '.', 'Ihre', 'ausführlichen', 'Bewerbungsunterlagen', 'senden', 'Sie', 'bitte', 'an', 'bewerbung@hallesche.de', 'oder', 'per', 'Post', 'an', ':', 'HALLESCHE', 'Krankenversicherung', 'a.', 'G.', 'Personal', 'und', 'Soziales', 'Reinsburgstraße', '10', '70178', 'Stuttgart', 'www.hallesche.de'], \n",
      "Doc length:75, \n",
      "Doc length (only alpha):59\n",
      "------------------------------------\n",
      "Tokens:['Qualifikation', 'zur', 'Heimleitung', 'gemäß', 'Heimpersonalverordnung'], \n",
      "Doc length:5, \n",
      "Doc length (only alpha):5\n",
      "------------------------------------\n",
      "Tokens:['Gute', 'organisatorische', 'und', 'konzeptionelle', 'Fähigkeiten'], \n",
      "Doc length:5, \n",
      "Doc length (only alpha):5\n",
      "------------------------------------\n",
      "Tokens:['Teamfähigkeit', ',', 'hohe', 'Flexibilität', 'und', 'Einsatzbereitschaft', '.'], \n",
      "Doc length:7, \n",
      "Doc length (only alpha):5\n",
      "------------------------------------\n"
     ]
    }
   ],
   "source": [
    "# Let's transform it to list \n",
    "raw_text = skills_df['text'].to_list()\n",
    "\n",
    "# Let's check the tokenizer of the first 5 sentences\n",
    "for doc in nlp.pipe(raw_text[:5]):\n",
    "    print(f\"Tokens:{[token.text for token in doc]}, \\nDoc length:{doc._.doc_len}, \\nDoc length (only alpha):{doc._.doc_len_alpha}\")\n",
    "    print(\"------------------------------------\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "31525e56",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 16.6 s, sys: 375 ms, total: 17 s\n",
      "Wall time: 17.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "docs = []\n",
    "tokens = []\n",
    "for doc in nlp.pipe(raw_text, batch_size=100):\n",
    "    docs.append(doc)\n",
    "    tokens.append([token.text for token in doc if token.is_alpha])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "fa3c475f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>doc</th>\n",
       "      <th>tokens</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Standort Trovarit AG München, Deutschland</td>\n",
       "      <td>none</td>\n",
       "      <td>(Standort, Trovarit, AG, München, ,, Deutschland)</td>\n",
       "      <td>[Standort, Trovarit, AG, München, Deutschland]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Wir freuen uns auf Ihre Bewerbung unter Angabe...</td>\n",
       "      <td>none</td>\n",
       "      <td>(Wir, freuen, uns, auf, Ihre, Bewerbung, unter...</td>\n",
       "      <td>[Wir, freuen, uns, auf, Ihre, Bewerbung, unter...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Qualifikation zur Heimleitung gemäß Heimperson...</td>\n",
       "      <td>tech</td>\n",
       "      <td>(Qualifikation, zur, Heimleitung, gemäß, Heimp...</td>\n",
       "      <td>[Qualifikation, zur, Heimleitung, gemäß, Heimp...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                text label  \\\n",
       "0          Standort Trovarit AG München, Deutschland  none   \n",
       "1  Wir freuen uns auf Ihre Bewerbung unter Angabe...  none   \n",
       "2  Qualifikation zur Heimleitung gemäß Heimperson...  tech   \n",
       "\n",
       "                                                 doc  \\\n",
       "0  (Standort, Trovarit, AG, München, ,, Deutschland)   \n",
       "1  (Wir, freuen, uns, auf, Ihre, Bewerbung, unter...   \n",
       "2  (Qualifikation, zur, Heimleitung, gemäß, Heimp...   \n",
       "\n",
       "                                              tokens  \n",
       "0     [Standort, Trovarit, AG, München, Deutschland]  \n",
       "1  [Wir, freuen, uns, auf, Ihre, Bewerbung, unter...  \n",
       "2  [Qualifikation, zur, Heimleitung, gemäß, Heimp...  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## Assign it to the DF \n",
    "skills_df['doc'] = pd.Series(docs)\n",
    "skills_df['tokens'] = pd.Series(tokens)\n",
    "skills_df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd131b3c",
   "metadata": {},
   "source": [
    "**Let's check the first sentence and how it was processed by the model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "9492281c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Wir freuen uns auf Ihre Bewerbung unter Angabe Ihrer Gehaltsvorstellung und Ihres möglichen Eintrittstermins. Der ALTE LEIPZIGER – HALLESCHE Konzern begrüßt ausdrücklich die Bewerbung von Menschen mit Behinderung. Bei Fragen und für weitere Informationen wenden Sie sich gerne an Elke Lang unter 0711-6603-2411 . Ihre ausführlichen Bewerbungsunterlagen senden Sie bitte an bewerbung@hallesche.de oder per Post an: HALLESCHE Krankenversicherung a. G. Personal und Soziales Reinsburgstraße 10 70178 Stuttgart www.hallesche.de'"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_text[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "4311146f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "59 ['Wir', 'freuen', 'uns', 'auf', 'Ihre', 'Bewerbung', 'unter', 'Angabe', 'Ihrer', 'Gehaltsvorstellung', 'und', 'Ihres', 'möglichen', 'Eintrittstermins', 'Der', 'ALTE', 'LEIPZIGER', 'HALLESCHE', 'Konzern', 'begrüßt', 'ausdrücklich', 'die', 'Bewerbung', 'von', 'Menschen', 'mit', 'Behinderung', 'Bei', 'Fragen', 'und', 'für', 'weitere', 'Informationen', 'wenden', 'Sie', 'sich', 'gerne', 'an', 'Elke', 'Lang', 'unter', 'Ihre', 'ausführlichen', 'Bewerbungsunterlagen', 'senden', 'Sie', 'bitte', 'an', 'oder', 'per', 'Post', 'an', 'HALLESCHE', 'Krankenversicherung', 'Personal', 'und', 'Soziales', 'Reinsburgstraße', 'Stuttgart']\n"
     ]
    }
   ],
   "source": [
    "print(len(tokens[1]), tokens[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0a5de42c",
   "metadata": {},
   "source": [
    "## 1.4.1 **Top-50 most Frequent Words**  <a name=\"top50\"></a>\n",
    "\n",
    "*Observation notes*: \n",
    "\n",
    "- We can see the most common words are usually words that are stopwords, like `und`, `mit`. \n",
    "- 62% of the Top 50 frequent words are stopwords. We should check this at the time we are performing modeling, sometimes this is not relevant for models, but for Classification it might, for example: \n",
    "    - If we are predicting something like having certain skills or not and someone puts that it doesn't perform well under pressure, and we remove the negation, the classifier might predict otherwise. \n",
    "    - On this particular case it might be Ok, but we will see in our set of experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "6a19bba0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1500x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "word_by_frequency = pd.Series(\" \".join(skills_df[\"tokens\"].apply(lambda t: \" \".join(t))).split()).value_counts()\n",
    "top_50_words = word_by_frequency[:50]\n",
    "plt.figure(figsize=(15,8))\n",
    "top_50_words.plot.bar(rot=0, grid=True, color=\"blue\", alpha=0.5)\n",
    "\n",
    "plt.xticks(rotation = 50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "62710f9c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>word</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>und</td>\n",
       "      <td>5732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>in</td>\n",
       "      <td>1778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sie</td>\n",
       "      <td>1717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>der</td>\n",
       "      <td>1347</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>mit</td>\n",
       "      <td>1248</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  word     0\n",
       "0  und  5732\n",
       "1   in  1778\n",
       "2  Sie  1717\n",
       "3  der  1347\n",
       "4  mit  1248"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_50_df = top_50_words.to_frame().reset_index()\n",
    "top_50_df.rename(columns = {'index':'word'}, inplace=True)\n",
    "top_50_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "6014eb44",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True     27\n",
       "False    23\n",
       "Name: word, dtype: int64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_50_df['word'].isin(de_stop_words).value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33ce4a16",
   "metadata": {},
   "source": [
    "**Check the words that are not stop words in the top 50** \n",
    "\n",
    "- `kenntnisse` --> Knowledge\n",
    "- `arbeitsweise` --> Way of working \n",
    "- `teamfähigkeit` --> Team Work \n",
    "- `englischkenntnisse` --> Knows english\n",
    "\n",
    "\n",
    "We can see this might be signals that could be used by the classifier to effectively perform its prediction towards certain skill"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "f72ed507",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>word</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Sie</td>\n",
       "      <td>1717</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Kenntnisse</td>\n",
       "      <td>1105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>GmbH</td>\n",
       "      <td>514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>Ihre</td>\n",
       "      <td>500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>Deutschland</td>\n",
       "      <td>486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>Arbeitsweise</td>\n",
       "      <td>474</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>Umgang</td>\n",
       "      <td>436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>Standort</td>\n",
       "      <td>426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>Gute</td>\n",
       "      <td>376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>Sehr</td>\n",
       "      <td>372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>m</td>\n",
       "      <td>343</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>w</td>\n",
       "      <td>336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>Teamfähigkeit</td>\n",
       "      <td>320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>Englischkenntnisse</td>\n",
       "      <td>296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>Wir</td>\n",
       "      <td>289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>Bereich</td>\n",
       "      <td>258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>Schrift</td>\n",
       "      <td>253</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>Wort</td>\n",
       "      <td>252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>MS</td>\n",
       "      <td>250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>Kontakt</td>\n",
       "      <td>245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>Flexibilität</td>\n",
       "      <td>244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>Ihr</td>\n",
       "      <td>236</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>Auftreten</td>\n",
       "      <td>228</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  word     0\n",
       "2                  Sie  1717\n",
       "5           Kenntnisse  1105\n",
       "16                GmbH   514\n",
       "17                Ihre   500\n",
       "18         Deutschland   486\n",
       "19        Arbeitsweise   474\n",
       "20              Umgang   436\n",
       "21            Standort   426\n",
       "28                Gute   376\n",
       "29                Sehr   372\n",
       "30                   m   343\n",
       "31                   w   336\n",
       "33       Teamfähigkeit   320\n",
       "35  Englischkenntnisse   296\n",
       "36                 Wir   289\n",
       "37             Bereich   258\n",
       "39             Schrift   253\n",
       "40                Wort   252\n",
       "42                  MS   250\n",
       "44             Kontakt   245\n",
       "45        Flexibilität   244\n",
       "47                 Ihr   236\n",
       "49           Auftreten   228"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_50_df[top_50_df['word'].isin(de_stop_words) == False]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "189e0432",
   "metadata": {},
   "source": [
    "### 1.4.2 Uncommon words <a name=\"uncommon\"></a>\n",
    "\n",
    "Let's check the uncommon words as it also might give us some guidance of words that might not be useful for the model.\n",
    "\n",
    "**Observation Notes**: \n",
    "\n",
    "- It seems a big part of the uncommon words refers to specifics of a skill such as:\n",
    "    - `Cloudservices`\n",
    "    - `FrameMaker`\n",
    "    - `NetApp`\n",
    "    - `Storagelösungen` --> Storage solutions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "bc3dca75",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Healthcheck                1\n",
       "Bypassing                  1\n",
       "Gültige                    1\n",
       "ESX                        1\n",
       "umsetzungsstarke           1\n",
       "Bürogebäude                1\n",
       "Oh                         1\n",
       "my                         1\n",
       "Maximalgröße               1\n",
       "überschreiten              1\n",
       "Cloudservices              1\n",
       "annehmen                   1\n",
       "FrameMaker                 1\n",
       "NetApp                     1\n",
       "Isilon                     1\n",
       "EMC                        1\n",
       "Storagelösungen            1\n",
       "Interfaces                 1\n",
       "Anwendungsmöglichkeiten    1\n",
       "Tatsachen                  1\n",
       "scheinbare                 1\n",
       "Damenschuhe                1\n",
       "Spitzenstellung            1\n",
       "Schuhherstellern           1\n",
       "bedeutendsten              1\n",
       "Gebäudebestand             1\n",
       "umfangreichen              1\n",
       "übergemeindlichen          1\n",
       "Kirchengemeinden           1\n",
       "WPR                        1\n",
       "dtype: int64"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uncommon_words = word_by_frequency[-30:]\n",
    "uncommon_words"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9ff08ac9",
   "metadata": {},
   "source": [
    "## 1.5 Distribution of Tokens per Document and per class <a name=\"distpertoken\"></a>\n",
    "\n",
    "**Notes**:\n",
    "- We can observe we have a long tailed distribution of tokens per document, where:\n",
    "    - Average doc len is around 8 tokens per Document \n",
    "    - Max tokens per doc is 213 \n",
    "    - Min tokens per doc is 0: (**We should take a look into this as we might need to remove it before training**)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "0d629a82",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "3f1c835c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA/YAAAK7CAYAAABYqRRSAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8o6BhiAAAACXBIWXMAAA9hAAAPYQGoP6dpAAB93klEQVR4nOzdeXxU5d3+8etkMpksJAMhZJNFVGQLuGFZ3JCdiqho0aKptIi2KorKY6t9VHzaHyitS4UWraWAgmK1otYlCIIoAopYFGikqCigCWEJ2fc5vz/CHBiSQJaZnJnM581rXmTOuefMd5KMeM29GaZpmgIAAAAAACEpwu4CAAAAAABA8xHsAQAAAAAIYQR7AAAAAABCGMEeAAAAAIAQRrAHAAAAACCEEewBAAAAAAhhBHsAAAAAAEIYwR4AAAAAgBBGsAcAAAAAIIQR7AEArW7RokUyDMO6RUdHKzU1VZdeeqlmz56tvLy8Oo+ZOXOmDMNo0vOUlpZq5syZev/995v0uPqe69RTT9W4ceOadJ2TeeGFF/Tkk0/We84wDM2cOdOvz+dv7733ngYMGKC4uDgZhqHXXnutTpuhQ4f6/KwbujXmtQ4dOlQZGRn+fyFt2LHf/4iICMXHx+uMM87QT37yE73yyivyeDx2lwgA8INIuwsAAISvhQsXqlevXqqqqlJeXp7WrVunRx99VH/84x/10ksvacSIEVbbm266SWPGjGnS9UtLS/Xwww9Lqg04jdWc52qOF154Qdu2bdP06dPrnNuwYYM6d+4c8BqayzRNTZw4UWeeeabeeOMNxcXFqWfPnnXa/eUvf1FhYaF1/6233tLvf/9762fvFcyvNdSddtppWrp0qSSppKREu3bt0muvvaaf/OQnuuiii/Svf/1Lbrfb5ioBAC1BsAcA2CYjI0MDBgyw7l999dW66667dOGFF2rChAnauXOnUlJSJNUGv0CHv9LSUsXGxrbKc53MoEGDbH3+k/nhhx906NAhXXXVVRo+fHiD7fr06eNz/8svv5RU92eP5jFNU+Xl5YqJiWmwTUxMTJ3fp5tuukkLFy7UL37xC91888166aWXAl0qACCAGIoPAAgqXbt21WOPPaaioiI988wz1vH6hsevXr1aQ4cOVceOHRUTE6OuXbvq6quvVmlpqb799lt16tRJkvTwww9bw5EnT57sc73PPvtM11xzjTp06KDTTz+9wefyWr58ufr376/o6Giddtppeuqpp3zOe6cZfPvttz7H33//fRmGYU0LGDp0qN566y199913PkPSveobnr5t2zZdccUV6tChg6Kjo3X22Wdr8eLF9T7Piy++qN/+9rdKT09XQkKCRowYoR07djT8jT/GunXrNHz4cMXHxys2NlZDhgzRW2+9ZZ2fOXOm9cHHr3/9axmGoVNPPbVR166Px+PRnDlz1KtXL7lcLiUnJ+tnP/uZ9u7de9LHLl++XLGxsbrppptUXV0tSfr00081fvx4JSYmKjo6Wuecc47+8Y9/+DzO+3Nas2aNfvWrXykpKUkdO3bUhAkT9MMPP/i0PdHv2Yl4p2+c7HdGkgoLCzVjxgx1795dUVFROuWUUzR9+nSVlJT4tDMMQ7fffruefvpp9e7dWy6Xq87vQGP9/Oc/149//GO9/PLL+u6776zj5eXluu+++3xque2223T48OE613jhhRc0ePBgtWvXTu3atdPZZ5+tBQsWNKseAEDzEewBAEHnxz/+sRwOhz744IMG23z77be67LLLFBUVpb///e/KysrSI488ori4OFVWViotLU1ZWVmSpClTpmjDhg3asGGDHnjgAZ/rTJgwQWeccYZefvllPf300yesa8uWLZo+fbruuusuLV++XEOGDNGdd96pP/7xj01+jX/5y190wQUXKDU11aptw4YNDbbfsWOHhgwZou3bt+upp57Sq6++qj59+mjy5MmaM2dOnfb333+/vvvuO/3tb3/TX//6V+3cuVOXX365ampqTljX2rVrNWzYMBUUFGjBggV68cUXFR8fr8svv9zq1b3pppv06quvSpKmTZumDRs2aPny5U3+Hnj96le/0q9//WuNHDlSb7zxhn73u98pKytLQ4YM0YEDBxp83BNPPKGf/OQnuv/++/W3v/1NkZGRWrNmjS644AIdPnxYTz/9tF5//XWdffbZuvbaa7Vo0aI617jpppvkdDr1wgsvaM6cOXr//fd1ww03WOdP9nt2Mo35nSktLdUll1yixYsX64477tA777yjX//611q0aJHGjx8v0zR9rvnaa69p/vz5evDBB7VixQpddNFFjfgu1897/Q8//FBS7QiAK6+8Un/84x+VmZmpt956S3fffbcWL16sYcOGqaKiwnrsgw8+qOuvv17p6elatGiRli9frhtvvNHnQwIAQCsxAQBoZQsXLjQlmZs2bWqwTUpKitm7d2/r/kMPPWQe+8/WK6+8Ykoyt2zZ0uA19u/fb0oyH3rooTrnvNd78MEHGzx3rG7dupmGYdR5vpEjR5oJCQlmSUmJz2vbtWuXT7s1a9aYksw1a9ZYxy677DKzW7du9dZ+fN3XXXed6XK5zN27d/u0Gzt2rBkbG2sePnzY53l+/OMf+7T7xz/+YUoyN2zYUO/zeQ0aNMhMTk42i4qKrGPV1dVmRkaG2blzZ9Pj8ZimaZq7du0yJZl/+MMfTni94x3/s8/OzjYlmbfeeqtPu48//tiUZN5///3WsUsuucTs27evWVNTY95+++1mVFSUuWTJEp/H9erVyzznnHPMqqoqn+Pjxo0z09LSzJqaGp86jn/eOXPmmJLMnJwc0zQb93vWkMb+zsyePduMiIio837wPvfbb79tHZNkut1u89ChQ42qwfs9a8g777xjSjIfffRR0zRNMysry5Rkzpkzx6fdSy+9ZEoy//rXv5qmaZrffPON6XA4zOuvv75RdQAAAoseewBAUDKP66U83tlnn62oqCjdfPPNWrx4sb755ptmPc/VV1/d6LZ9+/bVWWed5XNs0qRJKiws1Geffdas52+s1atXa/jw4erSpYvP8cmTJ6u0tLROb//48eN97vfv31+STtibWlJSoo8//ljXXHON2rVrZx13OBzKzMzU3r17Gz2cv7HWrFkjSdYUCa8f/ehH6t27t9577z2f4+Xl5bryyiu1dOlSvfvuu7r++uutc1999ZW+/PJL61h1dbV1+/GPf6ycnJw69Z/s+9TS37PG/M68+eabysjI0Nlnn+1T8+jRo32mb3gNGzZMHTp0aFIdDTn+fbZ69WpJdX8eP/nJTxQXF2f9PFauXKmamhrddtttfqkDANAyBHsAQNApKSnRwYMHlZ6e3mCb008/XatWrVJycrJuu+02nX766Tr99NP1pz/9qUnPlZaW1ui2qampDR47ePBgk563qQ4ePFhvrd7v0fHP37FjR5/7LpdLklRWVtbgc+Tn58s0zSY9T0t5r9fQcx7/fHl5eVqxYoUGDx6sIUOG+Jzbt2+fJGnGjBlyOp0+t1tvvVWS6gztP9n3qaW/Z435ndm3b5+++OKLOjXHx8fLNM06NTfld/ZkvB9gHPvzjYyMtNan8DIMQ6mpqVbN+/fvl8RuBgAQLFgVHwAQdN566y3V1NScdIu6iy66SBdddJFqamr06aefau7cuZo+fbpSUlJ03XXXNeq5Glokrz65ubkNHvMGxOjoaEnymYss1Q2UTdWxY0fl5OTUOe5d6C0pKalF15ekDh06KCIiIuDPcyzv9y0nJ6dOSPzhhx/qPF/Xrl31+OOP66qrrtKECRP08ssvW99zb9v77rtPEyZMqPf56tuS72Ra8nvWmN+ZpKQkxcTE6O9//3u91zj+e9CU39mTeeONN2QYhi6++GKrpurqau3fv98n3JumqdzcXJ1//vmSZJ3bu3dvnVEkAIDWR489ACCo7N69WzNmzJDb7dYtt9zSqMc4HA4NHDhQf/7znyXJGuLcmF7qpti+fbs+//xzn2MvvPCC4uPjde6550qStTr8F1984dPujTfeqHM9l8vV6NqGDx+u1atX11mx/bnnnlNsbKxftseLi4vTwIED9eqrr/rU5fF4tGTJEnXu3Flnnnlmi5/nWMOGDZMkLVmyxOf4pk2blJ2dXe9WeqNGjdKKFSv0wQcfaNy4cdbK8T179lSPHj30+eefa8CAAfXe4uPjm11rQ79nJ9KY35lx48bp66+/VseOHeutuSU7DpzIwoUL9c477+inP/2punbtKknW9/v4n8c///lPlZSUWOdHjRolh8Oh+fPnB6Q2AEDT0GMPALDNtm3brPnEeXl5+vDDD7Vw4UI5HA4tX768znDgYz399NNavXq1LrvsMnXt2lXl5eVWj+eIESMkSfHx8erWrZtef/11DR8+XImJiUpKSmp2UEpPT9f48eM1c+ZMpaWlacmSJVq5cqUeffRRxcbGSpLOP/989ezZUzNmzFB1dbU6dOig5cuXa926dXWu169fP7366quaP3++zjvvPEVERDS4t/tDDz2kN998U5deeqkefPBBJSYmaunSpXrrrbc0Z84cud3uZr2m482ePVsjR47UpZdeqhkzZigqKkp/+ctftG3bNr344ot+7S2WasP4zTffrLlz5yoiIkJjx47Vt99+qwceeEBdunTRXXfdVe/jLrzwQr333nsaM2aMRo0apbfffltut1vPPPOMxo4dq9GjR2vy5Mk65ZRTdOjQIWVnZ+uzzz7Tyy+/3KT6GvN7diKN+Z2ZPn26/vnPf+riiy/WXXfdpf79+8vj8Wj37t169913dc8992jgwIFNqvtYZWVl2rhxo/X1N998o9dee01vvvmmLrnkEp/dIEaOHKnRo0fr17/+tQoLC3XBBRfoiy++0EMPPaRzzjlHmZmZkmo/wLr//vv1u9/9TmVlZfrpT38qt9ut//znPzpw4IAefvjhZtcLAGgGW5fuAwCEJe+K5N5bVFSUmZycbF5yySXmrFmzzLy8vDqPOX6l+g0bNphXXXWV2a1bN9PlcpkdO3Y0L7nkEvONN97wedyqVavMc845x3S5XKYk88Ybb/S53v79+0/6XKZZu8L5ZZddZr7yyitm3759zaioKPPUU081H3/88TqP/+9//2uOGjXKTEhIMDt16mROmzbNfOutt+qsin/o0CHzmmuuMdu3b28ahuHznKpnNf+tW7eal19+uel2u82oqCjzrLPOMhcuXOjTxrsq/ssvv+xz3LuK/fHt6/Phhx+aw4YNM+Pi4syYmBhz0KBB5r/+9a96r9fSVfFN0zRramrMRx991DzzzDNNp9NpJiUlmTfccIO5Z88en8fWt8L7tm3bzNTUVPPcc8+1fpaff/65OXHiRDM5Odl0Op1mamqqOWzYMPPpp58+YR2mWXf3gsb+ntWnKb8zxcXF5v/+7/+aPXv2NKOioky3223269fPvOuuu8zc3FyrnSTztttuO+lzH/s9O/a9FhcXZ5522mnmNddcY7788svWLgHHKisrM3/961+b3bp1M51Op5mWlmb+6le/MvPz8+u0fe6558zzzz/fjI6ONtu1a2eec845jfodAwD4l2GaJ1l2GAAAAE126qmnKiMjQ2+++abdpQAA2jjm2AMAAAAAEMII9gAAAAAAhDCG4gMAAAAAEMLosQcAAAAAIIQR7AEAAAAACGEEewAAAAAAQlik3QWECo/Hox9++EHx8fEyDMPucgAAAAAAbZxpmioqKlJ6eroiIhrulyfYN9IPP/ygLl262F0GAAAAACDM7NmzR507d27wPMG+keLj4yXVfkMTEhJsrgYAAAAA0NYVFhaqS5cuVh5tCMG+kbzD7xMSEgj2AAAAAIBWc7Lp4CyeBwAAAABACCPYAwAAAAAQwgj2AAAAAACEMII9AAAAAAAhjGAPAAAAAEAII9gDAAAAABDCCPYAAAAAAIQwgj0AAAAAACGMYA8AAAAAQAgj2AMAAAAAEMII9gAAAAAAhDCCPQAAAAAAIYxgDwAAAABACCPYAwAAAAAQwgj2AAAAAACEMII9AAAAAAAhjGAPAAAAAEAII9gDAAAAABDCCPYAAAAAAIQwgj0AAAAAACGMYA8AAAAAQAgj2AMAAAAAEMII9gAAAAAAhDCCPQAAAAAAIYxgDwAAAABACCPY44TytuXpxfEv6sluTyr/m3y7ywEAAAAAHCfS7gIQnAr3FmrNg2v0+eLPZXpMSdKuNbvU4bQONlcGAAAAADgWwR51fDTnI73/0PuqLq+WJMUkxqjsUJmKvi+yuTIAAAAAwPEYig8fNVU1eu++91RdXq1uF3fTlA1T9KNpP5IkFX5faHN1AAAAAIDj0WMPH6X7S2V6TBkOQzeuuVFGhKF9W/dJEj32AAAAABCE6LGHj+J9xZKkuE5xMiIMSVLCKQmSCPYAAAAAEIwI9vBRklciSYpLjrOOxZ8SL4mh+AAAAAAQjAj28GEF+5Sjwd7bY1+6v1TVFdW21AUAAAAAqB/BHj5K9tXtsY/pGCOHyyFJKs4ptqUuAAAAAED9CPbwUd9QfMMwFJ/OcHwAAAAACEYEe/iweuyPGYovsYAeAAAAAAQrgj181NdjL7GAHgAAAAAEK4I9fHiDfbuUdj7HvcGeHnsAAAAACC4Ee/iw9rFPZig+AAAAAIQCgj0spmk2OBQ/oXNtsGcoPgAAAAAEF4I9LOWHy+Wp8khqeI49PfYAAAAAEFwI9rB4e+tdCS5FRkf6nPMOxS/8vlCmabZ6bQAAAACA+hHsYWloGL4kax/7mooalR0qa9W6AAAAAAANI9jD0tAe9pLkiHIotlOsJIbjAwAAAEAwIdjDcqIee8l3OD4AAAAAIDgQ7GGxgn09PfYSC+gBAAAAQDAi2MPS0B72Xt5gT489AAAAAAQPgj0spXmlkk4+FJ8eewAAAAAIHgR7WLw99u1S2tV7nqH4AAAAABB8CPawNHrxvL0MxQcAAACAYEGwh6Wxi+cxxx4AAAAAggfBHpKk6vJqVRRUSDp5j33ZwTJVl1e3Wm0AAAAAgIYR7CFJKtlf21sf4YxQdPvoettEd4hWZHSkJKnoB+bZAwAAAEAwINhDklSy7+j8esMw6m1jGAbD8QEAAAAgyBDsIenkC+d5seUdAAAAAAQXgj0kHQ32DW1150WPPQAAAAAEF4I9JB3dw/5kPfbsZQ8AAAAAwYVgD0lHe+xjk2NP2I6h+AAAAAAQXAj2kHR08TyG4gMAAABAaCHYQxKL5wEAAABAqCLYQ9IxwT6lkXPsfyiSaZoBrwsAAAAAcGIEe0jy3cf+ROLTaoN9TWWNSg+UBrwuAAAAAMCJEewh02OqZH/jgr0jymG1YTg+AAAAANjP1mA/f/589e/fXwkJCUpISNDgwYP1zjvvWOcnT54swzB8boMGDfK5RkVFhaZNm6akpCTFxcVp/Pjx2rt3r0+b/Px8ZWZmyu12y+12KzMzU4cPH26NlxgSyvLLZNbUDquP63TiYC+xgB4AAAAABBNbg33nzp31yCOP6NNPP9Wnn36qYcOG6YorrtD27dutNmPGjFFOTo51e/vtt32uMX36dC1fvlzLli3TunXrVFxcrHHjxqmmpsZqM2nSJG3ZskVZWVnKysrSli1blJmZ2WqvM9h5h+FHd4iWI8px0vYsoAcAAAAAwSPSzie//PLLfe7/v//3/zR//nxt3LhRffv2lSS5XC6lpqbW+/iCggItWLBAzz//vEaMGCFJWrJkibp06aJVq1Zp9OjRys7OVlZWljZu3KiBAwdKkp599lkNHjxYO3bsUM+ePQP4CkODd+G8k21150WPPQAAAAAEj6CZY19TU6Nly5appKREgwcPto6///77Sk5O1plnnqmpU6cqLy/POrd582ZVVVVp1KhR1rH09HRlZGRo/fr1kqQNGzbI7XZboV6SBg0aJLfbbbWpT0VFhQoLC31ubVXxvmJJJ59f79UurfYDgOLc4oDVBAAAAABoHNuD/datW9WuXTu5XC798pe/1PLly9WnTx9J0tixY7V06VKtXr1ajz32mDZt2qRhw4apoqJCkpSbm6uoqCh16NDB55opKSnKzc212iQnJ9d53uTkZKtNfWbPnm3NyXe73erSpYu/XnLQaewe9l4xHWIkSRUFFQGrCQAAAADQOLYOxZeknj17asuWLTp8+LD++c9/6sYbb9TatWvVp08fXXvttVa7jIwMDRgwQN26ddNbb72lCRMmNHhN0zRlGIZ1/9ivG2pzvPvuu0933323db+wsLDNhvvG7mHvFd0hWpJUfrg8YDUBAAAAABrH9mAfFRWlM844Q5I0YMAAbdq0SX/605/0zDPP1Gmblpambt26aefOnZKk1NRUVVZWKj8/36fXPi8vT0OGDLHa7Nu3r8619u/fr5SUlAbrcrlccrlcLXptoaKxe9h7Rbc/EuzzCfYAAAAAYDfbh+IfzzRNa6j98Q4ePKg9e/YoLS1NknTeeefJ6XRq5cqVVpucnBxt27bNCvaDBw9WQUGBPvnkE6vNxx9/rIKCAqtNuGtyj317euwBAAAAIFjY2mN///33a+zYserSpYuKioq0bNkyvf/++8rKylJxcbFmzpypq6++Wmlpafr22291//33KykpSVdddZUkye12a8qUKbrnnnvUsWNHJSYmasaMGerXr5+1Sn7v3r01ZswYTZ061RoFcPPNN2vcuHGsiH9Es3vsCfYAAAAAYDtbg/2+ffuUmZmpnJwcud1u9e/fX1lZWRo5cqTKysq0detWPffcczp8+LDS0tJ06aWX6qWXXlJ8fLx1jSeeeEKRkZGaOHGiysrKNHz4cC1atEgOx9H92JcuXao77rjDWj1//PjxmjdvXqu/3mDV1MXzjg32J1urAAAAAAAQWIZpmqbdRYSCwsJCud1uFRQUKCEhwe5y/Gp2/GxVFldq2s5pSjwj8aTtK4oq9EjCI5Kk+0vvlzPGGegSAQAAACDsNDaHBt0ce7SuqtIqVRZXSmp8j31UuygZEbW99AzHBwAAAAB7EezDnHcYfmR0pKLioxr1GMMwmGcPAAAAAEGCYB/mjp1f35S58gR7AAAAAAgOBPsw5w3m0R2im/Q4gj0AAAAABAeCfZirKKqQJLniXU16nPeDgPJ8gj0AAAAA2MnW7e5gP+/CeVHtfOfX7969WwcOHGjwceWqDfQ7v9ipql5V9bZJSkpS165d/VQpAAAAAKA+BPswV1+w3717t3r36q3SstIGHzde43WuztWTjz6pDx/9sN42sTGxyv4ym3APAAAAAAFEsA9z9QX7AwcOqLSsVL/7xe/UPa17vY87sOqADn98WFcPulq3DL+lzvldObv0wN8f0IEDBwj2AAAAABBABPswV1l0JNjXs9Vd97Tu6tW1V72P25myU4d1WAmRCQ22AQAAAAAEHovnhbmG5tifTGRs7WdCVWX1z68HAAAAALQOgn2Ya26wd8Y4JUlVpQR7AAAAALATwT7MNTvYxx4J9vTYAwAAAICtCPZh7kRz7E/EG+yrS6v9XhMAAAAAoPEI9mGuxT32DMUHAAAAAFsR7MNcsxfPizmyeF5plUzT9HtdAAAAAIDGIdiHuZb22JseUzWVNX6vCwAAAADQOAT7MNfcYO+IcsiIMCQxHB8AAAAA7ESwD3MVRRWSJFe8q0mPMwzj6AJ6ZSygBwAAAAB2IdiHMdM0m91jL0mRsUfn2QMAAAAA7EGwD2PVZdXSkXXvmhPsnTHsZQ8AAAAAdiPYhzFvb710dDG8pmDLOwAAAACwH8E+jHnn10e1i7IWwmsKb499dSlz7AEAAADALgT7MNaS+fUSPfYAAAAAEAwI9mGspcHeWjyPOfYAAAAAYBuCfRhrcY99DD32AAAAAGC3SLsLgP/t3r1bBw4cOGm7Hz7/QZJUaVTqs88+s45nZ2c36nmsofj02AMAAACAbQj2bczu3bvVu1dvlZaVnrTt2TpbV+pKffzvj3XHeXfUOV9QUHDCx3uDPYvnAQAAAIB9CPZtzIEDB1RaVqrf/eJ36p7W/YRtD286rAPvHtA5vc/R2AljreMfbf1I89+Yr9LSE384EBlzZI49Q/EBAAAAwDYE+zaqe1p39era64RtvvrPVzqgA+rQoYNP2105uxr1HAzFBwAAAAD7sXheGKsurx1C73A5mvV4Fs8DAAAAAPsR7MNYdUVtsI+Mbt7ADWuOfVm1TNP0W10AAAAAgMYj2IexmooaSVKkq2XB3vSY1rUAAAAAAK2LYB/GWjoUP8IZoYjI2l8hhuMDAAAAgD0I9mHMGorfzB57wzCOrozPAnoAAAAAYAuCfRirKa8dPu+Ibl6PvXTMyvj02AMAAACALQj2YaylPfbS0ZXxq0ur/VITAAAAAKBpCPZhzFo8r5mr4kvsZQ8AAAAAdiPYhzFvj31zF8+TdHSOPUPxAQAAAMAWBPsw5l0Vnx57AAAAAAhdBPswZXpM1VS2bB97iTn2AAAAAGA3gn2Yqqmqkczar1syFJ8eewAAAACwF8E+THkXzpMhOaL8EOxLCPYAAAAAYAeCfZiy5te7ImUYRrOvExl7ZPE8euwBAAAAwBYE+zDljxXxpaNz7FkVHwAAAADsQbAPU9Ye9i1YOE86OhS/uozF8wAAAADADgT7MOUdit/iHvtYeuwBAAAAwE4E+zDlHYrf0h77yJijc+xNj9niugAAAAAATUOwD1M15UeG4kf7Zyi+zKMfFgAAAAAAWg/BPkz5a/E8h9OhiMjaXyPm2QMAAABA6yPYhylr8bwW9thLzLMHAAAAADsR7MOUvxbPk46ZZ0+wBwAAAIBWR7APU/5aPE+ixx4AAAAA7ESwD1MBGYpfRrAHAAAAgNZGsA9T/lo8T6LHHgAAAADsRLAPU34dih9TG+yrS1kVHwAAAABaG8E+TPlrH3tJiow9sngeQ/EBAAAAoNUR7MOUX4fixzAUHwAAAADsQrAPUwFZFZ8eewAAAABodQT7MGUNxWeOPQAAAACENIJ9mLKG4ke3fCi+NceeofgAAAAA0OoI9mHI9JhH97H3Z499OT32AAAAANDaCPZhqKayxvraL6vix7AqPgAAAADYhWAfhrzD8GVIEc6W/wp4PxyoLquWaZotvh4AAAAAoPEI9mHIO2Q+0hUpwzBafD3vUHzTY8pT5Wnx9QAAAAAAjUewD0Pe+fX+WDhPkhxRDunI5wPMswcAAACA1kWwD0P+3MNekowIwxqOzzx7AAAAAGhdBPsw5M8V8b2OnWcPAAAAAGg9tgb7+fPnq3///kpISFBCQoIGDx6sd955xzpvmqZmzpyp9PR0xcTEaOjQodq+fbvPNSoqKjRt2jQlJSUpLi5O48eP1969e33a5OfnKzMzU263W263W5mZmTp8+HBrvMSg5B0u73D5Zyi+xJZ3AAAAAGAXW4N9586d9cgjj+jTTz/Vp59+qmHDhumKK66wwvucOXP0+OOPa968edq0aZNSU1M1cuRIFRUVWdeYPn26li9frmXLlmndunUqLi7WuHHjVFNzdEu3SZMmacuWLcrKylJWVpa2bNmizMzMVn+9wcIaiu+Hre682PIOAAAAAOzhv2TXDJdffrnP/f/3//6f5s+fr40bN6pPnz568skn9dvf/lYTJkyQJC1evFgpKSl64YUXdMstt6igoEALFizQ888/rxEjRkiSlixZoi5dumjVqlUaPXq0srOzlZWVpY0bN2rgwIGSpGeffVaDBw/Wjh071LNnz9Z90UEgoEPx6bEHAAAAgFYVNHPsa2pqtGzZMpWUlGjw4MHatWuXcnNzNWrUKKuNy+XSJZdcovXr10uSNm/erKqqKp826enpysjIsNps2LBBbrfbCvWSNGjQILndbqtNfSoqKlRYWOhzaysCMRSfOfYAAAAAYA/bg/3WrVvVrl07uVwu/fKXv9Ty5cvVp08f5ebmSpJSUlJ82qekpFjncnNzFRUVpQ4dOpywTXJycp3nTU5OttrUZ/bs2dacfLfbrS5durTodQaTgPTYx9BjDwAAAAB2sD3Y9+zZU1u2bNHGjRv1q1/9SjfeeKP+85//WOcNw/Bpb5pmnWPHO75Nfe1Pdp377rtPBQUF1m3Pnj2NfUlBL5CL5zHHHgAAAABal+3BPioqSmeccYYGDBig2bNn66yzztKf/vQnpaamSlKdXvW8vDyrFz81NVWVlZXKz88/YZt9+/bVed79+/fXGQ1wLJfLZa3W7721FQFZPI859gAAAABgC9uD/fFM01RFRYW6d++u1NRUrVy50jpXWVmptWvXasiQIZKk8847T06n06dNTk6Otm3bZrUZPHiwCgoK9Mknn1htPv74YxUUFFhtwg372AMAAABA22Hrqvj333+/xo4dqy5duqioqEjLli3T+++/r6ysLBmGoenTp2vWrFnq0aOHevTooVmzZik2NlaTJk2SJLndbk2ZMkX33HOPOnbsqMTERM2YMUP9+vWzVsnv3bu3xowZo6lTp+qZZ56RJN18880aN25cWK6ILx3tsXdEB2Afe4I9AAAAALQqW4P9vn37lJmZqZycHLndbvXv319ZWVkaOXKkJOnee+9VWVmZbr31VuXn52vgwIF69913FR8fb13jiSeeUGRkpCZOnKiysjINHz5cixYtksNxNLQuXbpUd9xxh7V6/vjx4zVv3rzWfbFBxDtcnsXzAAAAACD02RrsFyxYcMLzhmFo5syZmjlzZoNtoqOjNXfuXM2dO7fBNomJiVqyZElzy2xzrKH4AZhjX1XO4nkAAAAA0JqCbo49As8aiu/PfexjmGMPAAAAAHYg2IehQCyexxx7AAAAALAHwT4MBWIf+2O3uzNN02/XBQAAAACcGME+zJgeUzWVgZtjb3pMeao8frsuAAAAAODECPZhxhvqJf8OxXe4HJJR+3VVGQvoAQAAAEBrIdiHGe/CeTKkCKf/fvyGYTDPHgAAAABsQLAPM8fuYW8Yhl+vfew8ewAAAABA6yDYhxnviviOaP8tnOdlBXt67AEAAACg1RDsw8yxPfb+5t3Lnjn2AAAAANB6CPZhxjvHPpDBnqH4AAAAANB6CPZhpqY8cEPxndFHFs8j2AMAAABAqyHYh5mA9thHMxQfAAAAAFobwT7MWME+OoBD8Vk8DwAAAABaDcE+zFir4kcFYFV85tgDAAAAQKsj2IcZK9i7ArjdHcEeAAAAAFoNwT7M1FQdCfbOACyeF1O7eB5z7AEAAACg9RDsw0wgg73VY88cewAAAABoNQT7MOOp9Ehijj0AAAAAtBUE+zBTU1nbYx8R5f8fPXPsAQAAAKD1EezDTGvMsWcoPgAAAAC0HoJ9mPH22AdyKH5VWZVM0/T79QEAAAAAdRHsw4ynKoBz7I8MxZcpmVUEewAAAABoDQT7MGPNsXf6/0fviHLIiDAkSZ4Kj9+vDwAAAACoi2AfZgI5x94wDKvXnmAPAAAAAK2DYB9mAjnHXjo6z55gDwAAAACtg2AfZgI5x146Os/eU0mwBwAAAIDWQLAPM4GcYy8d3fKOHnsAAAAAaB0E+zBimubROfaB7rEn2AMAAABAqyDYhxFPtUc6sgtdIBbPk5hjDwAAAACtjWAfRrzz6yUWzwMAAACAtoJgH0a88+tlSIbDCMhzMBQfAAAAAFoXwT6MHDu/3jACE+yd0SyeBwAAAACtiWAfRqw97AM0v146Zig+290BAAAAQKsg2IeRQO9hLx0T7MsJ9gAAAADQGgj2YSTQe9hLx8yxp8ceAAAAAFoFwT6MWEPxA9hjzxx7AAAAAGhdBPswYi2e1xpz7An2AAAAANAqCPZhpFXn2BPsAQAAAKBVEOzDSKvOsSfYAwAAAECrINiHkVaZYx/jtL6OUlTAngcAAAAAUItgH0ZaY459hDNCRoQhSXLJFbDnAQAAAADUItiHEe8WdIHssTcMw5pnT7AHAAAAgMAj2IcRb499IOfYS0fn2RPsAQAAACDwCPZhpDXm2EtHV8aPVnRAnwcAAAAAQLAPK60xx16SnNG1C+jRYw8AAAAAgUewDyOtMcdeEnPsAQAAAKAVEezDiDXHPoo59gAAAADQVhDsw4g1xz7QQ/FjGIoPAAAAAK2FYB9GWmuOPT32AAAAANB6CPZhhDn2AAAAAND2EOzDCHPsAQAAAKDtIdiHkdaeY88+9gAAAAAQeAT7MGLNsWcoPgAAAAC0GQT7MNJqc+wZig8AAAAArYZgH0asOfbOAM+xp8ceAAAAAFoNwT5MmB5TnqojPfZsdwcAAAAAbQbBPkx4qj3W14Eeiu9dPM8ll0yPGdDnAgAAAIBwR7APE94V8aVWGIp/pMfekKHq0uqAPhcAAAAAhDuCfZjwzq83HIYiHIH9sUc4I6zfrOpigj0AAAAABBLBPky01h72kmQYhiKia3+1qoqrAv58AAAAABDOCPZhorW2uvNyuGqfhx57AAAAAAgsgn2Y8A7Fb61gH+Gixx4AAAAAWgPBPky01h72XtZQ/CKCPQAAAAAEEsE+TLTmHHvpaI89Q/EBAAAAILAI9mHCCvYMxQcAAACANoVgHyY8Va27eJ53KD499gAAAAAQWAT7MOHtsW+1OfYu5tgDAAAAQGuwNdjPnj1b559/vuLj45WcnKwrr7xSO3bs8GkzefJkGYbhcxs0aJBPm4qKCk2bNk1JSUmKi4vT+PHjtXfvXp82+fn5yszMlNvtltvtVmZmpg4fPhzolxg0WnsoPtvdAQAAAEDrsDXYr127Vrfddps2btyolStXqrq6WqNGjVJJSYlPuzFjxignJ8e6vf322z7np0+fruXLl2vZsmVat26diouLNW7cONXU1FhtJk2apC1btigrK0tZWVnasmWLMjMzW+V1BgNrKH5rLZ4XzRx7AAAAAGgNkXY+eVZWls/9hQsXKjk5WZs3b9bFF19sHXe5XEpNTa33GgUFBVqwYIGef/55jRgxQpK0ZMkSdenSRatWrdLo0aOVnZ2trKwsbdy4UQMHDpQkPfvssxo8eLB27Nihnj17BugVBg9ru7uoVh6KT7AHAAAAgIAKqjn2BQUFkqTExESf4++//76Sk5N15plnaurUqcrLy7PObd68WVVVVRo1apR1LD09XRkZGVq/fr0kacOGDXK73Vaol6RBgwbJ7XZbbY5XUVGhwsJCn1soY7s7AAAAAGibgibYm6apu+++WxdeeKEyMjKs42PHjtXSpUu1evVqPfbYY9q0aZOGDRumiooKSVJubq6ioqLUoUMHn+ulpKQoNzfXapOcnFznOZOTk602x5s9e7Y1H9/tdqtLly7+eqm2YLs7AAAAAGibbB2Kf6zbb79dX3zxhdatW+dz/Nprr7W+zsjI0IABA9StWze99dZbmjBhQoPXM01ThmFY94/9uqE2x7rvvvt09913W/cLCwtDOtzbNceeHnsAAAAACKyg6LGfNm2a3njjDa1Zs0adO3c+Ydu0tDR169ZNO3fulCSlpqaqsrJS+fn5Pu3y8vKUkpJitdm3b1+da+3fv99qczyXy6WEhASfWyhr7Tn21qr4JdXy1Hha5TkBAAAAIBzZGuxN09Ttt9+uV199VatXr1b37t1P+piDBw9qz549SktLkySdd955cjqdWrlypdUmJydH27Zt05AhQyRJgwcPVkFBgT755BOrzccff6yCggKrTVvX6nPso4/+alUUVrTKcwIAAABAOLJ1KP5tt92mF154Qa+//rri4+Ot+e5ut1sxMTEqLi7WzJkzdfXVVystLU3ffvut7r//fiUlJemqq66y2k6ZMkX33HOPOnbsqMTERM2YMUP9+vWzVsnv3bu3xowZo6lTp+qZZ56RJN18880aN25cWKyIL7X+HHvDYahKVXLKqYqCCsV0iGmV5wUAAACAcGNrsJ8/f74kaejQoT7HFy5cqMmTJ8vhcGjr1q167rnndPjwYaWlpenSSy/VSy+9pPj4eKv9E088ocjISE2cOFFlZWUaPny4Fi1aJIfjaIhdunSp7rjjDmv1/PHjx2vevHmBf5FBwjvHPsLZeoM0ylUup5wqP1zeas8JAAAAAOHG1mBvmuYJz8fExGjFihUnvU50dLTmzp2ruXPnNtgmMTFRS5YsaXKNbUVr99hLUoUqFK94lRcQ7AEAAAAgUIJi8TwEnnfxvNYM9uWqDfQVBcyxBwAAAIBAIdiHidZePE+q7bGXRI89AAAAAAQQwT5MtPY+9tLRHnvm2AMAAABA4BDsw4S3x7619rGXGIoPAAAAAK2BYB8m7Jhjz1B8AAAAAAg8gn0Y8NR4ZNbU7kDAUHwAAAAAaFsI9mHAO79eYlV8AAAAAGhrCPZhwDu/XpIiIlvvR+4dik+wBwAAAIDAIdiHAe/8+ghnhIwIo9We1xqKzxx7AAAAAAgYgn0YsGMPe4k59gAAAADQGgj2YcA7x741t7qTGIoPAAAAAK2BYB8GrB77Vlw4T2IoPgAAAAC0BoJ9GLD2sG/lofjeHvvqsmqfBfwAAAAAAP5DsA8Dds2x9wZ7iV57AAAAAAgUgn0Y8FTaM8feI48csbUfJjDPHgAAAAACg2AfBqyh+K08x16SnO2ckuixBwAAAIBAIdiHAbvm2EtSZLtISWx5BwAAAACBQrAPA3bNsZckZ3xtjz1D8QEAAAAgMAj2YcCuOfYSQ/EBAAAAINAI9mHAzjn23qH49NgDAAAAQGC0ONgXFhbqtddeU3Z2tj/qQQDYOhTf22PPHHsAAAAACIgmB/uJEydq3rx5kqSysjINGDBAEydOVP/+/fXPf/7T7wWi5excPI+h+AAAAAAQWE0O9h988IEuuugiSdLy5ctlmqYOHz6sp556Sr///e/9XiBazs459gzFBwAAAIDAanLSKygoUGJioiQpKytLV199tWJjY3XZZZdp586dfi8QLRcU+9gzFB8AAAAAAqLJwb5Lly7asGGDSkpKlJWVpVGjRkmS8vPzFR0d7fcC0XJ2zrGnxx4AAAAAAiuyqQ+YPn26rr/+erVr107dunXT0KFDJdUO0e/Xr5+/64MfWMHejh77eObYAwAAAEAgNTnY33rrrfrRj36kPXv2aOTIkYqIqO30P+2005hjH6Q8VUfm2DuZYw8AAAAAbU2Tg70kDRgwQAMGDPA5dtlll/mlIPgfc+wBAAAAoO1qchfuNddco0ceeaTO8T/84Q/6yU9+4pei4F+27mN/zFB80zRb/fkBAAAAoK1rcrBfu3Ztvb3zY8aM0QcffOCXouBfds6x9w7F91R5VF1e3erPDwAAAABtXZODfXFxsaKiouocdzqdKiws9EtR8C9b59jHRkpG7dcMxwcAAAAA/2ty0svIyNBLL71U5/iyZcvUp08fvxQF/7Jzjr0RYciV4JLEAnoAAAAAEAhNXjzvgQce0NVXX62vv/5aw4YNkyS99957evHFF/Xyyy/7vUC0jGmatg7Fl6To9tGqKKhgyzsAAAAACIAmB/vx48frtdde06xZs/TKK68oJiZG/fv316pVq3TJJZcEoka0gKfaIx1Zs86OxfMkKdodrQIV0GMPAAAAAAHQrO3uLrvsMra3CxHe+fWSPXPsJcnlrh2Kzxx7AAAAAPC/ZgV7Sdq8ebOys7NlGIb69Omjc845x591wU+88+tlSBGR9gT7aHe0JDEUHwAAAAACoMnBPi8vT9ddd53ef/99tW/fXqZpqqCgQJdeeqmWLVumTp06BaJONNOx8+sNw7Clhuj2tcGeofgAAAAA4H9N7sKdNm2aCgsLtX37dh06dEj5+fnatm2bCgsLdccddwSiRrSAFextml8vHTMUnx57AAAAAPC7JvfYZ2VladWqVerdu7d1rE+fPvrzn/+sUaNG+bU4tJyde9h7McceAAAAAAKnyWnP4/HI6XTWOe50OuXxeOp5BOxk91Z30tE59gzFBwAAAAD/a3KwHzZsmO6880798MMP1rHvv/9ed911l4YPH+7X4tBy3sXzbA32zLEHAAAAgIBpcrCfN2+eioqKdOqpp+r000/XGWecoe7du6uoqEhz584NRI1ogaCaY89QfAAAAADwuybPse/SpYs+++wzrVy5Ul9++aVM01SfPn00YsSIQNSHFgqGOfZsdwcAAAAAgdPsfexHjhypkSNH+rMWBEAwzLH39tgzFB8AAAAA/K9Jwd7j8WjRokV69dVX9e2338owDHXv3l3XXHONMjMzbdsnHQ0Lpjn29NgDAAAAgP81eny2aZoaP368brrpJn3//ffq16+f+vbtq++++06TJ0/WVVddFcg60UzBMMf+2FXxTdO0rQ4AAAAAaIsa3WO/aNEiffDBB3rvvfd06aWX+pxbvXq1rrzySj333HP62c9+5vci0XyeyiNz7KPs38fe9JiqLK6UK95lWy0AAAAA0NY0Ou29+OKLuv/+++uEeql2C7zf/OY3Wrp0qV+LQ8tZQ/Ft7LF3xjoVEVn7q8Y8ewAAAADwr0YH+y+++EJjxoxp8PzYsWP1+eef+6Uo+E8wzLE3DIMt7wAAAAAgQBod7A8dOqSUlJQGz6ekpCg/P98vRcF/gmGOvcSWdwAAAAAQKI0O9jU1NYqMbHhKvsPhUHV1tV+Kgv8Ewxx7iS3vAAAAACBQGr14nmmamjx5slyu+hc+q6ggsAWjYBiKL0kxHWIkSWX5ZbbWAQAAAABtTaOD/Y033njSNqyIH3yCYfE8SYpJrA325fkMxQcAAAAAf2p0sF+4cGEg60CAeIfi2x3soxNr59iXHaLHHgAAAAD8yd6J1wg4b4+93XPsvT32BHsAAAAA8C+CfRsXNEPxOxDsAQAAACAQCPZtnLUqvpMeewAAAABoiwj2bVzQ9NgT7AEAAAAgIBoV7M8991zl5+dLkv7v//5PpaWlAS0K/uOpOrJ4nt3b3RHsAQAAACAgGhXss7OzVVJSIkl6+OGHVVxcHNCi4D/W4nkMxQcAAACANqlR292dffbZ+vnPf64LL7xQpmnqj3/8o9q1a1dv2wcffNCvBaL5TNNUTWXwDcU3TVOGYdhaDwAAAAC0FY0K9osWLdJDDz2kN998U4Zh6J133lFkZN2HGoZBsA8ipseUzNqvg6XH3qwxVVlUKVeCy9Z6AAAAAKCtaFSw79mzp5YtWyZJioiI0Hvvvafk5OSAFoaW8/bWS/bPsY+MiZTD5VBNRY3KDpUR7AEAAADAT5rcjevxeAj1IcK7cJ4kRUTa22NvGAbz7AEAAAAgABrVY3+8r7/+Wk8++aSys7NlGIZ69+6tO++8U6effrq/60MLHLtwXjDMaY9JjFFxTjHBHgAAAAD8qMnduCtWrFCfPn30ySefqH///srIyNDHH3+svn37auXKlYGoEc1kbXVn88J5XvTYAwAAAID/NbnH/je/+Y3uuusuPfLII3WO//rXv9bIkSP9VhxaxjvHPiLK3mH4XgR7AAAAAPC/Jie+7OxsTZkypc7xX/ziF/rPf/7TpGvNnj1b559/vuLj45WcnKwrr7xSO3bs8GljmqZmzpyp9PR0xcTEaOjQodq+fbtPm4qKCk2bNk1JSUmKi4vT+PHjtXfvXp82+fn5yszMlNvtltvtVmZmpg4fPtykekONdyi+I5IeewAAAABoq5oc7Dt16qQtW7bUOb5ly5YmL6q3du1a3Xbbbdq4caNWrlyp6upqjRo1SiUlJVabOXPm6PHHH9e8efO0adMmpaamauTIkSoqKrLaTJ8+XcuXL9eyZcu0bt06FRcXa9y4caqpOboq/KRJk7RlyxZlZWUpKytLW7ZsUWZmZlNffkixhuLbvCK+F8EeAAAAAPyvyUPxp06dqptvvlnffPONhgwZIsMwtG7dOj366KO65557mnStrKwsn/sLFy5UcnKyNm/erIsvvlimaerJJ5/Ub3/7W02YMEGStHjxYqWkpOiFF17QLbfcooKCAi1YsEDPP/+8RowYIUlasmSJunTpolWrVmn06NHKzs5WVlaWNm7cqIEDB0qSnn32WQ0ePFg7duxQz549m/ptCAnHLp4XDAj2AAAAAOB/TQ72DzzwgOLj4/XYY4/pvvvukySlp6dr5syZuuOOO1pUTEFBgSQpMTFRkrRr1y7l5uZq1KhRVhuXy6VLLrlE69ev1y233KLNmzerqqrKp016eroyMjK0fv16jR49Whs2bJDb7bZCvSQNGjRIbrdb69evrzfYV1RUqKKiwrpfWFjYotdmB+8cexbPAwAAAIC2q8lduYZh6K677tLevXtVUFCggoIC7d27V3feeWeLtlQzTVN33323LrzwQmVkZEiScnNzJUkpKSk+bVNSUqxzubm5ioqKUocOHU7Ypr5pAsnJyVab482ePduaj+92u9WlS5dmvza7eIfiB1uPfXl+uc2VAAAAAEDb0aLEFx8fr/j4eL8Ucvvtt+uLL77Qiy++WOfc8R8YmKZ50g8Rjm9TX/sTXee+++6zPrgoKCjQnj17GvMygoq1eB5z7AEAAACgzQqKrtxp06bpjTfe0Jo1a9S5c2freGpqqiTV6VXPy8uzevFTU1NVWVmp/Pz8E7bZt29fnefdv39/ndEAXi6XSwkJCT63UMM+9gAAAADQ9tka7E3T1O23365XX31Vq1evVvfu3X3Od+/eXampqVq5cqV1rLKyUmvXrtWQIUMkSeedd56cTqdPm5ycHG3bts1qM3jwYBUUFOiTTz6x2nz88ccqKCiw2rRF1j72QTYUn2APAAAAAP7T5MXz/Om2227TCy+8oNdff13x8fFWz7zb7VZMTIwMw9D06dM1a9Ys9ejRQz169NCsWbMUGxurSZMmWW2nTJmie+65Rx07dlRiYqJmzJihfv36Wavk9+7dW2PGjNHUqVP1zDPPSJJuvvlmjRs3rs2uiC8Fb499dXm1qsqq5Ixx2lwRAAAAAIS+JnXlVlVV6dJLL9V///tfvzz5/PnzVVBQoKFDhyotLc26vfTSS1abe++9V9OnT9ett96qAQMG6Pvvv9e7777rM7f/iSee0JVXXqmJEyfqggsuUGxsrP71r3/J4TgaaJcuXap+/fpp1KhRGjVqlPr376/nn3/eL68jWAXbdndR8VEyHLVrGtBrDwAAAAD+0aQee6fTqW3btrVo9ftjmaZ50jaGYWjmzJmaOXNmg22io6M1d+5czZ07t8E2iYmJWrJkSXPKDFnW4nlB0mNvGIZiEmNUur9UZYfKlHBK6K1bAAAAAADBpslduT/72c+0YMGCQNQCP/NUHtnuLio4euwl5tkDAAAAgL81eY59ZWWl/va3v2nlypUaMGCA4uLifM4//vjjfisOLVNTHVw99hLBHgAAAAD8rcnBftu2bTr33HMlqc5ce38N0Yd/eHvsCfYAAAAA0HY1OdivWbMmEHUgAIJt8TyJYA8AAAAA/tbsxPfVV19pxYoVKiurDWiNWQgPrcva7i6KHnsAAAAAaKuaHOwPHjyo4cOH68wzz9SPf/xj5eTkSJJuuukm3XPPPX4vEM1Hjz0AAAAAtH1NTnx33XWXnE6ndu/erdjYWOv4tddeq6ysLL8Wh5apqQzexfPKD5XbXAkAAAAAtA1NnmP/7rvvasWKFercubPP8R49eui7777zW2FoOe9QfHrsAQAAAKDtanLiKykp8emp9zpw4IBcLpdfioJ/eIfiM8ceAAAAANquJgf7iy++WM8995x13zAMeTwe/eEPf9Cll17q1+LQMtbieUE4FJ9gDwAAAAD+0eSh+H/4wx80dOhQffrpp6qsrNS9996r7du369ChQ/roo48CUSOayTvHnqH4AAAAANB2NTnx9enTR1988YV+9KMfaeTIkSopKdGECRP073//W6effnogakQzWUPxg7DHvrK40vrgAQAAAADQfE3usZek1NRUPfzww/6uBX5kekyZNaak4Jpj73K7JEOSKZXll6ldSju7SwIAAACAkNasYJ+fn68FCxYoOztbhmGod+/e+vnPf67ExER/14dmMqtN6+tgGoof4YhQdPtoleeXq+wQwR4AAAAAWqrJiW/t2rXq3r27nnrqKeXn5+vQoUN66qmn1L17d61duzYQNaIZjg32wTQUX2KePQAAAAD4U5N77G+77TZNnDhR8+fPl8NRGxhramp066236rbbbtO2bdv8XiSazhvsDYchI8KwuRpfMYkxyv86n2APAAAAAH7Q5B77r7/+Wvfcc48V6iXJ4XDo7rvv1tdff+3X4tB8nurg2+rOix57AAAAAPCfJgf7c889V9nZ2XWOZ2dn6+yzz/ZHTfADsyr4Fs7zItgDAAAAgP80aij+F198YX19xx136M4779RXX32lQYMGSZI2btyoP//5z3rkkUcCUyWazDsUP5gWzvMi2AMAAACA/zQq2J999tkyDEOmeXRBtnvvvbdOu0mTJunaa6/1X3VoNm+wZyg+AAAAALRtjQr2u3btCnQd8DPvHPtg7rEvP1RucyUAAAAAEPoaFey7desW6DrgZ9Yce3rsAQAAAKBNa/J2d5L0/fff66OPPlJeXp48Ho/PuTvuuMMvhaFlmGMPAAAAAOGhycF+4cKF+uUvf6moqCh17NhRhnF0j3TDMAj2QcKaY8+q+AAAAADQpjU52D/44IN68MEHdd999ykiIvh6g1GLfewBAAAAIDw0OZmXlpbquuuuI9QHuVAYil9+uFyeGs9JWgMAAAAATqTJqW/KlCl6+eWXA1EL/CiYt7uL7hBtfV1+mJXxAQAAAKAlmjwUf/bs2Ro3bpyysrLUr18/OZ1On/OPP/6434pD83lXxY+ICr4ee4fToaj4KFUWVarsUJliO8baXRIAAAAAhKwmB/tZs2ZpxYoV6tmzpyTVWTwPwSGY59hLtcPxvcEeAAAAANB8TQ72jz/+uP7+979r8uTJASgH/hLMc+yl2mBf8F0BwR4AAAAAWqjJqc/lcumCCy4IRC3wo2CeYy8ds4BePnPsAQAAAKAlmhzs77zzTs2dOzcQtcCPgnkfe4kt7wAAAADAX5o8FP+TTz7R6tWr9eabb6pv3751Fs979dVX/VYcms9aPC8yeIfiSwR7AAAAAGipJgf79u3ba8KECYGoBX7k3R8+2HvsSw+W2lwJAAAAAIS2Jgf7hQsXBqIO+JnVYx+ki+fFJtVucVe6n2APAAAAAC0RnKkPLRbsc+zjkuMkSSV5JTZXAgAAAAChrck99t27dz/hfvXffPNNiwqCf1jBPjK4gz099gAAAADQMk0O9tOnT/e5X1VVpX//+9/KysrS//zP//irLrSQtY99VHAOyojtVDsUnx57AAAAAGiZJgf7O++8s97jf/7zn/Xpp5+2uCD4h6fqyOJ5QbqPvTUUf3+JTI8pI6LhUSAAAAAAgIb5rTt37Nix+uc//+mvy6GFrB77IF08L65TbbA3a0yV5bPlHQAAAAA0l99S3yuvvKLExER/XQ4tZM2xD9Iee0eUQ9HtoyUxzx4AAAAAWqLJQ/HPOeccn8XzTNNUbm6u9u/fr7/85S9+LQ7NF+yr4ku18+zLD5erJK9ESb2S7C4HAAAAAEJSk4P9lVde6XM/IiJCnTp10tChQ9WrVy9/1YUWiDzmxxqsQ/Gl2nn2h3YeYgE9AAAAAGiBJgf7hx56KBB1wI+ODfbBOhRfYi97AAAAAPCH4O3ORbNZwd6QDEfwrjZ/7Mr4AAAAAIDmaXSPfUREhM/c+voYhqHq6uoWF4WWccopqXZ+/cl+ZnZiL3sAAAAAaLlGB/vly5c3eG79+vWaO3euTNP0S1FoGW+PfTAMw8/Ozm7wXH5FviTp+x3f67PPPqu3TVJSkrp27RqQ2gAAAACgLWh0sL/iiivqHPvyyy9133336V//+peuv/56/e53v/NrcWgeb4+9nQvnHSg4IEm64YYbGmzTV331E/1EG97boFvOu6XeNrExscr+MptwDwAAAAANaPLieZL0ww8/6KGHHtLixYs1evRobdmyRRkZGf6uDc0UDD32RaVFkqQ7r7hT52ecX2+b0m9L9cPSH9QjqYeW3LKkzvldObv0wN8f0IEDBwj2AAAAANCAJgX7goICzZo1S3PnztXZZ5+t9957TxdddFGgakMzeYN9RJT9ayN27thZvbrWvw1iUWSRftAPUpkabAMAAAAAOLFGB/s5c+bo0UcfVWpqql588cV6h+YjOFiL50XaP8f+RKLaRUmSqkqq5KnxKMJh/wcRAAAAABBqGh3sf/Ob3ygmJkZnnHGGFi9erMWLF9fb7tVXX/VbcWieYOqxP5GodlGSIcmUKosrFe2OtrskAAAAAAg5jQ72P/vZz4J66zQcFQxz7BvDiDAU1S5KlUWVBHsAAAAAaKZGB/tFixYFsAz4kzUUP8iDvSS54l21wb6wUjrF7moAAAAAIPQE91htNIs1FN/G7e4ayzvPvqKowuZKAAAAACA0BX/yQ5NZQ/Gjgr/HPiqeYA8AAAAALUGwb4O8Q/FDocfeFe+SJFUWVdpcCQAAAACEpuBPfmiyUFk8T5KiEmp77An2AAAAANA8BPs2KKQWz2tX22PPUHwAAAAAaB6CfRsUUovnxdNjDwAAAAAtEfzJD00WSovnuRLosQcAAACAliDYt0GhtHgePfYAAAAA0DLBn/zQZCG1eN6Rfeyry6tVU1VjczUAAAAAEHoI9m1QKAV7Z6xTRoQhiV57AAAAAGgOgn0bZA3Fjwr+H69hGNaWd8yzBwAAAICmC/7khyYLpR57SXLF1y6gR489AAAAADQdwb4NCqXt7qSj8+wJ9gAAAADQdLYmvw8++ECXX3650tPTZRiGXnvtNZ/zkydPlmEYPrdBgwb5tKmoqNC0adOUlJSkuLg4jR8/Xnv37vVpk5+fr8zMTLndbrndbmVmZurw4cMBfnX28Q7FD7Uee4biAwAAAEDT2RrsS0pKdNZZZ2nevHkNthkzZoxycnKs29tvv+1zfvr06Vq+fLmWLVumdevWqbi4WOPGjVNNzdEV1idNmqQtW7YoKytLWVlZ2rJlizIzMwP2uuwWcj328cyxBwAAAIDmirTzyceOHauxY8eesI3L5VJqamq95woKCrRgwQI9//zzGjFihCRpyZIl6tKli1atWqXRo0crOztbWVlZ2rhxowYOHChJevbZZzV48GDt2LFDPXv29O+LCgJWj31UiPTYJzDHHgAAAACaK+i7dN9//30lJyfrzDPP1NSpU5WXl2ed27x5s6qqqjRq1CjrWHp6ujIyMrR+/XpJ0oYNG+R2u61QL0mDBg2S2+222tSnoqJChYWFPrdQEWqL53l77An2AAAAANB0QR3sx44dq6VLl2r16tV67LHHtGnTJg0bNkwVFbVDtnNzcxUVFaUOHTr4PC4lJUW5ublWm+Tk5DrXTk5OttrUZ/bs2dacfLfbrS5duvjxlQWOp8qjiCM/1lAZiu9qxxx7AAAAAGguW4fin8y1115rfZ2RkaEBAwaoW7dueuuttzRhwoQGH2eapgzDsO4f+3VDbY5333336e6777buFxYWhkS4r6k4urZAyPTYJ9BjDwAAAADNFRpdukekpaWpW7du2rlzpyQpNTVVlZWVys/P92mXl5enlJQUq82+ffvqXGv//v1Wm/q4XC4lJCT43EKBp9JjfR0yPfbHrIpvmqbN1QAAAABAaAmN5HfEwYMHtWfPHqWlpUmSzjvvPDmdTq1cudJqk5OTo23btmnIkCGSpMGDB6ugoECffPKJ1ebjjz9WQUGB1aYt8fbYG5HGCUckBBPvHHtPlcdnxAEAAAAA4ORsHYpfXFysr776yrq/a9cubdmyRYmJiUpMTNTMmTN19dVXKy0tTd9++63uv/9+JSUl6aqrrpIkud1uTZkyRffcc486duyoxMREzZgxQ/369bNWye/du7fGjBmjqVOn6plnnpEk3XzzzRo3blybXBG/pvxosA8Vka5IRTgj5KnyqKKoQpHRQT1DBAAAAACCiq0J6tNPP9Wll15q3ffOab/xxhs1f/58bd26Vc8995wOHz6stLQ0XXrppXrppZcUHx9vPeaJJ55QZGSkJk6cqLKyMg0fPlyLFi2Sw3F0fvnSpUt1xx13WKvnjx8/XvPmzWulV9m6PBW1Q/FDKdhLtVvelR0sU2VRpeI6xdldDgAAAACEDFuD/dChQ084p3rFihUnvUZ0dLTmzp2ruXPnNtgmMTFRS5YsaVaNoaamMvR67KXa4fjeYA8AAAAAaLyQmmOPk/OU1/bYh8rCeV7HLqAHAAAAAGi80Ep/OClr8TxHiPXYt2PLOwAAAABoDoJ9G2MFe2doBXt67AEAAACgeQj2bYx3H/uQm2OfQI89AAAAADQHwb6N8W53FxEZWj9aeuwBAAAAoHlCK/3hpKyh+KHWYx9f22NPsAcAAACApiHYtzHWUPwQm2Mf7Y6WJFUcJtgDAAAAQFMQ7NuYUO2xj25fG+wriytVU1VjczUAAAAAEDoI9m2Mdx/7UAv2zjintS5ARQG99gAAAADQWAT7NsbbYx9qi+cZhmH12pcfLre5GgAAAAAIHaGV/nBSNZWhORRfEsEeAAAAAJqBYN/GeCpCc/E8iWAPAAAAAM1BsG9jvPvYGw6CPQAAAACEA4J9G+Pd7i7CGXo/WoI9AAAAADRd6KU/nFCobncnHRPs8wn2AAAAANBYBPs2JqSDfQd67AEAAACgqQj2bYw1xz4Ug/0xQ/FN07S5GgAAAAAIDQT7NsY7xz4UV8V3uV2SJE+1R1UlVTZXAwAAAAChgWDfxniH4kdEht6P1uF0KKpdlCSG4wMAAABAY4Ve+sMJWfvYh+BQfElyta/ttSfYAwAAAEDjEOzbmFCeYy+x5R0AAAAANBXBvo0J5Tn2EsEeAAAAAJqKYN+GeGo88lTVBvtQnGMvEewBAAAAoKlCM/2hXoZh6OIXLtbf9DdFRIfmj5ZgDwAAAABNE5rpD/UyIgy5e7q1V3tlRDAUHwAAAADCAcEeQSWmfYwkgj0AAAAANBbBHkElukNtj31lUaXMatPmagAAAAAg+BHsEVSccU5r4b/q4mqbqwEAAACA4EewR1AxDMOaZ19dRLAHAAAAgJMh2CPoEOwBAAAAoPEI9gg63mBfU1xjcyUAAAAAEPwI9gg6LrdLEj32AAAAANAYBHsEHe/K+AR7AAAAADg5gj2CDnPsAQAAAKDxCPYIOtYc+yLm2AMAAADAyRDsEXTosQcAAACAxiPYI+h4g71ZbSpGMTZXAwAAAADBjWCPoONwOuSMc0qS4hVvczUAAAAAENwI9ghK3pXxE5RgcyUAAAAAENwI9ghK3uH49NgDAAAAwIkR7BGUvMGeHnsAAAAAODGCPYISPfYAAAAA0DgEewQlgj0AAAAANA7BHkGJofgAAAAA0DgEewQleuwBAAAAoHEI9ghK3mDfTu3kqfLYXA0AAAAABC+CPYJSVLsoGZGGJKkst8zmagAAAAAgeBHsEZQMw5Czg1OSVLKnxOZqAAAAACB4EewRtKxgv5dgDwAAAAANIdgjaHmDfeneUpsrAQAAAIDgRbBH0GIoPgAAAACcHMEeQYtgDwAAAAAnR7BH0HImHhmK/32pTI9pczUAAAAAEJwI9ghakQmRqlGNPJUeFX5faHc5AAAAABCUCPYIWkaEoXzlS5IOfXXI5moAAAAAIDgR7BHUCPYAAAAAcGIEewS1Q6oN9Plf59tcCQAAAAAEJ4I9gpo32NNjDwAAAAD1I9gjqBHsAQAAAODECPYIascGe9NkyzsAAAAAOB7BHkHtsA5LEVJVSZVK8krsLgcAAAAAgg7BHkGtRjWKSY2RxHB8AAAAAKgPwR5BL65znCSCPQAAAADUh2CPoBfXhWAPAAAAAA0h2CPoeXvs2cseAAAAAOoi2CPoxXaOlUSPPQAAAADUh2CPoMdQfAAAAABomK3B/oMPPtDll1+u9PR0GYah1157zee8aZqaOXOm0tPTFRMTo6FDh2r79u0+bSoqKjRt2jQlJSUpLi5O48eP1969e33a5OfnKzMzU263W263W5mZmTp8+HCAXx38xdtjX55frrJDZTZXAwAAAADBxdZgX1JSorPOOkvz5s2r9/ycOXP0+OOPa968edq0aZNSU1M1cuRIFRUVWW2mT5+u5cuXa9myZVq3bp2Ki4s1btw41dTUWG0mTZqkLVu2KCsrS1lZWdqyZYsyMzMD/vrgH5ExkYpPj5ckHfqaXnsAAAAAOFaknU8+duxYjR07tt5zpmnqySef1G9/+1tNmDBBkrR48WKlpKTohRde0C233KKCggItWLBAzz//vEaMGCFJWrJkibp06aJVq1Zp9OjRys7OVlZWljZu3KiBAwdKkp599lkNHjxYO3bsUM+ePVvnxaJFOpzeQUU/FOnQV4d0yvmn2F0OAAAAAASNoJ1jv2vXLuXm5mrUqFHWMZfLpUsuuUTr16+XJG3evFlVVVU+bdLT05WRkWG12bBhg9xutxXqJWnQoEFyu91Wm/pUVFSosLDQ5wb7JJ6RKIl59gAAAABwvKAN9rm5uZKklJQUn+MpKSnWudzcXEVFRalDhw4nbJOcnFzn+snJyVab+syePduak+92u9WlS5cWvR60jDfYs+UdAAAAAPgK2mDvZRiGz33TNOscO97xbeprf7Lr3HfffSooKLBue/bsaWLl8KcOp9d+eEOPPQAAAAD4Ctpgn5qaKkl1etXz8vKsXvzU1FRVVlYqPz//hG327dtX5/r79++vMxrgWC6XSwkJCT432Ieh+AAAAABQv6AN9t27d1dqaqpWrlxpHausrNTatWs1ZMgQSdJ5550np9Pp0yYnJ0fbtm2z2gwePFgFBQX65JNPrDYff/yxCgoKrDYIfomn1wb7kn0lqiiqsLkaAAAAAAgetq6KX1xcrK+++sq6v2vXLm3ZskWJiYnq2rWrpk+frlmzZqlHjx7q0aOHZs2apdjYWE2aNEmS5Ha7NWXKFN1zzz3q2LGjEhMTNWPGDPXr189aJb93794aM2aMpk6dqmeeeUaSdPPNN2vcuHGsiB9CottHKzYpVqUHSpX/Tb5Sz0q1uyQAAAAACAq2BvtPP/1Ul156qXX/7rvvliTdeOONWrRoke69916VlZXp1ltvVX5+vgYOHKh3331X8fHx1mOeeOIJRUZGauLEiSorK9Pw4cO1aNEiORwOq83SpUt1xx13WKvnjx8/XvPmzWulVwl/STwjUaUHSnVwx0GCPQAAAAAcYWuwHzp0qEzTbPC8YRiaOXOmZs6c2WCb6OhozZ07V3Pnzm2wTWJiopYsWdKSUhEEkvsna+/Gvcr5LEd9J/a1uxwAAAAACAq2BnugMbKzsyVJ1Z2qJUk73t+hxM8SG/34pKQkde3aNSC1AQAAAIDdCPYIWgcKDkiSbrjhBklSmtJ0i27R7o936/bzbm/0dWJjYpX9ZTbhHgAAAECbRLBH0CoqLZIk3XnFnTo/43yZ1aa+/sPXivXEauFtC+Vs7zzpNXbl7NIDf39ABw4cINgDAAAAaJMI9gh6nTt2Vq+uvSRJ+0/Zr8I9hUqqTlJa1zSbKwMAAAAA+wXtPvZAfdzd3JKkgu8KbK4EAAAAAIIDwR4hpX3X9pKkgt0EewAAAACQCPYIMVaP/e6CE26VCAAAAADhgmCPkBJ/SryMCEOVxZUqzy+3uxwAAAAAsB3BHiHF4XQoPj1eknT4u8P2FgMAAAAAQYBgj5DjHY5fuLvQ5koAAAAAwH4Ee4Qcd9faYH9492F7CwEAAACAIECwR8jxBvuC71hADwAAAAAI9gg5CZ0TahfQK6pU+WEW0AMAAAAQ3gj2CDmOKIfapbWTxH72AAAAAECwR0g6djg+AAAAAIQzgj1CkhXs6bEHAAAAEOYI9ghJ3i3v6LEHAAAAEO4I9ghJ7i5uyZAqCitUXsACegAAAADCF8EeIckR5VB8Wrwkeu0BAAAAhDeCPUJWQtcESdLh7w7bWwgAAAAA2Ihgj5DVoXsHSdKhnYdsrgQAAAAA7EOwR8hK6pUkScr/Ol81lTU2VwMAAAAA9iDYI2TFpcQpun20PNUeHfqaXnsAAAAA4Ylgj5BlGIbVa3/gywM2VwMAAAAA9iDYI6RZwT6bYA8AAAAgPBHsEdI69uooSSrYXaDKkkqbqwEAAACA1kewR0iL6RCjuJQ4yZQO/veg3eUAAAAAQKsj2CPkMc8eAAAAQDgj2CPkJfWuDfYHv6THHgAAAED4Idgj5HU8s6NkSMW5xSrLL7O7HAAAAABoVQR7hLyouCi5u7olSQd30GsPAAAAILwQ7NEmMM8eAAAAQLgi2KNNOHY/e9M0ba4GAAAAAFoPwR5tQuIZiYqIjFD54XKV7CuxuxwAAAAAaDUEe7QJjiiHOpzWQRLD8QEAAACEF4I92gzm2QMAAAAIRwR7tBnWfvY7DspT47G5GgAAAABoHQR7tBntT20vZ6xTVaVVOrzrsN3lAAAAAECrINijzTAiDHXq00mStH/7fpurAQAAAIDWQbBHm9Kpb22wz9ueZ3MlAAAAANA6CPZoU7zBvuC7AlUUVthcDQAAAAAEHsEebUq0O1oJnRMkSfuzGY4PAAAAoO0j2KPN6ZTBPHsAAAAA4YNgjzYnuW+yJGn/f/bLNE2bqwEAAACAwCLYo83pcFoHRUZHqrKoUhW5zLMHAAAA0LYR7NHmRERGqGOvjpKk0q9Lba4GAAAAAAKLYI82yTscv/Qbgj0AAACAto1gjzapU5/aBfTK95YrWtE2VwMAAAAAgUOwR5sUmxSrdqntJFPqru52lwMAAAAAAUOwR5vVqW9tr/0ZOsPmSgAAAAAgcAj2aLOODfZsewcAAACgrSLYo83q2KOjjEhDbrlV/E2x3eUAAAAAQEAQ7NFmOaIciukaI0nKW59nczUAAAAAEBgEe7RpsafHSpLyNhDsAQAAALRNBHu0ad5gf+izQ6osqbS5GgAAAADwP4I92jRnolP5ypenyqNv3//W7nIAAAAAwO8I9mjTDMPQV/pKkvRV1lc2VwMAAAAA/kewR5vnDfZfZ31tcyUAAAAA4H8Ee7R5u7RLRqShQ18d0qGvDtldDgAAAAD4FcEebV6lKpV4dqIkhuMDAAAAaHsI9ggLyUOSJRHsAQAAALQ9BHuEBW+w/3bNt6our7a5GgAAAADwH4I9wkL8GfFql9ZOVaVV2r1ut93lAAAAAIDfEOwRFgzD0BljzpDEcHwAAAAAbQvBHmGDYA8AAACgLSLYI2ycNvI0GRGG9m/fr4I9BXaXAwAAAAB+QbBH2IjpEKPOgzpLkna+vdPmagAAAADAP4I62M+cOVOGYfjcUlNTrfOmaWrmzJlKT09XTEyMhg4dqu3bt/tco6KiQtOmTVNSUpLi4uI0fvx47d27t7VfCoLEmZefKUn67xv/tbkSAAAAAPCPoA72ktS3b1/l5ORYt61bt1rn5syZo8cff1zz5s3Tpk2blJqaqpEjR6qoqMhqM336dC1fvlzLli3TunXrVFxcrHHjxqmmpsaOlwOb9RzfU5L0zXvfqLK40uZqAAAAAKDlgj7YR0ZGKjU11bp16tRJUm1v/ZNPPqnf/va3mjBhgjIyMrR48WKVlpbqhRdekCQVFBRowYIFeuyxxzRixAidc845WrJkibZu3apVq1bZ+bJgk6TeSepwegfVVNTo65Vf210OAAAAALRY0Af7nTt3Kj09Xd27d9d1112nb775RpK0a9cu5ebmatSoUVZbl8ulSy65ROvXr5ckbd68WVVVVT5t0tPTlZGRYbVpSEVFhQoLC31uCH2GYVi99gzHBwAAANAWBHWwHzhwoJ577jmtWLFCzz77rHJzczVkyBAdPHhQubm5kqSUlBSfx6SkpFjncnNzFRUVpQ4dOjTYpiGzZ8+W2+22bl26dPHjK4OdrGD/5n/lqfHYXA0AAAAAtExQB/uxY8fq6quvVr9+/TRixAi99dZbkqTFixdbbQzD8HmMaZp1jh2vMW3uu+8+FRQUWLc9e/Y081Ug2HS9sKuiO0Sr9ECp9m5kIUUAAAAAoS2og/3x4uLi1K9fP+3cudNaHf/4nve8vDyrFz81NVWVlZXKz89vsE1DXC6XEhISfG5oGyIiI3TmZbWr4+94fYfN1QAAAABAy4RUsK+oqFB2drbS0tLUvXt3paamauXKldb5yspKrV27VkOGDJEknXfeeXI6nT5tcnJytG3bNqsNwtOZ448E+zcI9gAAAABCW6TdBZzIjBkzdPnll6tr167Ky8vT73//exUWFurGG2+UYRiaPn26Zs2apR49eqhHjx6aNWuWYmNjNWnSJEmS2+3WlClTdM8996hjx45KTEzUjBkzrKH9CF9njD5DEc4IHdxxUAd2HFBSzyS7SwIAAACAZgnqYL9371799Kc/1YEDB9SpUycNGjRIGzduVLdu3SRJ9957r8rKynTrrbcqPz9fAwcO1Lvvvqv4+HjrGk888YQiIyM1ceJElZWVafjw4Vq0aJEcDoddLwtBwJXgUvdLu+vrd7/Wf//1X4I9AAAAgJAV1MF+2bJlJzxvGIZmzpypmTNnNtgmOjpac+fO1dy5c/1cHULdmePP1Nfvfq0db+zQkBlMzQAAAAAQmkJqjj3gTz0vr932bs9He1R6oNTmagAAAACgeQj2CFvurm6lnpMq02Pqv2/91+5yAAAAAKBZCPYIaz2vqO21375su82VAAAAAEDzEOwR1vrf0F+S9NWKr1Swp8DmagAAAACg6Qj2CGuJpyeq2yXdJFP6/LnP7S4HAAAAAJqMYI+wd84vzpEkbVm4RaZp2lwNAAAAADQNwR5hr/fVvRUVH6X8r/O1+8PddpcDAAAAAE1CsEfYi4qLUt9r+0qS/v33f9tcDQAAAAA0DcEekHTOz2uH4//n5f+ooqjC5moAAAAAoPEI9oCkzoM7q2PPjqoqrdL2f7D1HQAAAIDQQbAHJBmG4bOIHgAAAACECoI9cET/zP4yHIb2fLRHB3YcsLscAAAAAGiUSLsLAFpDdnZ2o9olD0nWvg/3acXsFep9R2/reFJSkrp27Rqo8gAAAACg2Qj2aNMOFNT2vN9www2Nat9LvXSdrtMXi7/Qzxf/XFWqkiTFxsQq+8tswj0AAACAoEOwR5tWVFokSbrzijt1fsb5J21vekztfnq3YvNj9fjwx9VhUAftytmlB/7+gA4cOECwBwAAABB0CPYIC507dlavrr0a1TZuXJy+eP4LFW8q1o+u/FGAKwMAAACAlmHxPOA4nQd1VkxijCoKK7T7o912lwMAAAAAJ0SwB44TERmh00efLkn6esXXMmtMmysCAAAAgIYR7IF6dLmgi1xul8rzy1X4RaHd5QAAAABAgwj2QD0cTodOH1Xba5+/Pl8RvFUAAAAABCnSCtCAbhd3U1R8lKoPVytDGXaXAwAAAAD1ItgDDXBEOXTaiNMkSRfrYnmqPTZXBAAAAAB1EeyBEzh16KmKiIlQkpK05409dpcDAAAAAHUQ7IETiIyOVOKFiZKkHU/vUEVRhc0VAQAAAIAvgj1wEu7z3Dqog6o4WKH1f1hvdzkAAAAA4INgD5yE4TC0SqskSev/uF6Fe9n+DgAAAEDwINgDjZCtbCWenajqsmqt/t/VdpcDAAAAABaCPdBIfe7qI0n6/LnPlfPvHJurAQAAAIBaBHugkTpkdFDGTzMkU1o5Y6VM07S7JAAAAAAg2ANNMXzWcDlcDu1avUs73thhdzkAAAAAQLAHmqL9qe016K5BkqQV01eoqqzK5ooAAAAAhDuCPdBEF//2YiV0TtDhbw/rozkf2V0OAAAAgDBHsAeaKKpdlEY9NkqS9NEjHyl/V77NFQEAAAAIZwR7oBn6/KSPug/rruryaq2YvsLucgAAAACEMYI90AyGYWjs3LGKiIzQjjd2aOfbO+0uCQAAAECYItgDzdSpTycNvHOgJCnrzixVV1TbXBEAAACAcESwB1rgkocuUbu0djr01SGte2Sd3eUAAAAACEORdhcAhIrs7Ox6j5857Ux9dv9n+uD3H0i9pIQeCfW2S0pKUteuXQNZIgAAAIAwRLAHTuJAwQFJ0g033NBgm2t1rXpX99ay65bpWT0rjzx12sTGxCr7y2zCPQAAAAC/ItgDJ1FUWiRJuvOKO3V+xvn1tqkurtbuv+5WWlmanrz4SSVelOhzflfOLj3w9wd04MABgj0AAAAAvyLYA43UuWNn9eraq8HzHSZ10L8X/Fv5H+Ur45IMJXSuf0g+AAAAAPgTi+cBfpJ+frpSzkqRWWNqy6It8tTUHY4PAAAAAP5GsAf8xDAM9bu+n5yxThXuKdRX73xld0kAAAAAwgDBHvCjaHe0+l7XV5L03zf/qwM7DthcEQAAAIC2jmAP+NkpPzpFnQd3lkzps2c/U/nhcrtLAgAAANCGEewBPzMMQ/0m9VN853hVFlVq8183y6wx7S4LAAAAQBtFsAcCwBHl0IBbBigyOlL5X+frwGqG5AMAAAAIDII9ECBxyXE6++dnS5IKPilQH/WxtyAAAAAAbRLBHgig1LNTdfqo0yVJV+pKHf7PYXsLAgAAANDmEOyBAOt5ZU/FdI9RlKL08bSPdfC/B+0uCQAAAEAbQrAHAizCEaG0q9P0g35Q5eFKPT/qeRX9UGR3WQAAAADaCII90AoiXBFaqqWK6xqngu8KtGT0EpXll9ldFgAAAIA2gGAPtJISlWjQvEFql9pOedvytGz8MlUWV9pdFgAAAIAQR7AHWlHsKbG6YcUNcrld2r1utxZetFCFewvtLgsAAABACCPYA60spX+KMt/NVFxynHK35OpvA/+mnM9y7C4LAAAAQIgi2AM2OOVHp+imj29Spz6dVPRDkRZetFBfvv6l3WUBAAAACEEEe8Am7U9tr1+s/4VOH3W6qkqr9NJVLynrriwW1QMAAADQJJF2FwCEk+zs7DrHev2ul6oTqvXdK9/p4yc/1mcLP1OvX/VS16u6KiLy6GdvSUlJ6tq1a2uWCwAAACAEEOyBVnCg4IAk6YYbbmiwzek6XWM0Rp0KOmnrI1u16pFVWqu12qEdqlGNYmNilf1lNuEeAAAAgA+CPdAKikqLJEl3XnGnzs84v8F2psdUwWcFOvTBIaWUpWiiJioiOkLmqab++uVftX//foI9AAAAAB8Ee6AVde7YWb269jpxo1OlytGV+mblN9q7Ya/KD5dLX0pTNVWrr1ytfVfu0xljztCpQ09VVFxUq9QNAAAAIHgR7IEgFBUXpV5X9lLP8T114MsD2v7eduVvy1fp3lJtmrdJm+ZtUoQzQolnJ6pD/w61t34dFOU+cdBnnj4AAADQ9hDsgSBmRBjq1KeTImsi9Ydtf1B3ddcZOkM91EPtq9rrwKYDOrDpgNX+gA7ooA7qkA4pX/k6pEMqUYnKjvxxRDuUvYN5+gAAAEBbQrAHQkBRaZEqVakxV4zR+RnnyzRNVR2qUtm3ZSr/vlzl35er6lCVko78aYin3KPFfRcrun20nAlORSVEyZnglDPe6Xv/mGPOBKeiO0VbK/TT6w8AAAAEF4I9EEJ85uh3k3TO0XOVxZUq2F2gkrwSlR4oVcn+2r8riypVVVolT5VHEYqQp9ij0uLSJj1vtap1UAe1X/t1OPKwHpj7gHpf0luJZyTK4XT47wW2gGmakiTDMGyuBAAAAGhdBHugjYhqF6VOfTqpU59O9Z5/+6O3Nee5OZp66VT16dJHnjKPaspr5Ck/8neZ5+jXxxyrKatRpCdSKUf+qFpa86s1WqM1inBGqOOZHZXUM0kJXRIUf0q8EjonKD4tXpExkXI4HXJEORThjJCnyqPq8uoGb1VlVXWPlVapuqxa1WW15733q8qqVFZYpsqSStVU1KimokaeCo8kyel2Kqp9lFwdXHIlutTu1HaKPyNeCWckKK5LnAzH0eDP6AMAAAC0BQR7IEwYkYaKVaz07uk6Z+A5J3/AEabHVNmhMhXlFOmb/3yjVatXadBpg1SdU62ashrt375f+7fvD2DlTVN5qFKVhypVrOI656pUpf3arzzlaZ/26XDUYb3y/ivqdX4va6oBAAAAEGrCKtj/5S9/0R/+8Afl5OSob9++evLJJ3XRRRfZXRYQ1IwIQ7FJsYpNitVO7dTrq1/X69+8LkOGEpSgTuqkjuqoeMUr4cifdmqnSEUqQhFyHPlToxpVN/CnSlUnPOc9f+zfVarST4f/VD1O6yEj0lCEM0IypZqyGtWU1tT+XVSjiv0VqsyrVOWBSjmrnEo/8keSVCm9PORlGRGG2qW1k7urW+1S2ykqLkqRsZG1f0fX/mfSNE2ZHlMyjwz7N2s/9CgsLFRZaZlkyvfcMX8bhlG7ZoH7yDoGbmftiIKOtbfUrqmMHAAAAECzhU2wf+mllzR9+nT95S9/0QUXXKBnnnlGY8eO1X/+8x/+hxpopKLSIknSnVfcqfMzzm/WNT7a+pHmvzHfL9fo2K2jzhpwVqMeY3pMlR4oVeHeQhV9X6Tvv/pe3335nZIcSTJrTBV9X6Si74uaVU9LVapSiV0S1f6U9mqX2k5xKXGKTYqVM84pZ6xTUXFRcsY6rfs+x4457ohyNHqNAdM0a6c8HJna4P3brDFP/uCTPEVEZIQcUQ5FuiLlcB35O8ohh8uhCEfTRkbU90FJY/6WJMNh1E4FiYwI6bUXaqpqVLr/yLoZR/4uzy9XZUmlqkqqVFlSqeryauv77ohyWNNgvFNhvD+PyJhIRUZHyhnjVGR07f1jvz72nBERut8zwJ9M01R1WbXKC8pVUVChisIKVZVW1b5fYp2176NYp6Ld0XLGOUP6vzcAQlfYBPvHH39cU6ZM0U033SRJevLJJ7VixQrNnz9fs2fPtrk6ILT4LOLXRLtydvntGk1hRBiKS45TXHKc0s5NU97WPM37cp6MGkNxipNbbp/RBs5j/kiSeeRPfV8P6jVIyR2TjzyRrL+t/7kzVbsOwJE1C2rKalRTUnszq0xFKUrFe4pVvKfu9IEmvUaHIUe0w7pFOCNkekyZNUdvx69J0NoMhyEj8uj35fgwbn3diM8XGv+kUkRUhCIiI2r/dh69HTv6wqrFo5Met85J1oiRY29GpCFXrEsx7WKOhu3jA3dUbeCOcESouqL2Qxbv2hJlB8tUsr9ExfuKVVlY6cdvRuNFOCMU4YqQw+WQEWXUhv0ja1RYv9vH/L5bxyNq32+Gw6j9+8jNlFk7skby/dkfue/9XkvHff+Pua4pU45IR+3768h1vV/XOdZAuKrx1MjhOMGinyfIZBGRtT/b2HaxSuiQIIfz6AcnDqej9vvTzA+iTnROku/rOvbvCKNxxwyj1a8hSWaNKU+N54R/Hzp4SCVFJbX/nfL+N8vj+7VqZB0zIo78nhlH/46Ni1VC+wSf3znvLcIR4XNfkjw1Hnmqj9yqPKooqtChHw6p+GCxqourVVVc5fN3oz70VO1/46LbR9d7c8Y5a2txHFOXw2jasSPvrWPPN+qDhEZ+1tDoDyX88dlFI76l1n8juA7Xac51PEf+O3PMf0uOv3/KwFN02vDTTn6xEBAWwb6yslKbN2/Wb37zG5/jo0aN0vr16+t9TEVFhSoqKqz7BQUFkqTCwsLAFeoHxcW1wSD7u2yVVjRt5XOvb3O+lSR9nfu13DvdXCMIrhEMNbS1a2z9eqskafzA8Trz1DObdY3t327X2x+/rR6n9VC3Pt2adY0vvvxCS95cojjFKVaxaqd2ilOcohVtfbAQpag6HzYc+ydCR8JSjaSSI7cmMGVa0xw8alngN478iVSkNQ3DR82RW2syJVUcuTXxexMsTJkqU5lKVaoSlahCFao88sc7bSXimD/e7/2xX3t/Jt7fJe9979fe6TOWqiO3ln3eBLQZpkxVHPlTrWqf90+UomTIkGqksoNl0kG7qwXQGBk/z1DS+Q1vFR0MvPnzZB94GGajPhIJbT/88INOOeUUffTRRxoyZIh1fNasWVq8eLF27NhR5zEzZ87Uww8/3JplAgAAAABQx549e9S5c+cGz4dFj73X8cOLvIta1ee+++7T3Xffbd33eDw6dOiQOnbsGNRzpwoLC9WlSxft2bNHCQkJdpcDBA3eG0BdvC+AunhfAPXjvWEP0zRVVFSk9PT0E7YLi2CflJQkh8Oh3Nxcn+N5eXlKSUmp9zEul0sul8vnWPv27QNVot8lJCTwhgPqwXsDqIv3BVAX7wugfrw3Wp/bffKpp2GxcXNUVJTOO+88rVy50uf4ypUrfYbmAwAAAAAQasKix16S7r77bmVmZmrAgAEaPHiw/vrXv2r37t365S9/aXdpAAAAAAA0W9gE+2uvvVYHDx7U//3f/yknJ0cZGRl6++231a1b81ayDlYul0sPPfRQnWkEQLjjvQHUxfsCqIv3BVA/3hvBLSxWxQcAAAAAoK0Kizn2AAAAAAC0VQR7AAAAAABCGMEeAAAAAIAQRrAHAAAAACCEEezbkL/85S/q3r27oqOjdd555+nDDz+0uySgVc2cOVOGYfjcUlNTrfOmaWrmzJlKT09XTEyMhg4dqu3bt9tYMeB/H3zwgS6//HKlp6fLMAy99tprPucb8z6oqKjQtGnTlJSUpLi4OI0fP1579+5txVcB+N/J3huTJ0+u82/IoEGDfNrw3kBbM3v2bJ1//vmKj49XcnKyrrzySu3YscOnDf9uhAaCfRvx0ksvafr06frtb3+rf//737rooos0duxY7d692+7SgFbVt29f5eTkWLetW7da5+bMmaPHH39c8+bN06ZNm5SamqqRI0eqqKjIxooB/yopKdFZZ52lefPm1Xu+Me+D6dOna/ny5Vq2bJnWrVun4uJijRs3TjU1Na31MgC/O9l7Q5LGjBnj82/I22+/7XOe9wbamrVr1+q2227Txo0btXLlSlVXV2vUqFEqKSmx2vDvRogw0Sb86Ec/Mn/5y1/6HOvVq5f5m9/8xqaKgNb30EMPmWeddVa95zwej5mammo+8sgj1rHy8nLT7XabTz/9dCtVCLQuSeby5cut+415Hxw+fNh0Op3msmXLrDbff/+9GRERYWZlZbVa7UAgHf/eME3TvPHGG80rrriiwcfw3kA4yMvLMyWZa9euNU2TfzdCCT32bUBlZaU2b96sUaNG+RwfNWqU1q9fb1NVgD127typ9PR0de/eXdddd52++eYbSdKuXbuUm5vr8z5xuVy65JJLeJ8gbDTmfbB582ZVVVX5tElPT1dGRgbvFbR577//vpKTk3XmmWdq6tSpysvLs87x3kA4KCgokCQlJiZK4t+NUEKwbwMOHDigmpoapaSk+BxPSUlRbm6uTVUBrW/gwIF67rnntGLFCj377LPKzc3VkCFDdPDgQeu9wPsE4awx74Pc3FxFRUWpQ4cODbYB2qKxY8dq6dKlWr16tR577DFt2rRJw4YNU0VFhSTeG2j7TNPU3XffrQsvvFAZGRmS+HcjlETaXQD8xzAMn/umadY5BrRlY8eOtb7u16+fBg8erNNPP12LFy+2FkDifQI0733AewVt3bXXXmt9nZGRoQEDBqhbt2566623NGHChAYfx3sDbcXtt9+uL774QuvWratzjn83gh899m1AUlKSHA5HnU/E8vLy6ny6BoSTuLg49evXTzt37rRWx+d9gnDWmPdBamqqKisrlZ+f32AbIBykpaWpW7du2rlzpyTeG2jbpk2bpjfeeENr1qxR586dreP8uxE6CPZtQFRUlM477zytXLnS5/jKlSs1ZMgQm6oC7FdRUaHs7GylpaWpe/fuSk1N9XmfVFZWau3atbxPEDYa8z4477zz5HQ6fdrk5ORo27ZtvFcQVg4ePKg9e/YoLS1NEu8NtE2maer222/Xq6++qtWrV6t79+4+5/l3I3QwFL+NuPvuu5WZmakBAwZo8ODB+utf/6rdu3frl7/8pd2lAa1mxowZuvzyy9W1a1fl5eXp97//vQoLC3XjjTfKMAxNnz5ds2bNUo8ePdSjRw/NmjVLsbGxmjRpkt2lA35TXFysr776yrq/a9cubdmyRYmJieratetJ3wdut1tTpkzRPffco44dOyoxMVEzZsxQv379NGLECLteFtBiJ3pvJCYmaubMmbr66quVlpamb7/9Vvfff7+SkpJ01VVXSeK9gbbptttu0wsvvKDXX39d8fHxVs+82+1WTExMo/7/ifdGkLBtPX743Z///GezW7duZlRUlHnuueda21QA4eLaa68109LSTKfTaaanp5sTJkwwt2/fbp33eDzmQw89ZKamppoul8u8+OKLza1bt9pYMeB/a9asMSXVud14442maTbufVBWVmbefvvtZmJiohkTE2OOGzfO3L17tw2vBvCfE703SktLzVGjRpmdOnUynU6n2bVrV/PGG2+s83vPewNtTX3vCUnmwoULrTb8uxEaDNM0zdb/OAEAAAAAAPgDc+wBAAAAAAhhBHsAAAAAAEIYwR4AAAAAgBBGsAcAAAAAIIQR7AEAAAAACGEEewAAAAAAQhjBHgAAAACAEEawBwAAAAAghBHsAQBoYwzD0GuvvWZ3GQAAoJUQ7AEACDKGYZzwNnnyZLtLDGrHfq/i4uLUo0cPTZ48WZs3b7a7NAAAAoJgDwBAkMnJybFuTz75pBISEnyO/elPf7K7RNuZpqnq6uoGzy9cuFA5OTnavn27/vznP6u4uFgDBw7Uc88914pVAgDQOgj2AAAEmdTUVOvmdrtlGIbPsRdeeEGnn366oqKi1LNnTz3//PMnvN7//d//KSUlRVu2bJEkrV+/XhdffLFiYmLUpUsX3XHHHSopKbHan3rqqZo1a5Z+8YtfKD4+Xl27dtVf//pX63xlZaVuv/12paWlKTo6Wqeeeqpmz57d4PNPnjxZV155pR5++GElJycrISFBt9xyiyorK602pmlqzpw5Ou200xQTE6OzzjpLr7zyinX+/fffl2EYWrFihQYMGCCXy6UPP/ywweds3769UlNTdeqpp2rUqFF65ZVXdP311+v2229Xfn6+1e6f//yn+vbtK5fLpVNPPVWPPfaYz3UqKip07733qkuXLnK5XOrRo4cWLFhwwu83AACtjWAPAEAIWb58ue68807dc8892rZtm2655Rb9/Oc/15o1a+q0NU1Td955pxYsWKB169bp7LPP1tatWzV69GhNmDBBX3zxhV566SWtW7dOt99+u89jH3vsMQ0YMED//ve/deutt+pXv/qVvvzyS0nSU089pTfeeEP/+Mc/tGPHDi1ZskSnnnrqCet+7733lJ2drTVr1ujFF1/U8uXL9fDDD1vn//d//1cLFy7U/PnztX37dt1111264YYbtHbtWp/r3HvvvZo9e7ays7PVv3//Jn3v7rrrLhUVFWnlypWSpM2bN2vixIm67rrrtHXrVs2cOVMPPPCAFi1aZD3mZz/7mZYtW6annnpK2dnZevrpp9WuXbsmPS8AAAFnAgCAoLVw4ULT7XZb94cMGWJOnTrVp81PfvIT88c//rF1X5L58ssvmzfccIPZq1cvc8+ePda5zMxM8+abb/Z5/IcffmhGRESYZWVlpmmaZrdu3cwbbrjBOu/xeMzk5GRz/vz5pmma5rRp08xhw4aZHo+nUa/hxhtvNBMTE82SkhLr2Pz588127dqZNTU1ZnFxsRkdHW2uX7/e53FTpkwxf/rTn5qmaZpr1qwxJZmvvfbaSZ9Pkrl8+fI6x8vKykxJ5qOPPmqapmlOmjTJHDlypE+b//mf/zH79OljmqZp7tixw5Rkrly5slGvEwAAu9BjDwBACMnOztYFF1zgc+yCCy5Qdna2z7G77rpLGzZs0IcffqjOnTtbxzdv3qxFixapXbt21m306NHyeDzatWuX1e7Y3nDvVIC8vDxJtUPrt2zZop49e+qOO+7Qu+++e9K6zzrrLMXGxlr3Bw8erOLiYu3Zs0f/+c9/VF5erpEjR/rU9dxzz+nrr7/2uc6AAQMa8V2qn2ma1uuRGv5e7ty5UzU1NdqyZYscDocuueSSZj8nAACtIdLuAgAAQNN4g6mXaZp1jo0cOVIvvviiVqxYoeuvv9467vF4dMstt+iOO+6oc92uXbtaXzudzjrP6fF4JEnnnnuudu3apXfeeUerVq3SxIkTNWLECJ858U15Ld7rvvXWWzrllFN8zrtcLp/7cXFxTX4OL++HH927d5dU//fNG/4lKSYmptnPBQBAayLYAwAQQnr37q1169bpZz/7mXVs/fr16t27t0+78ePH6/LLL9ekSZPkcDh03XXXSaoN5du3b9cZZ5zRojoSEhJ07bXX6tprr9U111yjMWPG6NChQ0pMTKy3/eeff66ysjIrLG/cuFHt2rVT586d1aFDB7lcLu3evTugvePeHQZGjBghSerTp4/WrVvn02b9+vU688wz5XA41K9fP3k8Hq1du9Z6DAAAwYhgDwBACPmf//kfTZw4Ueeee66GDx+uf/3rX3r11Ve1atWqOm2vuuoqPf/888rMzFRkZKSuueYa/frXv9agQYN02223aerUqYqLi1N2drZWrlypuXPnNqqGJ554QmlpaTr77LMVERGhl19+WampqWrfvn2Dj6msrNSUKVP0v//7v/ruu+/00EMP6fbbb1dERITi4+M1Y8YM3XXXXfJ4PLrwwgtVWFio/9/eHbJE1oZhAL53LIYJJkEEQcQsHJimIvgHRMSxyIxGg8U2gogniJgUUUyC0TTJblHwJygINhGEqVN0wwey4VvYXXZZznJd9Tzh5rSb95znvb29Tb1eT6vV+un31Ov18vLykn6/n4eHh5yfn6fb7eby8vIz59bWVhqNRsqyTLPZzN3dXU5OTnJ6eprkv9sBWq1W1tfXc3x8nKmpqTw/P+f19TXLy8s/nQkA/hTFHgAqZGFhIUdHRzk8PMzm5mbGx8dzcXGRubm5/51fWlrK+/t7VldXU6vVsri4mJubm2xvb2dmZiYfHx+ZmJhIs9n84Qz1ej0HBwd5fHzMwMBAGo1Grq+vU6t9f3XP/Px8JicnMzs7m36/n5WVlezu7n4+L8syw8PD2d/fz9PTU4aGhlIURTqdzg/n+tba2lqSZHBwMKOjo5mens79/X2KovicKYoiV1dX2dnZSVmWGRkZyd7eXtrt9ufM2dlZOp1ONjY28vb2lrGxsV/OBAB/ypePb38mAwD4zdrtdnq9Xrrd7t+OAgD/JFvxAQAAoMIUewAAAKgwn+IDAABAhTmxBwAAgApT7AEAAKDCFHsAAACoMMUeAAAAKkyxBwAAgApT7AEAAKDCFHsAAACoMMUeAAAAKuwrVc0BvyBJ4nYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1200x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_dist_tokens_per_doc = skills_df['doc'].apply(lambda d: d._.doc_len_alpha).to_frame()\n",
    "plt.figure(figsize=(12,8))\n",
    "sns.histplot(df_dist_tokens_per_doc[\"doc\"], kde=True, stat=\"count\", color='purple', alpha=0.4, bins=50)\n",
    "plt.xlabel(\"Tokens per Doc\")\n",
    "plt.ylabel(\"Number of Docs\")\n",
    "plt.title(\"Distribution of Tokens per Doc\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "24356613",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    10249.000000\n",
       "mean         8.855596\n",
       "std         10.781777\n",
       "min          0.000000\n",
       "25%          4.000000\n",
       "50%          6.000000\n",
       "75%         11.000000\n",
       "max        213.000000\n",
       "Name: doc, dtype: float64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_dist_tokens_per_doc[\"doc\"].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f565428c",
   "metadata": {},
   "source": [
    "Let's calculate the number of Docs where the size of tokens is empty and also check how many per class\n",
    "\n",
    "*Notes*\n",
    "- 2% of Documents ( around 200~) are empty, we will remove this at cleaning time. \n",
    "- Observing a sample of this documents, we can see that much of them were removed because we have filtered by alpha and they contain hyphens, or only numbers. \n",
    "- Some of this examples with hyphens are skills like: `EDV-Anwenderkenntnisse` --> IT user skills, this might add value to our classifier but for now let's leave it out. ( We can revisit later on )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "a8c56529",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ratio of empty docs after cleaning: 2%\n"
     ]
    }
   ],
   "source": [
    "ratio_empty = df_dist_tokens_per_doc[df_dist_tokens_per_doc['doc'] < 1].shape[0]/skills_df.shape[0] * 100\n",
    "print(f\"Ratio of empty docs after cleaning: {round(ratio_empty)}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "ee25264b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>text</th>\n",
       "      <th>label</th>\n",
       "      <th>doc</th>\n",
       "      <th>tokens</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>Pkw-Führerschein</td>\n",
       "      <td>soft</td>\n",
       "      <td>(Pkw-Führerschein)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>420</th>\n",
       "      <td>IT-Affinität.</td>\n",
       "      <td>tech</td>\n",
       "      <td>(IT-Affinität, .)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>424</th>\n",
       "      <td>Www.bode-chemie.de.</td>\n",
       "      <td>none</td>\n",
       "      <td>(Www.bode-chemie.de, .)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>427</th>\n",
       "      <td>KFZ-Führerschein</td>\n",
       "      <td>tech</td>\n",
       "      <td>(KFZ-Führerschein)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>527</th>\n",
       "      <td>Hands-on-Mentalität.</td>\n",
       "      <td>soft</td>\n",
       "      <td>(Hands-on-Mentalität, .)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>528</th>\n",
       "      <td>Junior-Vertriebsmitarbeiter/-in</td>\n",
       "      <td>none</td>\n",
       "      <td>(Junior-Vertriebsmitarbeiter/-in)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>604</th>\n",
       "      <td>EDV-Anwenderkenntnisse</td>\n",
       "      <td>tech</td>\n",
       "      <td>(EDV-Anwenderkenntnisse)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>616</th>\n",
       "      <td>MS-Office-Know-how.</td>\n",
       "      <td>tech</td>\n",
       "      <td>(MS-Office-Know-how, .)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>699</th>\n",
       "      <td>VPN-Bausteine Internet-Bausteine Telefonie-Bau...</td>\n",
       "      <td>none</td>\n",
       "      <td>(VPN-Bausteine, Internet-Bausteine, Telefonie-...</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>876</th>\n",
       "      <td>+49 (0) 49 54 - 89 3 88-100</td>\n",
       "      <td>none</td>\n",
       "      <td>(+49, (, 0, ), 49, 54, -, 89, 3, 88, -, 100)</td>\n",
       "      <td>[]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                  text label  \\\n",
       "39                                    Pkw-Führerschein  soft   \n",
       "420                                      IT-Affinität.  tech   \n",
       "424                                Www.bode-chemie.de.  none   \n",
       "427                                   KFZ-Führerschein  tech   \n",
       "527                               Hands-on-Mentalität.  soft   \n",
       "528                    Junior-Vertriebsmitarbeiter/-in  none   \n",
       "604                             EDV-Anwenderkenntnisse  tech   \n",
       "616                                MS-Office-Know-how.  tech   \n",
       "699  VPN-Bausteine Internet-Bausteine Telefonie-Bau...  none   \n",
       "876                        +49 (0) 49 54 - 89 3 88-100  none   \n",
       "\n",
       "                                                   doc tokens  \n",
       "39                                  (Pkw-Führerschein)     []  \n",
       "420                                  (IT-Affinität, .)     []  \n",
       "424                            (Www.bode-chemie.de, .)     []  \n",
       "427                                 (KFZ-Führerschein)     []  \n",
       "527                           (Hands-on-Mentalität, .)     []  \n",
       "528                  (Junior-Vertriebsmitarbeiter/-in)     []  \n",
       "604                           (EDV-Anwenderkenntnisse)     []  \n",
       "616                            (MS-Office-Know-how, .)     []  \n",
       "699  (VPN-Bausteine, Internet-Bausteine, Telefonie-...     []  \n",
       "876       (+49, (, 0, ), 49, 54, -, 89, 3, 88, -, 100)     []  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "skills_df.iloc[df_dist_tokens_per_doc[df_dist_tokens_per_doc['doc'] < 1].index].head(10)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}