{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "1b68c7d8",
   "metadata": {},
   "source": [
    "# Kryssvalidering\n",
    "\n",
    "Vi vil brukte et datasett som inneholder informasjon om 4000 epler og hvor hvert eple er karakterisert som enten bra eller dårlig. Vi vil lage en modell basert på beslutningstrær som kan klassifisere et eple som bra eller dårlig basert på 7 forskjellige parametere. Hovedfokuset i denne notebooken vil derimot være mer på hvordan vi kan bruke de tilgjengelig dataene til å validere modellen og unngå overtrening.\n",
    "\n",
    "Først leser vi inn datasettet, fjerner ugyldige verdier, gjør om *god* og *dårlig* til 0 og 1 og plukker ut variablene/predikatorene samt target (god eller dårlig) i egne numpy-arrays. Dette bør være kjent stoff fra tidligere så vil ikke kommentere så mye på det som blir gjort. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "28b5d6cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "#%matplotlib notebook\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from scipy.special import expit\n",
    "from scipy.stats import expon\n",
    "import math\n",
    "from sklearn import tree\n",
    "from sklearn.metrics import accuracy_score, \\\n",
    "                            precision_score, recall_score, f1_score, mean_squared_error, \\\n",
    "                            mean_absolute_error, ConfusionMatrixDisplay, confusion_matrix\n",
    "from IPython.display import Image\n",
    "from IPython.core.display import HTML "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "33f4c11c",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"./apple_quality.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "94888b49",
   "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>A_id</th>\n",
       "      <th>Size</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Sweetness</th>\n",
       "      <th>Crunchiness</th>\n",
       "      <th>Juiciness</th>\n",
       "      <th>Ripeness</th>\n",
       "      <th>Acidity</th>\n",
       "      <th>Quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.970049</td>\n",
       "      <td>-2.512336</td>\n",
       "      <td>5.346330</td>\n",
       "      <td>-1.012009</td>\n",
       "      <td>1.844900</td>\n",
       "      <td>0.329840</td>\n",
       "      <td>-0.491590483</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.195217</td>\n",
       "      <td>-2.839257</td>\n",
       "      <td>3.664059</td>\n",
       "      <td>1.588232</td>\n",
       "      <td>0.853286</td>\n",
       "      <td>0.867530</td>\n",
       "      <td>-0.722809367</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>-0.292024</td>\n",
       "      <td>-1.351282</td>\n",
       "      <td>-1.738429</td>\n",
       "      <td>-0.342616</td>\n",
       "      <td>2.838636</td>\n",
       "      <td>-0.038033</td>\n",
       "      <td>2.621636473</td>\n",
       "      <td>bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>-0.657196</td>\n",
       "      <td>-2.271627</td>\n",
       "      <td>1.324874</td>\n",
       "      <td>-0.097875</td>\n",
       "      <td>3.637970</td>\n",
       "      <td>-3.413761</td>\n",
       "      <td>0.790723217</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.364217</td>\n",
       "      <td>-1.296612</td>\n",
       "      <td>-0.384658</td>\n",
       "      <td>-0.553006</td>\n",
       "      <td>3.030874</td>\n",
       "      <td>-1.303849</td>\n",
       "      <td>0.501984036</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   A_id      Size    Weight  Sweetness  Crunchiness  Juiciness  Ripeness  \\\n",
       "0   0.0 -3.970049 -2.512336   5.346330    -1.012009   1.844900  0.329840   \n",
       "1   1.0 -1.195217 -2.839257   3.664059     1.588232   0.853286  0.867530   \n",
       "2   2.0 -0.292024 -1.351282  -1.738429    -0.342616   2.838636 -0.038033   \n",
       "3   3.0 -0.657196 -2.271627   1.324874    -0.097875   3.637970 -3.413761   \n",
       "4   4.0  1.364217 -1.296612  -0.384658    -0.553006   3.030874 -1.303849   \n",
       "\n",
       "        Acidity Quality  \n",
       "0  -0.491590483    good  \n",
       "1  -0.722809367    good  \n",
       "2   2.621636473     bad  \n",
       "3   0.790723217    good  \n",
       "4   0.501984036    good  "
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "075763ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "e13ff817",
   "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>A_id</th>\n",
       "      <th>Size</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Sweetness</th>\n",
       "      <th>Crunchiness</th>\n",
       "      <th>Juiciness</th>\n",
       "      <th>Ripeness</th>\n",
       "      <th>Acidity</th>\n",
       "      <th>Quality</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.970049</td>\n",
       "      <td>-2.512336</td>\n",
       "      <td>5.346330</td>\n",
       "      <td>-1.012009</td>\n",
       "      <td>1.844900</td>\n",
       "      <td>0.329840</td>\n",
       "      <td>-0.491590483</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.195217</td>\n",
       "      <td>-2.839257</td>\n",
       "      <td>3.664059</td>\n",
       "      <td>1.588232</td>\n",
       "      <td>0.853286</td>\n",
       "      <td>0.867530</td>\n",
       "      <td>-0.722809367</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>-0.292024</td>\n",
       "      <td>-1.351282</td>\n",
       "      <td>-1.738429</td>\n",
       "      <td>-0.342616</td>\n",
       "      <td>2.838636</td>\n",
       "      <td>-0.038033</td>\n",
       "      <td>2.621636473</td>\n",
       "      <td>bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>-0.657196</td>\n",
       "      <td>-2.271627</td>\n",
       "      <td>1.324874</td>\n",
       "      <td>-0.097875</td>\n",
       "      <td>3.637970</td>\n",
       "      <td>-3.413761</td>\n",
       "      <td>0.790723217</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.364217</td>\n",
       "      <td>-1.296612</td>\n",
       "      <td>-0.384658</td>\n",
       "      <td>-0.553006</td>\n",
       "      <td>3.030874</td>\n",
       "      <td>-1.303849</td>\n",
       "      <td>0.501984036</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3995</th>\n",
       "      <td>3995.0</td>\n",
       "      <td>0.059386</td>\n",
       "      <td>-1.067408</td>\n",
       "      <td>-3.714549</td>\n",
       "      <td>0.473052</td>\n",
       "      <td>1.697986</td>\n",
       "      <td>2.244055</td>\n",
       "      <td>0.137784369</td>\n",
       "      <td>bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3996</th>\n",
       "      <td>3996.0</td>\n",
       "      <td>-0.293118</td>\n",
       "      <td>1.949253</td>\n",
       "      <td>-0.204020</td>\n",
       "      <td>-0.640196</td>\n",
       "      <td>0.024523</td>\n",
       "      <td>-1.087900</td>\n",
       "      <td>1.854235285</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3997</th>\n",
       "      <td>3997.0</td>\n",
       "      <td>-2.634515</td>\n",
       "      <td>-2.138247</td>\n",
       "      <td>-2.440461</td>\n",
       "      <td>0.657223</td>\n",
       "      <td>2.199709</td>\n",
       "      <td>4.763859</td>\n",
       "      <td>-1.334611391</td>\n",
       "      <td>bad</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3998</th>\n",
       "      <td>3998.0</td>\n",
       "      <td>-4.008004</td>\n",
       "      <td>-1.779337</td>\n",
       "      <td>2.366397</td>\n",
       "      <td>-0.200329</td>\n",
       "      <td>2.161435</td>\n",
       "      <td>0.214488</td>\n",
       "      <td>-2.229719806</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3999</th>\n",
       "      <td>3999.0</td>\n",
       "      <td>0.278540</td>\n",
       "      <td>-1.715505</td>\n",
       "      <td>0.121217</td>\n",
       "      <td>-1.154075</td>\n",
       "      <td>1.266677</td>\n",
       "      <td>-0.776571</td>\n",
       "      <td>1.599796456</td>\n",
       "      <td>good</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4000 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        A_id      Size    Weight  Sweetness  Crunchiness  Juiciness  Ripeness  \\\n",
       "0        0.0 -3.970049 -2.512336   5.346330    -1.012009   1.844900  0.329840   \n",
       "1        1.0 -1.195217 -2.839257   3.664059     1.588232   0.853286  0.867530   \n",
       "2        2.0 -0.292024 -1.351282  -1.738429    -0.342616   2.838636 -0.038033   \n",
       "3        3.0 -0.657196 -2.271627   1.324874    -0.097875   3.637970 -3.413761   \n",
       "4        4.0  1.364217 -1.296612  -0.384658    -0.553006   3.030874 -1.303849   \n",
       "...      ...       ...       ...        ...          ...        ...       ...   \n",
       "3995  3995.0  0.059386 -1.067408  -3.714549     0.473052   1.697986  2.244055   \n",
       "3996  3996.0 -0.293118  1.949253  -0.204020    -0.640196   0.024523 -1.087900   \n",
       "3997  3997.0 -2.634515 -2.138247  -2.440461     0.657223   2.199709  4.763859   \n",
       "3998  3998.0 -4.008004 -1.779337   2.366397    -0.200329   2.161435  0.214488   \n",
       "3999  3999.0  0.278540 -1.715505   0.121217    -1.154075   1.266677 -0.776571   \n",
       "\n",
       "           Acidity Quality  \n",
       "0     -0.491590483    good  \n",
       "1     -0.722809367    good  \n",
       "2      2.621636473     bad  \n",
       "3      0.790723217    good  \n",
       "4      0.501984036    good  \n",
       "...            ...     ...  \n",
       "3995   0.137784369     bad  \n",
       "3996   1.854235285    good  \n",
       "3997  -1.334611391     bad  \n",
       "3998  -2.229719806    good  \n",
       "3999   1.599796456    good  \n",
       "\n",
       "[4000 rows x 9 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "61cdd501",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.drop(\"A_id\",axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "f913899c",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['quality_num'] = df['Quality'].apply(lambda x: 1.0 if x == \"good\" else 0.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "f033f47f",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictors = list(df.columns)\n",
    "predictors.remove('Quality')\n",
    "predictors.remove('quality_num')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "474fd4ed",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = df[predictors].values\n",
    "Y = df.quality_num.to_numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a644a141",
   "metadata": {},
   "source": [
    "Først skal vi se på tilfellet hvor vi delere dataene inn i et treningssett og et testsett. Vi bruker allerede eksisterende funskjon i sklearn for dette. Den returnerer X- og Y-vektorer med $80\\%$ av dataene for trening og $20\\%$ til testing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "ed216adb",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.2, random_state=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "86806b0b",
   "metadata": {},
   "source": [
    "Sjekk dimensjonen til treningssettet og testsettet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "bfe7c111",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3200, 7)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "871aad3b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800, 7)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3dfefa88",
   "metadata": {},
   "source": [
    "## Trenings- og testsett\n",
    "\n",
    "Vi lager en modell hvor vi lar dybden til treet vårt være forskjellig (alle verider mellom 1 og 21) siden dette er en parameter som, om vi lar den bli veldig stor, fort fører til overtrening. For å validere metoden bruker vi *accuracy*, som for en klassifiseringsmodell er definert ved"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7d4b3ef3",
   "metadata": {},
   "source": [
    "$$\\texttt{accuracy}(y, \\hat{y}) = \\frac{1}{n_\\text{samples}} \\sum_{i=0}^{n_\\text{samples}-1} 1(\\hat{y}_i = y_i)$$\n",
    "\n",
    "hvor\n",
    "\n",
    "$$ 1(\\hat{y}_i = y_i) = \\begin{cases}\n",
    "1 & \\text{hvis}\\, \\hat{y}_i = y_i, \\\\\n",
    "0 & \\text{hvis}\\, \\hat{y}_i \\neq y_i.\n",
    "\\end{cases} $$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e17a88cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Definerer numpy-arrays til å lagre informasjon om hvert enkelt tre (med ulik dybde)\n",
    "treedepth = np.zeros((20,))\n",
    "\n",
    "\n",
    "mean_scores_test = np.zeros((20,))\n",
    "mean_scores_train = np.zeros((20,))\n",
    "\n",
    "i = 0\n",
    "for depth in range(1,21):\n",
    "    clf = tree.DecisionTreeClassifier(max_depth=depth)\n",
    "    clf = clf.fit(X_train,y_train)\n",
    "    \n",
    "    # Bruk modellen på test-data og trenings-data\n",
    "    y_pred_test = clf.predict(X_test)\n",
    "    y_pred_train = clf.predict(X_train)\n",
    "    \n",
    "    # Hent score (basert på trening og test) etter formelen over \n",
    "    # for denne modellen og lagre til senere for plotting\n",
    "    mean_scores_train[i] = clf.score(X_train, y_train)\n",
    "    mean_scores_test[i] = clf.score(X_test, y_test)\n",
    "    \n",
    "    # For å definere x-aksen\n",
    "    treedepth[i] = (depth)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "021c3c91",
   "metadata": {},
   "source": [
    "Så kan vi plotte score som en funksjon av tredybde for de to tilfellene hvor vi \n",
    "\n",
    "1. har validert modellen på treningssettet \n",
    "2. har validert modellen på testsettet."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "c813fe55",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fc0609724d0>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "\n",
    "plt.plot(treedepth,mean_scores_train,color='blue', marker='^', linestyle='dashed',label='Treningssett')\n",
    "plt.plot(treedepth,mean_scores_test,color='red', marker='^', linestyle='dashed',label='Testsett')\n",
    "\n",
    "\n",
    "ax.set_title('')\n",
    "ax.set_xlabel('Tredybde')\n",
    "ax.set_ylabel('accuracy')\n",
    "ax.set_xticks(treedepth) \n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f5adfcb3",
   "metadata": {},
   "source": [
    "Her ser vi klare tegn til overtrenining siden accuracy for modellen øker og øker når vi bruker treningssettet, mens accuracy ser ut til å flate ut på en tredybde på rundt 9 når vi tester den på testsettet (dvs. at modellen egentlig ikke blir noe bedre på å klassifisere data den ikke har sett etter en dybde på ca. 9). Derimot blir accuarcy nesten 1 dersom vi lar treet vokse helt til en dybde på 21. Det at disse to kurvene er såpass forskjellig er et klart tegn på overtrening og at modellen overoptimaliserer for tilfeldige fluktuasjoner i treningssettet som ikke nødvendigvis er der i testsettet. Altså er modellen vår (med dybde > 9) dårlig til å generalisere og unødvendig komplisert.\n",
    "\n",
    "En bedre måte å dele opp datasettet vårt i trening og validering er å bruke såkalt kryssvalidering. La oss først se på k-fold kryssvalidering hvor datasettet deles opp på følgende måte:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "d216aa14",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<img src=\"https://scikit-learn.org/stable/_images/grid_search_cross_validation.png\"/>"
      ],
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(url= \"https://scikit-learn.org/stable/_images/grid_search_cross_validation.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5cd52b07",
   "metadata": {},
   "source": [
    "Her har vi valgt en k-fold kryssvalidering med 5 splits. Vi holder forstatt en del av dataene utenfor, og som brukes som testsett og ikke på noe punkt inngår i treningen av modellen. De resterende dataene brukes i kryssvalideringen hvor vi deler dataene opp i 5 blokker med like mange datapunkter. Vi tilpasser så modellen vår ved å bruke første \"fold\" til validering og de resterende til trening. Fold-1 brukes altså til å validere modellen oppnådd ved å trene på de fire andre \"foldene\", og basert på validerings-folden kan vi finne MSE, accuracy og andre parametere. Vi gjentar så prosessen (*Split2*) hvor vi endrer hvilken fold vi bruker til validering. Når vi har gjort dette for alle 5 splittene sitter vi altså igjen med 5 målinger på modellen vår, trent på 5 ulike versjoner av treningssettet. \n",
    "\n",
    "Alt dette er enkelt tilgjengelig via- følgende funksjon:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "0d8fe013",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "53df069c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Numpy-arrays for a lagre resultatene for hver tredybde\n",
    "mean_scores = np.zeros((20,))\n",
    "std_scores = np.zeros((20,))\n",
    "\n",
    "i = 0\n",
    "# Loop over de ulike tredybpdene \n",
    "for depth in range(1,21):\n",
    "    clf = tree.DecisionTreeClassifier(max_depth=depth)\n",
    "    # Denne gjør selve kryssvalideringen og returnerer \n",
    "    # score/accuracy bergenet på de ulike valdierings-delene/foldene \n",
    "    # av datasettet (i dette tilfeller altså 5 verdier)\n",
    "    scores = cross_val_score(clf, X_train, y_train, cv=5)\n",
    "    # Vi finner gjennomsnitt og standardavvik til de 5 målingene\n",
    "    mean_scores[i] = (scores.mean())\n",
    "    std_scores[i] = (scores.std())\n",
    "    \n",
    "    treedepth[i] = (depth)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "739d6005",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fc0548bc190>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkAAAAGwCAYAAABB4NqyAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAACs6ElEQVR4nOzdd3xUVfr48c9Meg8hIQWS0Amh9yYIiBBQigoBRVwsuLgqYtdVd61ff7quBQtiRZSlKAKKEHqXDqFIb0kggZDe28z9/XGZIZPMJJNkkknI83695kVy55Yzk5D7zDnPeY5GURQFIYQQQohGRGvvBgghhBBC1DUJgIQQQgjR6EgAJIQQQohGRwIgIYQQQjQ6EgAJIYQQotGRAEgIIYQQjY4EQEIIIYRodBzt3YD6SK/Xk5iYiJeXFxqNxt7NEUIIIYQVFEUhOzubkJAQtNqK+3gkADIjMTGR0NBQezdDCCGEENWQkJBAixYtKtxHAiAzvLy8APUN9Pb2tnNrhBBCCGGNrKwsQkNDjffxikgAZIZh2Mvb21sCICGEEKKBsSZ9RZKghRBCCNHoSAAkhBBCiEZHAiAhhBBCNDoSAAkhhBCi0ZEASAghhBCNjgRAQgghhGh0JAASQgghRKMjAZAQQgghGh0JgIQQQgjR6EgAJIQQQohGRwIgIYQQQjQ6EgAJIYQQotGRAEgIIYQQjY6sBi+EEKJRys/P5+rVq7i5ueHu7o67uzsODg72bpaoIxIACSFqlU6nIysrCycnJzw9Pe3dHNHI5eXlcfLkSf766y8uXLiAoigmzzs5OeHm5oaHhweenp54eHgYg6PSD8N2FxcXNBqNnV6NqAkJgIQQ1aYoCvn5+WRmZpo8srKySEtLIzMzk7y8PAAcHBwYO3Ys3bp1s3OrRWOTk5NjDHri4uJQFAWNRlMu+AEoLi6muLiYrKwsALRaNVNEURSz+2s0GlxdXXF3d8fT0xNPT09jAOXu7k5gYCBhYWESJNVDEgAJISwqKSkhKyurXICTmZlJeno62dnZlJSUmByj1WrN3ix0Oh0rVqwgISGBqKgoHB3lz4+oPdnZ2Zw4cYJjx46RkJAAYBL0mAtmzNHr9RU+b/gQkJ+fT2pqKhqNxhjsGI718vKie/fudOvWjaZNm1b3JQkb0yjW/hY0IllZWfj4+JCZmYm3t7e9myNErdPpdMTGxpKSkmLSe5Ofn2+yn+HTcGU3hYpoNBqCgoKYPHkyPj4+NWq3EKVlZmYag57Lly8DWOzpqWuGdoSEhNCjRw86deqEm5ubvZtVJwxBYl5eHrm5ucZ/vb29ad++vU2vVZX7twRAZkgAJBqbrVu3smXLFou9N7am0WhwdnZm4sSJtG3btlavJW5u6enpxqAnKSmp3gQ8ldFqtbRv357u3bvTtm3bBpV8rdfrzQY0hq/z8vLIyckhJyeHvLw8CgoKzP5MXF1defHFF23atqrcv6UPWohGLjMzk+3btwM169mpCkVRKCoqYuHChQwdOpQhQ4ZIjoSwWlpaGsePH+fYsWNcvXq1WkNb9qbX6zl9+jQnT57E1dWVrl270q1bN4KDg+3+fyE7O5vLly9z9epVYxCTnZ1tDHAKCwvNvs+V5UuVZe+flQRAQjRy69evr7PApzTDH78tW7aQkJDAPffc0yCGBAx/3Ct66PV6q/YrvX9JSQnFxcVm/61om7mH4fmSkhJ0Oh2Ojo4EBgYSHBxMYGAgQUFBNGvWDCcnJ3u/nVZLSUkxBj3Xrl1rkEFPWYb/dwUFBezfv5+9e/fStGlTevToQZcuXepkBCI3N5fExETj49KlS8aJC1UNaOzxd6QmZAjMDBkCE41FXFwc8+fPt3cz0Gg0eHl5MWXKFIKDg+3dHKPCwkL27NnDrl27LH7qrW2G3oCyvQLVGarUarXGm5RGo8HX15fg4GCCgoKMgZGXl5ddeyD0er0x8T4jI4PU1FSOHz9uTDBuDLcsw+ts1aoV3bt3JyIiAmdn5xqft6CggKSkJC5fvmwMdrKzs43XhLoNJl1cXHjppZdsek4ZAhNCVEqv17N69ep6cVNRFIXs7Gy++eYb7rjjDnr27GnX9hQXF7N37162b99OUVGRXd8fW/ZylP6ErigK6enpZGRkcOLECeP5XVxcCAoKMj4CAwMJCAiw2ay9oqIiMjIyys0qNCTe5+bmmrzW0kGbvX9P64rhdV68eJELFy7g6OhIp06d6NatGy1btrQqQC0qKuLKlStcvnyZpKQkEhISyMjIAMwHO43lvS1NAiAhGqlDhw6RnJxs72YYGXo0fv/9dxISEhgzZkydD9GUlJRw4MABtm7dWm4G3M2q7I2vsLCQuLg4EhISTHqL/Pz8CAkJMfYUBQYGlitsqSgKubm5xqCmdKCTlpZGVlYWhYWFJsdUlnjf0IZVbMnwnpSUlHD06FEOHz6Ml5cX3bp1o1u3bvj7+xufv3r1qkmwk5aWZqx3VPpcZb9uzCQAEqIRys/PZ8OGDfZuhkWHDx8mMTGRKVOm0KRJk1q/nk6n49ChQ2zdupWcnJxav15DULa3KDU1lbS0NI4dO2a8gbq5uREUFASoicnZ2dkmxxlq4lQUxDTmAKcqDO9TdnY2O3fuZMeOHQQFBaEoCteuXTM+X7rHDCTYqYgEQEI0Qlu3bi33Sbw+URSFlJQUvvzyS+655x6b1wox0Ov1HD58mC1bthgr/wrLyt5M8/PzuXDhQoX7yw3Y9gzv6ZUrV8o9JwGl9SQAEqKRuXbtGnv37q33Nya9Xk9RURGLFi1i8ODBDB061DgrxRbn/uuvv9i0aZMxL0II0bhIACREI6IoCmvWrKkXic9VsX37di5dusTEiRNxd3ev9nkUReHEiRNs2rSJ1NRUG7ZQCNHQSAAkRCNy6tSpCocs6rOLFy8yd+5cJk+eTIsWLap0rKIonD59mk2bNpGcnGz3QnNCCPuTAEiIRqKkpISYmJgG1/tjYJhh9P333xMVFUXv3r0rDWQUReH8+fNs3LjRuEyCYXtjoNHoCQuLw8srh+xsT+Ljw1EU2wwjCtHQSQAkRCOxa9cuMjMz7d2MGjEk1a5evZqEhATGjh1rcar8xYsX2bhxI5cuXWp0gQ9ARMRxoqJi8PG5kdydmelNTEwUJ09G2rFlQtQPEgAJ0QhkZ2ezbds2ezfDpgyLX06ZMoWmTZsatyckJLBp0yYuXrzYKAMfUIOf6Oil5bZ7e2cRHb2UpUujJQgSjZ4EQEI0AuvXr0en09m7GTZlqE0zb9487r77bry9vdm0aRPnzp1rtIEPqMNeUVEx178u+xwoCkRFxXDqVIQMh4lGze6//V988QWtWrXC1dWVXr16GVeltmThwoV069YNd3d3goODefDBB8vN5li2bBmRkZG4uLgQGRnJ8uXLa/MlCFGvXbp0iaNHj96UwYCiKBQXF7NkyRK+/vprY4J3Q3mtGo2e8PALdO58lPDwC2g0Na3hoqdTp6P4+GSVC35uXBN8fLIIC4ur4bWEaNjs2gO0ZMkSZs+ezRdffMGgQYOYN28eo0eP5vjx44SFhZXbf8eOHTzwwAN89NFHjB07lsuXLzNz5kweeeQRY5Cza9cuJk+ezFtvvcVdd93F8uXLiY6OZseOHfTr16+uX6IQdqUoCn/88UeDTXyuKlsXgavNJGJb5Og0bXqNNm3OERiYTLNmV2nWLBln52KrjvXyatwVryVBXNh1Nfh+/frRs2dP5s6da9zWsWNHJkyYwLvvvltu/w8++IC5c+dy7tw547ZPP/2U999/n4SEBAAmT55MVlYWa9asMe4TFRVFkyZNWLRokVXtktXgxc3i0KFD/Pbbb/ZuRoNUm0nEpXN0SvfUGP4al87RcXAowd8/hcDAqzRrdpWDB3uRlqbmPPXtu4fRo9eYnLukRIujY+WB4MqV44iNte+is/YiCeIVq+3g0HD+Jk0KGD9+JjAYcLDJuRvEavBFRUUcOHCAl156yWT7yJEj+fPPP80eM3DgQF555RVWr17N6NGjSU5O5pdffuGOO+4w7rNr1y6efvppk+NGjRrFxx9/bLEthYWFJssCSEl8cTMoLCxk/fr19m5Gg1SbScTW5OiMH7+SyMi/CAxMpmnTVBwcbgQ0ycnNjAHQpUvNOXmyA8nJgVy9qj7S032ZNWsO3t6Wh8EAxo//jYiIk+zYMZhLl0Kr9VpqU23dhCVBvGK1HRyWP/8SoAXwCXB3jc9fFXYLgFJSUtDpdAQGBppsDwwMNLu+CagB0MKFC5k8eTIFBQWUlJQwbtw4Pv30U+M+V65cqdI5Ad59913eeOONGrwaIeqfrVu3UlBQYO9mNDhVTyJWcHDQ4eRUbPLIyfEkN1ddLd3DI5t27c7g5FRMs2ZXTW4u5a8Prq6FdOnyl3Fbfr4rV68GmgQ/AImJLViy5N5y54iJiSI6eimKYr6HKSGhBaGhl+jQ4TQdOpxm27bBbN58WxXfqdpTWzdhSRCvWG0Hh5bOD5eBicAv1GUQZPdZYGULmSmKYrG42fHjx5k1axb/+te/GDVqFElJSTz//PPMnDmTb7/9tlrnBHj55Zd55plnjN9nZWURGlr/PhEJYa3U1FT27NnTKPJ+bC0sLK7SAMXHJ4u+ffcwbNhmnJyK0WrLv88xMaPYs2cAAE2apDN+fNWGIo8e7cSRI924ejWQ7GxvwPrq1SdPRrJ0aXS5ICIr60YQ4eeXwqBBO+nW7TBnztxYbNbJqYjiYkfsNUemZjdhBWfnQry8csjPdyMvzwOAgIBkbrllOwEByVb9bAcM+JO//upMZqY31X0fGlqOkW2CQz0ODnq0Wp3Jv7m5HiiKxuL5QUH9/Z4NjMdWw2GVsVsA5O/vj4ODQ7memeTk5HI9OAbvvvsugwYN4vnnnwega9eueHh4MHjwYN5++22Cg4MJCgqq0jkBXFxccHFxqeErEqL+KJ0DV5m6Gu9vKDcCL69sq/ZzdS3AxaXIZJter6GoyJniYid0uht/xHNzPTh9uh3FxU64uBTStu25sqcr58CB3sTFtapa40s5eTKSU6ciLL73aWn+/P77eDZtus3YUwVw220bad36HDt3DuLo0a7o9XVzM4Kq3YR9fdPp23cfnp7ZeHll4+mZg5dXtjEJPCYmij17+gPg7FxE165HrW7H7bdv4PbbN1BU5ERqalOuXQsgJcWflBR/Ll0KvR6QWlbbw0i18X+qTZszVgWHTz/9X/R6B7755hFyctT3YcSIdQwYsMvsBwGAzz57Ak/P7ArPrwZBCcB2YGh1X0aV2C0AcnZ2plevXqxfv5677rrLuH39+vWMHz/e7DF5eXk4Opo22cFB/c9p+KQ7YMAA1q9fb5IHtG7dOgYOHGjrlyBEvXTmzBmTiQIVqfvx/vqdbOrllcWIEdblTSUkhDJnziyKi52MD0vBQnp6UxYtmgqoN6+nnvrYYo6Ooqg9NfHx4dV+HTfOpa00iCod/Gi1JXTseBxv72wmTFjJ0KFb2LVrIAcP9qCkxNnkuNq4Cbdqdd6qm3BYWBzFxU7077/b7H6Fhc44OJQYv09N9WPdupG4uORz660Vl1oBSE/3wdtbDaaCg68QHHzjQ/WqVXdw4EAfQJ2F16PHIVJS1ADp2jV/WrW6UOvDSNX/P6Xg5ZVFQEAK/v6GxzX8/VOsnhXo5ZULgKOjaV0xS8GPTqdFq9VXYdZhkpX71Zxdh8CeeeYZpk2bRu/evRkwYABfffUV8fHxzJw5E1CHpi5fvsyCBQsAGDt2LDNmzGDu3LnGIbDZs2fTt29fQkJCAHjqqacYMmQI7733HuPHj2flypVs2LCBHTt22O11ClFXdDqd1au922u8v74lm3p4ZJOb6wVAdrYXxcVOxlyZigKUCxdaV+uGryjaSnN0YmKi7NJLptc78vnnj9O79wEGDPgTX99MRo9ew5AhW9m9uz/79vWhsNDNBoGtHl/fDAIDk8nM9OHKlWAAQkMTrGqnl1cO5861ZufOgWRne5GT40V2tqfx6+Ji02CtoMCdXbsGotHo6d79cKXB56efPoVGo+Drm24MFpo2TSEg4BpXr94YTQgNvcSgQaaTdvR69cSWerDGjFnNxYstKShwparDa9b+n9JqS/DzS8ff/xpnz7Y1Bq+jRsXQv/+eKl2zrD/+GMPlyy3IzvYybtu+fTC7d/dHr3dAp9Oa/GsYunV3z7XyCsE1al9V2DUAmjx5Mqmpqbz55pskJSXRuXNnVq9eTXi4+sknKSmJ+Ph44/7Tp08nOzubzz77jGeffRZfX1+GDx/Oe++9Z9xn4MCBLF68mFdffZXXXnuNNm3asGTJEqkBJBqFPXv2kJ6eXul+tZ0MWt+TTR0ciunU6Th9+uyladNUPvzwWUpKnAANS5dOJjDwKnff/WutBSjW5OjYS1GRK3/+OYg9e/rSvXssgwbtpEmTDG67bRMuLoVcvty8SoGtg0MJLVpcolmzq9en8ifTrFmycfhw9+5+xgAoMTHEqjZmZ3uSn+/Bhg0jq/TaqhJ8Koo6TJiW5s+pU+bPl5LSlL17+xqDI2/vbIs9IaBez8srhxdffB9FgcJCFwoLXSgocGXNmjHExbUEIDj4MpGRJygocKGw0PX6v87ccccq43nKnldR4O67fyUzcwNNmmQYZw5+9dUMkpKaA5CW5oderyEtze/6kN6Nnqu0ND9mzvyy0uDwwIHe5X73CwvdKCx0s/i6AeLjw8nM9K5gdqIGdTbY4ArPY0t2rQNUX0kdINEQ5eTkMGfOHIqLKy+EFx5+genTf6jyNU6e7GAy6+jVV980maJdVfPn/61GeS5V5eubTu/e++nR4yDu7vmA2kX/44/TyrWjLobvGkJ+lEajo3Pnv+jffxeLF0/h4Ye/rfAmmZ/vxvLlEzh7tgOg9rA999x/y+1bUuLAtWsBHDvWmT//vOX6tawbHvzkk9k1ep9q62fbvfvBKie7GyxY8AAXLrQGoFev/dx556pqtwPUYcCUFH/Wro0iIUEtLOzoWIRer0WvN9/3UZX6VNVh6fw3EvxrPgusQdQBEkLY1saNGykpKal8R+pPFeA77ljFmTMdOHeuNefPt63y8dYGEAEByYwYsZ527c4Y//BmZnqzf39vDh3qaZIHY1BZErEtWJOjY2+K4sDRo105erQr4eEXKs3RcXfPZ8iQbcYAKDfXk8TEYLKzvUrVK1Kn85fNmaqr4cHa+tmmpzexar8ff7yfK1eCcHUtwNW1EBeXApKSgozPJycHsGdPP1xcCnBxKcTVtQBf3wyaNMmo9Nzbtt3C/v19rw9RmUaRZfO4yqrtnklL51d7fj6m0dQBEkLYTmJiIrGxsVbvn51d/oZvzuLFk42fHgGTmU0AH374rNnjQkPjmTJlSaXnDwhIJSDgTzw9s0sFQHoGDNjN5cshJCWFlMvnMKj8U7xhaq3ay9O+/RkAzp5tw/79fTh9uh2KUvEMp4YQoNQlawNn01lSGr7++u9WX6Ouhgdr42db2TBP2fyxvDzz/w8TEsJJSDBNgre21/b8+TaVzlKrSG0H/qXPXxuVoKtCAiAhGrjqrPeVmBhCSYlDuZkcN86p/qE+fbpDhX/4DHVWyjp9ukOlN4KcHE82bryNFi0uc+FCS+Nz/v4pjBy5DlATSq9eDeTSpRZcvtyCS5dakJrqR0TEyQrzUC5eDCcry4cVK9RPlGlp/vzxxx2cP9/apJCgqBprA+e9e/vW6Dp10ftWG2qzB8va4KquZg/a4vxXrrgwfvzQWrtOZSQAEqKBO3r0KImJiVbvr9WWMHHiLzg66szOdrLFUIM1N4LVq8dw8mQkhw/3MDlWo1E4frwjLVpcwts72zgNuU+f/QBs3nwrPXseKtfu0t+3ahWHXg/r1480Dm/t39+nWq9F3HAz3YRrS231YNXn2YMNlQRAQjRgRUVFrFu3zur9NRo9d921nPbtz1Bc7Mj27YPp1etAnY73V3b+a9cC+fnnyQB4eWXSosUlWrS4RPPmlwkJSaSkxLGSgmqqNWvGkJtrvoeqMdFq1RuiXl/9ZHUDuQlbp7Z6sOrz7MGGSAIgIRqw7du3k5eXZ/X+Y8aspnPnv9DptCxZMplz59qxY8fgOhnvr875s7N9OHHChxMnOgGg1ero1OmYVccWFLhRleUjbiZarRa9Xo+rqysdO3YkISGB1NRUmyyNIjdh69RWD1ZDHR6sjyQAEqKBSktL488//6zSTS0+PpTu3Q+xfPndnDvXDqi78X5b0OsdyMqyLsHT2nyVm4Uh6PHw8KBTp0507NiRsLAwtFotxcXFrFq1iiNHjtjkWnITtq+GOjxY30gAJEQDtXbt2iofc/RoNy5ebEl2to/J9lbnzjF6zRrWjB7NhTZtbNXEWlGXeSj1nSHx3cfHh86dO9OxY0dCQkLKLf7s5OTEhAkTCA8P548//kBRlBr3BslNWDR0EgAJ0QCdO3eO06dPW7Vv585HuXgx3LhwYdngB0Xhto0bCUhJ4baNG/mmdWvza0DUE409D8UQ9AQEBBAZGUlkZCQBAQHlgh5zx/Xs2ZPg4GCWLFlCVlaWTYbEhGioJAASooHR6XSsXr3aqmnvXboc4a67fiU9vQnffDOD/Hz3cvu0OXeO5tdnkTVPTKTNuXOca1v1ooR1qTHloZT+OYeEhBAZGUnHjh3x8/Or1vmCg4OZOXMmK1as4JSlNR6EaAQkABKigdm3bx9paWmV7te+/SkmTFiORgNnzrQjP9/MWj2KwrBNm9BrNGgVBb1Gw5hVq/h50iRSAwIodq64cqw93cx5KKWDnrCwMCIjI4mIiLDZ0jyurq5MnjyZXbt2sWHDBgDpDRKNjgRAQjQgubm5bN68udL9wsMvMGnSUrRahcOHuxITE4W5GVEdjx839v4AaBUFv4wM/v711wBk+PiQ4u9vfBzt0oUiV9cqt7u2coxupjwUQ9Cj1Wpp3bo1kZGRdOjQAXf38r12trrewIEDad68OUuXLiU/P1+CINGoSAAkRAOyefPmShc7DQm5zL33LsLRUceJExGsXDkeKN8r4p+czN2//lpqwQiVAug1GhwUBd/MTHwzM2l77hwAf3XubNyv7+7dBF29agyOrvn7k9GkCYq2zLUaWI6RrWlLvR+WavE4OjrSrl07OnbsSPv27XFxcamr5hEeHs5jjz3GL7/8QlxcXJ1dVwh7kwBIiAbiypUrHDhwoMJ9AgKSmTr1J1xcijh/vhXLlt1jdr2r8IsXuXfhQhx15ZfC0AAOisLSSZPI8fLC/9o1/FNS8MrOpsDtxjBauzNnjIGRQYmDA6lNm5Li78/yu+5C5+TUIHOMzLEmkNFqtbi4uODq6oq7uzseHh64u7vj6uqKm5tbuYe7uztubm44OztXmsRcmzw9PXnggQfYsmUL27dvt1s7hKhLEgAJ0QDk5+fz+++/G2u9WFJQ4EJurgdpaX4sWTIFnc6p3D5djhxh/PLlOChKud4fAz0waOdOvpkxg4SwMDN7wN5+/bgUGmoMkJqmpuJUUkJgcjJe2dnoHB2NOUaG6+g1GoZt2sS5Nm3qfS9Q6TycFi1a4O/vbzaIKR3IODk52TWQqQmtVsvw4cMJDQ1l2bJlFBcX26R6tBD1lQRAQtRjiqJw7NgxVq9eTWFhYaU5GtnZPsyf/yB6vYaiojLDKIrC4O3bGb5pE6D21pjrAQJ1wMw7KwsHnU4NZMw40749Z9q3v7FBr8c3MxP/lBRcCwpAo6HN2bPlcozqcy9Q6TycVq1aGfNwPDwaz5Ia7dq147HHHmPJkiVcuXJF8oLETUsCICHqqfT0dFatWsX58+cr3M/NLY8WLRI4c6YDYHmF9tGrV9N33z4A/hwwgD19++Ken2/xvLkeHhaDH7O0WjKaNCGjSRP1+zIzzAwUYPjGjfWmF8jQq+bo6Ej79u3p2LEj7dq1q9M8nPrGx8eHhx9+mHXr1rF37157N0eIWiEBkBD1jF6vZ/fu3WzatKnST9/OzgVMnfoTISGJrFhxF0eOdLO4b3x4OL337ydm9Gj29e0LQJYhWKkFpXN/StMAIUlJtD95ktMdO9ba9StSdq2siIgIWrdujWNVAr6bnIODA6NHjyY0NJSVK1ei0+nqrDfIMIyoKEq5elcajQaNRiPDczZW3ffVcJyBLaqM1xX53y5EPZKYmMjKlStJTk6udF9Hx2KmTFlM8+aJ5OW5kZgYXH6nUqWS/+rcmcshIWRUs4BelRh6fzA3/0x196+/8tHTT1NYS9O8yzLcSL28vIx1dQxrZQnLOnfuTFBQEIsXLyYtLa1Wbm6lc9tcXV0JDw8nPDycsLAwgoKCKC4uJj09nYyMDNLT042P1NRUsrKyTG7aWq22Qd2E64Kl4MbBwQFPT098fX3x9fXF29vb5OHp6YlOp6O4uJiioqJy/5rbVlxcTHFxMYWFhRQVFRn/NWwvKSkx/mzsPbQsAZAQ9UBRURGbNm1iz549ViXRarU6Jk78mVatLlJY6MxPP91PSkozk30Cr1zhzt9/Z2l0NNk+6vIXdRL8AA46HT6ZmRaDHwCX4mIiT5zgUK9etdYOQ9Dj5+dnXCA0KCioaonKGzbArFkwZw6MGFFrba3P/P39efTRR1m1ahVHjx6t8flKBzw+Pj60atWK0NBQwsPD8fPzK/fzcXBwIDg4mODg8kG+oihkZ2ebBEcZGRmkpqaSnp5Obm5uuWvfTAGSNcGNj4+PMagp/bWbm5tdkvZ1Oh1FRUU4OZWfpFGXNMrN8ltgQ1lZWfj4+JCZmWmzyqtCWHL69GlWrVpFTk6O2T/KGo2+TLXjUCZMWEnXrkcpLnZk4cL7iYtraXJMm7NnmbR0KS5FRRzr1IllkybV0au5wTszE/cyNx+DpqmpNEtOZvNtt9n0mqWHS4KDg43LRjRt2rR6J1QU6NcP9u2DPn1gz556kbdkL4qicPDgQVavXl2lIMIQ8Gg0Gpo1a2bs3QkLC8PLy6tW21xSUkJmZqZJcGToPcrIyKCoqMikjfVd6XY2a9aMoKAgk14bQ4Bjr+DG3qpy/5YAyAwJgERdyMnJYc2aNRw/ftziul4REcfLrXdVWOiEi0sxOp2WJUumcOZMe5Njehw4wJ2rVqFVFC60bMnSyZNN6vfURy4FBbgWFJDp61vlYw3vnUajITw83JjTY5P/u2vXQlTUje9jYmDUqJqft4FLSkpi8eLFZGdnl/u9LZ2/4+DgQPPmzY1DWi1atKh3yeXZ2dnEx8cTHx/PxYsXjcPP9SUgKt2OwMBAWrduTcuWLQkLC8O1GlXZb3YSANWQBECiNhk+Ra9bt47i4mKLn6IjIo4THb0UML/i+e7d/Vm3LsrkieGbNjH4eiG7w1278vu4cVWbyWUHTkVFTP3pJ5qkp7PggQdIDQio9JjSQU/Lli3p3LkzERERtl02IjkZIiMhNfXGtt69Ye/eRt0LZJCfn8+KFSs4ffq0cZuLi4uxdyc8PJzg4GAcHMoX4qzPCgsLSUhIID4+nri4OC5fvoxOpzMJ7KxVnSVgJOCpmarcv+v3X0YhbjLXrl3jt99+49KlSxXup9HoiYqKuf512efUICgy8jjr149EUbQ4lJQwbuVKul7Pz9g6ZAhbhg1rEDdq58JC3PLz8c7O5sHvv+fHBx7galBQhceEhobSpUsXOnbsaPtESkWBBQvgySchO9v0uf37Yd06GDlSDYz8/W177QbEzc2NKVOmcOTIEYqLiwkPD8ff37/BD7u4uLjQtm1b2l6vU6XT6UhMTDQGRHFxccZhM0s9t4DVS8BIwGM/EgAJUQdKSkrYsWOH1csMhIXFmQx7laXRgI9PFmFhccTFtcKhpITAq1fRabWsuvNOYnv2tFXTa12ulxfzp0/n/p9+IiQpib/Nn8/CqVO5HBoK3LjJNG/enC5duhAZGVl7eSPnz8Ojj8LGjeafd3CA114DJye44w546CF4/nlo2bJ22lPPaTQaunWzXHqhxupBArqDgwOhoaGEhoYyaNAgFEXh2rVrxmGzCxcukJOTA5gGM5aWgJGAp/6QITAzZAhM2FJcXBwrV64kPT3d6mM6dz7KPfcsq3S/Zcvu4dixLgB4ZWbin5rKhdatq91We3IpKOC+hQsJS0igyNmZRffeS8GAAXTp0oVOnTrhc30mW606ehR69gStFq5/yjdrwgRYsUL92sEB7rsPXnpJHTITttGAEtAzMzNNAqLUa9d44rPP8EtLMy4BkxQczO+vvkrrNm0k4KlFkgNUQxIACVvIz89n/fr1HDp0qOKucjPCwy8wffoPFe+0F3Z91o91bUbXsKUqQz0ceyV+ajQaHAsLeWDZMlqcOoXi6opm+XLTJOTacPkyNG9+4/sffoD//hf++gvMvRdarRokvfcevPuu2kthMGECvPwyXC80KWqgoSag79uH7sEHcfjrr/LPNZTX0IBJAFRDEgCJmlAUhb/++ovVq1dTUFBQrXojGo2ep576GG/vLLMfepXlwFTQ5MPCqVM5265dtdpqCMxcXFzo3LkzJSUlxmnCOTk5JsFQbVTgNVy/adOmdOnShc6dO9PUwwMmT4Y//4StW2uvVyUrC/75T/jqK7WXwTCUU1gI4eFw9arlY4OC4OJFcHFRj333XVi+XH2ufXs4cUINlIT1SkrgwgX1vTtxAv7znxsJ6Iags74noCuK2lt14ID559u2hdOn6/draOAkCVoIO8nIyGDVqlWcO3euRudRFC0JCaF07vyXWsx5IzALmAPKMeAZ0Chwpm1b4i2s1l4RQx5CUFAQ/fr1o1OnTuWWgVAUhdzcXDIzM8s90tPTyczMpKCgoNx5K6sPYwh6fH19jUFPs2amRRz55Rc1wKhmYFepFSvgiSfU3h+AP/64EQAZgppr1ywf36yZuh+oN7xff1Vv2u+9B8OH3wh+Cgpg/Xo1X+hmDIiqk6OTmwunTsHJk9CxI/TooW7ftg0s1YXS628koNe3HpSMDDUnzMNDDWyioy0HQGfP1s/XYE49yL+qbdIDZIb0AImq0Ol0XLp0iTNnzrBnzx6brJnUuvVZ7r//JzQayMt1wX1YIewDAoHrHRP7e/Vi9ZgxKFWcZuzg4ECXLl3o06cPISEhNWpncXExWVlZxsAoIyODrKwsY8G57Oxskx4jb29vY05PlSoyb9qkBhiPP16j9nL5sjq7y9Bb06YNfPll7f2BnzcPZs6ETp3UHKEpU6CelyWwmjU5OmlpsHSpGuycOKH+Gx9/4/mXXlJ7zwASE9UekvbtISEB0tNv1HwANYDs1Uu9TnIy+PreCELtoagI5s6FN99UA4V///vGe3LggOXhU8NriImBYcOgPuYBNaD8q7KkB0iIWqQoCmlpaZw7d46zZ89y4cIFSkpKbFY4TavVceedq9Bo4MCBXpz8pANT9/1PffJ68LN+xAj+HDTIqj9Khnb5+PjQr18/unfvjpuNCiM6OTnRtGlTi5WWS/ciabXaqi9DAWpP0Lhxas9BZqY6bFUd338Ps2erQ1+Ojursrddeg9osEllQAN7eaj7RtGnq9Z5/Hh588MZ1G+on7XXr1BskqP/OmKHe9AcNgocfVrdnZ8Njj5U/NiBA7f0p3XsZHAw5OWqPmbm8L71evc66dWrQuncvPPecel1PT9u/PksUBZYtU4M3Q0/v6tXqz7a4WA3wLP0d0OvV4G73brjzTjUAnzdPDYTqk7I/24bSa1VF0gNkhvQAibLy8/O5cOEC586d48yZM2Rfrw9T1eRma/n7X2PIkK2sXDGOB7+cT0hiIhpAAdKbNOHTWbMqDX4MbWvfvj19+vShTZs2DbNGi6LAG2+oD1BvPP/3f1X/RPrpp2qg0a8ffP01dOli+7aak5kJX3wBH310Y1gtMBCeeUa9gffv3/A+aSuK2pNx6FD55yZOhJ9/Vr/W6+Huu9WenYgINeiJiABLS5NY04PSvbuan2UYvvTzg6eeUoc0a3utu1274Nln1X9B/Tm+9ZYa0Bp69hISKh8+PXUKHnhA7fUC9fj//Mfy+1KX8vPVHLjSr8HbW21jx47q/5uBA+3XvkpIEnQNSQAkdDodly9fNgY8SUlJQN2Xx29z9iz3//RTue0/3X8/564XaivNEPS4urrSu3dvevXqhW81lpeolz74QO09AfVm98knFefVFBRAXBx06KB+r9OpwzHR0erU9bqWlwfffafe6OLjYcwYNSBriDOd5s6Ff/yj/Pb771cf1X0N1iagnzql/izfe0/NqwG1F2jmTDWwNLNoao3NmaMGWgDu7mrw+vzz1e99MvRmzp2rBn4BAfDxx3DvvXUfBBcXq3lMoP4Ojq5gZmmPHnDw4I3v//UvNUAyBLgtW1r3/6uWej4lAKohCYAaH0VRSE9PNxnWKi4urrUeHnPatz9JQYEr8fEtDY3ika+/JjgpCW2pNhhqinwzY4bxD6UhMAsJCaFfv35ERkaWS2q+Kcybpw6pKApMn6725Jh7nZs3w9//rgY9R4+qN6z6orgY/vc/9WbxxBPqzUSnM80Pqc+9QIqi9rRkZJhud3BQZ2rVtP3W9KC0aKF+rdOpCfPvvguHD6vb5sxR87xs7cIFNZfrvvvUvJ8a5s8Z/fmnWnzTMG0+OhqWLLHNuSuiKOow4ldfqZMAzpxRgzlzPXAajfq+9+yp/t7+97/qdp1O/b9VumaWi4uaxxURoQ7tmRsCrcUcIwmAakgCoMahoKDAZFgrK0utvFyXQY+Bn18Kjz76FU5OxSxY8Dfi4lpa7P0xMPQCOTg40LVrV/r06UNwbXzyrW9++kkNfnQ69WbXseONT5I9eqifyr//Xt03OFj9RNu1q12bbFbZOjcGK1bA+PF13hyrWWq3gT16sRQF1qxRc4MWL74R8G7friZLV3W4s6AAPvtMDXo+//zG9mvX1J4aWysqUnsG33pL7QWaOdP21zBIT1f/D339tfrhwOCnn9SlXarys83Lg/ffv5HkfuqU2otnUHY4tHNntYfP1fVGIVFz560BCYBqSAKgm5OiKCQkJBh7eRKvj7/be9VnR8ciHnnkGwIDk7l4MZwFCx5A0Wt55Ouvjbk/ZemB5NBQLi5ZQvcePRpfRdnly9VelIUL4ZZb1E+SrVurCc4pKeqnycceU3OF6qKCdFUZPgEben9Kc3GBnTvV3qD64upV9b2ePVvNWbJmlpO9e7H0ejXwOX5cTTj+5z9hwADTfcoOwyiK2vvy8stq8j2oeU7du9dNmy9cUAMEw9Du9u1qsnzv3jU/9/nz6ky1X35RAzxQA5FJk9RE8kGDav6z1enUYWdDQNS27Y1gPj5efW1l2arn8DqZBSZEGXq9nmXLlnH8+PFyPTz2DH5A4Y47/iAwMJmcHA+WLZuIojjgoCvBJzPTbPADoAUCi4sJ6tnTvlOB7eWuu9TH2rU3ZqucP6/+26mT2q1fjxM1TWbZlFVYqAZH776rJtzau37Q7t1wzz1qwq6rq3WznIqK7P97mZ2t9jicOAGrVqmPW29VA6Hbb1f3+ec/1ef/+U81B+b552/8XEJC4O236y5ZHqBVqxtf5+SoMwcTEtQg7a23qp5vpCg3ggpHRzWIVRS1R3TGDJg6FZo0UZ8vLKz5z9bBQf0g0rq1muNWWrNmaq2nZcvU/D0Dnc5uM82kB8gM6QG6ueh0OpYtW8aJEyfs3ZRyevY8wNixv6PXa1iw4AHi4m78Aex09CgTly2jyNGRn+6/H52rKxMnTqSJ4Q9W6VyIxshcL0rz5urUZHvffCtS2Uyn0oYPV4cm7DG0qShqIPnkk2reUseOas+bu7v1OTr1wenT6jDNggXq6wC1J2PyZHjhhfL7e3rCiy/C00+rxQ3tJT1dfe8XLlS/DwtTZxPecYf6vaUkYr1erZv19ddqsGKoeQVq4DFggJp3Y663pSr5V9VhqefThr1AMgRWQxIA3Tzqc/ATFJTIww9/i6Ojjg0bRrBz5y0mz0cvXkzHkyfZ26cPMXfeSZ8+fRhd0eyMxsZSLkp9n0llzUwnb2/1Zu3lBUeOqNOt61JBgZqg/e236vf33KPmVXl51W07bCkhQU3e/eordap3+/ZqsFz6Rvz3v6vlFur6/a7I2rVqTpBhSC46Ws0TGj/eNIn4yhWYPx+++eZGb6hGA5cu2S5hu6bqIH9MAqAakgDo5qDT6fjll184efKkvZti1u23r2XgwF2cOtWexYunoA5sqbwzMnjqk0/QKgqfP/446UFBPPXUU3g15BuQLdXBJ8laZc0n7bw89aY2ZMiN7YWFtd+7lZCgBjz79qnDb//3f2pPSX1+P6vi2jU1F2bu3PLP1dfgOTdXDcw+/FD9fffwULcZDBigzugy/F/w9laHt2bMuLHUiL1VpUp2DX7XJAdINHr1PfgBWL9+JCkpAZw40ZHSwQ9A7/370SoK51u1IrVZM/r27i3BT2mWcmjsmE9QJaGh6qMy7dvf+HrJEnj1VXVIpDZXmz9xQl13y89PnVFlyJe5Wfj7q6/PwaF88PzaazByZP0L9jw81GG8e++FRx5Re3gM7ddobhRmHDhQDXomTbLv8J05RUX1Ln9MeoDMkB6ghq0hBD8V0ep0PPPf/+KRl8eSyZM507kzs2fPxrMuy/3XZ3X0SbJe0evVT/JHjqg3vjfeUCti11ZBx/nzYehQtajdzaY+TuOvitWrb+QBlTZvnlpPqD6r7RwjZAisxiQAarh0Oh0///wzp06dsndTzAoPv0ifPnv5/fdxFBZanrrul5JCt8OH2TpsGP0GDWLkyJF12Mp6ztpqwRcv1u9k6KpKT1dzVAx1VQYPhh9/ND+1uCpyc9Xp7S++qE5bvpk19OC5oQ/91gEZAhONkiH4OX36tL2bYpaHRzb33PMLXl45ZGT4smGD5aAmzd+fzbfdhoODA4MGDarDVjYALi7qMFdlnyRvpuAH1OnKS5aon/6feEKtEdOtm1r8b8qU6p3z7Fm1nMCxY+pN1ZD3c7Oqh8MwVdLQh37rGQmAxE1Bp9OxdOlSzpw5U+dVnK2h0ei4555leHnlkJwcwNatQ83vWKpuh0ajoV+/fnjUt7H8+sDaHJqbjUYDf/ubWvxx6lT1E/+996pT1Lt1q9q5/vhDPUdmptpjVtnaajeDhhw8K4qao6TVWu69qq85TPWUBECiwavvwQ/A8OGbadXqIoWFzixdOpniYmez+41buRLnoiK2Dh1KekiI9P4I89q0UXuA3n5bLfhXleBHr1ePe/119aY6cKA6rFZfpkrXtoYaPDf03qt6SAIg0aDpdDqWLFnC2bNn623w0779KW65ZQcAv/02jtRUf7P7uefk0OXoURx1Onbdcgv9+/fHvT4t4inqFycnNRm69O99XJxaOPHFF80vEpudrfb6/P67+v0//gEffQTO5gNyUY805N6rekoCINFglZSUsHTp0nod/Pj6pjNhglqJdc+efhw/3tnivj0PHsRRp+NS8+Ykh4dzX9l1i4QwxzDcoderSyds364Oby1cqC6tULpi8ODBkJys3iS//FJdVFY0HA2196qekgBINEglJSUsWbKEc+fO1dvgB8DNLZ/iYidSU5uybp3leioanY7e+/cDsK9fP+n9EVWn1aqzxA4fVuvCdOumrmQ+Z86N9a727FHXYrpypX4ttiqEHUgAJBqchhL8ACQlhTBv3kwcHHTo9Zb/u3U4dQqfrCxy3d053bUrs6T3R1TH1Knqqt7336+uKP/AAzeeKz1LqHlz+7VRiHriJk/5FzebhhL8ODkVGb/Oy/MgO7viehR9rk9tPdirF/1uvRU3N7dabZ+4ibVsCVu2wJtvmm43zBKqx/9vhKhLEgCJBqOkpITFixfXu+BHo9ETHn6Bzp2PEh5+gWbNrvDUUx/Ttethq473T06m9YUL6DUaDg8YQP/+/Wu5xeKm5+hYfrkMvf5GL5AQQobARMNgCH7Onz9fr4KfiIjjREXF4OOTZdym02lxcNDTtethjhzpQmWfM7K9vYmJisI3M5MuY8bg6mq5QrQQVjHUjGlI610JUcckABL1XklJCYsWLeLChQvVDn40Gj1hYXF4eeWQne1JfHw4ilKzDtCIiONERy8tt93BQY+iwNGjlQc/AIWuruzp3x8XFxeelt4fYQtSMViISkkAJOo1WwQ/5nppMjO9iYmJ4uTJyGqdU6PRExUVc/1r8/sMG7aZI0e6WRVoaTQaBg0ahIvU8BA1JRWDhbCK5ACJestWwU909FK8vbNMtnt7ZxEdvZSIiONWnUej0ePmlkvTptcIDY2ja9fD+PhkWbx/aDTg45NFWFic5ZPq9dzz8890jY3FTaulb9mcDSGqoyoVg4VoxKQHSNRLxcXFLFq0iIsXL9Zo2MtSL41Go35QvuOOP9DpHHBzyycjw5f4+JYAuLnlMnnyEjw8cnF3z8PNLd/kHBcvWrcCt5dXjsXnWl+4QOe//qLdmTMcfOwx6f0RtiEVg4WwigRAot6xRfADEBYWZzLsVZZGA56eudx33yIADh3qbgyAioudCA+PL3dMfr4reXnuZGd7WtWGivbrs3cvAEd796bX0KFWnU8Iq0jFYCEqJQGQqFdsFfxAxb0vpaWn+5Ka2pRr15oZt5WUOLN0aTR5eW7k5bmTl+dBfr4ber0DYEiqTsDb2/wwmKJAVpY38fHme4p80tPpcOoUAA5PPIGzrMUkhBB1SgIgUW/odDr+97//ERcXZ5Op7tb20qxcOZ64uFbltp84YTlBWlG0xMREER29FEUxHWIzND0mJspiAnSf/fvRABfataPTPfdY1U4hhBC2I0nQot7Yv3+/TXp+DPz9U1AUy4VvFUWdDWapl6YyJ09GsnRpNFlZplWes7K8Wbo02uIMM8fiYnocPAhA4SOPSO+PEELYgfQAiXqhsLCQLVu22OhsCsOGbWbIkG3qd9eDoKr20ljj5MlITp2KqFKNoU7HjuGen09mkya0mTWr2tcWQghRfRIAiXph9+7dFBYW1vg8Wq2OO+/8nR49YgHYsmUoV682K1cHKCurZnWASlMUrdkhNEsyfH0536oVjmPG4CNVn4UQwi4kABJ2l5uby44dO2o89OXkVEh09FLatj2HXq9h1ao7OXSoF0CVe2lqU1yrVvzauTOzn3rKLtcXQghRD3KAvvjiC1q1aoWrqyu9evVi+/btFvedPn06Go2m3KNTp07GfebPn292n4KCgrp4OaIatm/fjq70ekXVojB16kLatj1HcbEjixdPMQY/cKOX5tixLsTFtbJb8GMwZMgQHJ2c7NoGIYRozOx6F1iyZAmzZ8/mlVde4dChQwwePJjRo0cTH1++/grAJ598QlJSkvGRkJCAn58fkyZNMtnP29vbZL+kpCRZYLKeysjIYN++fTZIfNawa9dAcnI8+OGH6Zw508Em7bMlj5wcbt2yhUC9np49e9q7OUII0ajZdQjsww8/5OGHH+aRRx4B4OOPP2bt2rXMnTuXd999t9z+Pj4++Pj4GL9fsWIF6enpPPjggyb7aTQagoKCarfxwiZqmvis1eqMtXlOnYrg/PlWFBfXzwq3PQ8cYOiWLfTKyMDxjTfs3RwhhGjU7NYDVFRUxIEDBxg5cqTJ9pEjR/Lnn39adY5vv/2WESNGEB5uOo05JyeH8PBwWrRowZ133smhQ4cqPE9hYSFZWVkmD1H7kpOTOXz4MHpLaxZVon37Uzz++Gf4+KQbt9XX4Eer09F7/34APJ55xs6tEUIIYbcAKCUlBZ1OR2BgoMn2wMBArly5UunxSUlJrFmzxth7ZBAREcH8+fP57bffWLRoEa6urgwaNIgzZ85YPNe7775r7F3y8fEhVErI14mNGzei1VbvV7Bnz/1MnrwYP790Bg60LmC2pw4nT+KdnU2xnx/a6Gh7N0cIIRo9uydBa8qsI6AoSrlt5syfPx9fX18mTJhgsr1///7cf//9dOvWjcGDB7N06VLat2/Pp59+avFcL7/8MpmZmcZHQkJCtV6LsF5CQgKnT5+uRu+PwtChmxg7dhVarcKhQz1YuzaqVtpoS32vr/vlMHOmLEIphBD1gN1ygPz9/XFwcCjX25OcnFyuV6gsRVH47rvvmDZtWqVVdLVaLX369KmwB8jFxUVW4q5DiqKwfv16NBpNlZKfy9b42br1VrZsGQpUHjDbU7OrV2kZF4fi4ID2scfs3RwhhBDYsQfI2dmZXr16sX79epPt69evZ+DAgRUeu3XrVs6ePcvDDz9c6XUURSE2Npbg4OAatVfYztmzZ0lISKhS8OPkVMiUKYvo0SMWvV7D77+PZcuWYdT34Aegz7596hcTJkCLFnZtixBCCJVdZ4E988wzTJs2jd69ezNgwAC++uor4uPjmTlzJqAOTV2+fJkFCxaYHPftt9/Sr18/OnfuXO6cb7zxBv3796ddu3ZkZWUxZ84cYmNj+fzzz+vkNYmKVbf3R6MBD49ciosd+eWXSZw+Xf+muVuic3BA7+KC9skn7d0UIYQQ19k1AJo8eTKpqam8+eabJCUl0blzZ1avXm2c1ZWUlFSuJlBmZibLli3jk08+MXvOjIwMHn30Ua5cuYKPjw89evRg27Zt9O3bt9Zfj6jcsWPHuHbtWpWPKypy4X//m4qPTwaJiQ2rF2X3vffS548/wM/P3k0RQghxnUax1dLbN5GsrCx8fHzIzMzE29u78gOEVXQ6HZ9++imZmZnlntNo9OWWqggOTiQ09BJ79vS3Q2ttZ8KECXTr1s3ezRBCiJteVe7fshaYqDMHDx40G/xERBwvt1hpbq4bzs5FODnpyMjw4dSpjnXZVJsIvHIFH09PunTpYu+mCCGEKEMCIFEnioqK2Lx5c7ntERHHiY5eWm67u3s+Gg0kJgZx/nzrumiizQ3fuJH2Z85AUBA8+6y9myOEEKIUu9cBEo3Dnj17yi1Iq9HoiYqKuf41ZZ4DRQEPjzxKShreoqFN0tNpd/as+s3YsfZtjBBCiHIkABK1Li8vj+3bt5eb9RUWFoePT1a54MdAowEfnyzCwuLqoJW21XvfPjSKAqNGQfv29m6OEEKIMiQAErVux44dlJSUlNvu5ZVj1fHW7ldfOBUX0zM2Vv3miSfs2hYhhBDmSQAkalVWVhZ79uwxW/MnO9vTqnNYu581qrv2WFV0PnoU17w8aNUKRo+u9esJIYSoOkmCFrVqy5YtFgsexseHk5npjbe3+WEwRYGsLG/i48Nt1p527dpRUlJCRkYGWVlZFBcXmzyv1WpRFKVKRRpL0wADDh5Uv/nHP8DBoYYtFkIIURskABK1JiUlhUOHDll8XlG0xMREER29FEUxTYQ2xB8xMVEoSs16bTQaDQ4ODowfP75c9fDCwkLjIrhZWVnGf9PT08nIyCAnJ8dkwVaNRmOsYm0uSPLIzsa3uBhcXeGhh2rUbiGEELVHAiBRazZu3IhWq61wxffz51uTnu6Lm1sBbm43ZollZXkTExPFyZORNWqDRqPB19eXKVOm0KxZs3LPu7i40KxZM7PPgbp0R25urkmAVDpIyszMJC8vz7i/e5s2OCYkwJEjUvlZCCHqMQmARK24fPkyJ0+erHS/nj0P4eeXwbVrTVmyJBovr1xjJeia9vwAtG/fngkTJuDq6lqt4zUaDZ6ennh6etK8eXOz++h0OvJ//x3n55+n5L//RePkBL161aTZQgghapkEQKJWbNiwodLeH41GR79+uwHYvXsgcXG2KXhoGKK67bbbGDRoEBpL8+xtxEGrxfP//g/OnsX5rbfUuj+1fE0hhBA1I7PAhM2dO3eOixcvVhj8AHTseAJf30xyc905cqSrTa6t1WpxcXFh2rRp3HLLLbUe/ACwbh3s26d+vX8/fP117V9TCCFEjUgAJGxKURTWr19vReChMGDALgD27etjk2rPGo2GZs2aMXPmTFq3rqPlMxQFXnjBtMfnq69uZHELIYSolyQAEjZ1/Phxrl69Wuk08tDQeFq0uExJiQP79vWxybV79OjBww8/jI+Pj03OV6HiYlixAvr1UxOeS7/eAwfUXiEhhBD1luQACZvR6XRs2LDBmINTEUPvz5Ej3cjLq36hQ8O09DvuuIOePXtW+zxVFh2tBkDmODjAa6/ByJGSCySEEPWU9AAJm4mNjSUjI8OqIoK7dg3gxIkIdu3qX+3rGWZoPfzww7Ub/BQVwc8/Q2rqjW0TJoClniadTs0Jkl4gIYSotyQAEjZRXFzM5s2brd4/ISGcpUunkJJivv6ONVq2bMnMmTMJCQmp9jkqdOoUPP88tGih9vgsWHDjuSlToG1bsLS0hlar9gJJLpAQQtRLMgQmbGLv3r3k5ubW2fUGDx7M0KFDbb+2V34+LFumzuTatu3G9uBgtbpzaZcugaWZbno9JCSovUcuLrZtoxBCiBqTAEjUWH5+PttKBwsV6Nt3N02apLN79wAyM32rdB2NRoOjoyP33HMPHTp0qEZLK1FQoC5gevWq+r1WC2PGwIwZ6r+Opf67uLiow1zXrlk+X7NmEvwIIUQ9JQGQqLGdO3eWW1TUHK22hFtu2YGXVw5JScEcOdLd6mtoNBqaNm3KlClTaNq0afUaumEDzJoFc+bAiBGQmwubNqmFC0Ht4RkyBPbsgYcfVtfyatHC8vlCQ9WHEEKIBkcCIFEj2dnZ7N6926rE586dj+HllUN2tifHjnWudP/SIiMjGTduHM7OztVrqKLAP/8JJ07AU0/B4MGwaBFkZanbIiLU/b78Uk1ullXchRDipiYBkKiRrVu3VlrxWXWj8OHevf3Q6yv/1TMUUxw1ahR9+/atWVXnNWtuVGs+flx9ALRpA4mJNwIgWcBUCCEaBQmARLWlpqZy8OBBq3p/WrW6QFDQVYqKnNi/v/KFQjUaDW5ubkyePJmwsLCaNfTSJZg82XSbnx8sXQrDhlmeySWEEOKmJQGQqLZNmzZZVfQQYMCAPwGIje1OQYF7pfs3b96c6OhovLy8atxOdu2CnBzTbWlpUFIiwY8QQjRS8tdfVEtSUhLHjx+3avjL3z+Zdu3Ooiiwe3flhQ/79u3L9OnTaxb85OWpeT+KAv/5T/lAx1CtWer0CCFEoyQBkKiWDRs2WF2DJz/fjT//HEhsbHfS0yueweXr60tUVBQONUlC/vNP6NwZfvzxxkrtZQM1qdYshBCNmgyBiSq7cOEC58+ft3r/3Fwv1q8fWel+Go2Gnj17Vj/ZuaQE3n4b3npLDXg++ECtw6PVmi9YaKjWLGt2CSFEoyM9QKJKFEVh/fr1NZuRVcG5u3fvXr2Dz59Xa/i88YYa7EybBhs3qtWYranWLIQQolGRHiBRJSdPniQpKcmqfR0dixk3biUHD/bi4sWWgOWgSaPR0KZNm6rn/SgK/PQTPP44ZGerNXzmzoV771Wfl2rNQgghzJAASFgtOzubtWvXWj3zq2vXI3TpcozQ0ATmzJmFoljO61EUhV69Kp8eX85ff8Hf/qYGQoMHq3k/4eE3npdqzUIIIcyQAEhYJS0tjR9++IGcnByrgh/Q07+/Wvhwz57+FQY/AG5ubrRr167qDevcGV5+Gdzd4aWXpIKzEEIIq0gAJCp19epVfvjhBwoLC62s+gzt2p0lICCFggIXDh7sUeG+Go2GHj16WDfzq7gY3nkH7r8f2rZVt73zjlVtEkIIIQwkABIVio+PZ+HChRQXF1vZ86My9P4cPNiToiLXCvdVFIUePSoOkgA4cwamTlXzetasUae7S4+PEEKIapBZYMKis2fPsmDBgioHP4GBSbRufQG9XsOePf0q3Fej0dC8eXP8/f0t76Qo8N130KOHGvw0aQLPPy/BjxBCiGqTHiBh1rFjx/j111+rFPgYGBY9/euvTmRl+Va4b6XJz2lp8OijsGyZ+v3QobBggSQ2CyGEqBEJgEQ5+/fv548//qj28RcvtqR588vs2jWg0n0dHR3p1KmT+SdPn4bhw+HyZXB0VHN9nn1Wen6EEELUmARAwkhRFHbs2MGmTZtqdJ7Y2J7ExnanshFWrVZLly5dcHZ2Vjds2ACzZsGcOTBiBLRsCYGB4OEB//sfVGeavBBCCGGGBEACuFHhedeuXTY6Y+XpZXq9/kbys6LAP/8JJ06o09n37QNnZ1ixAvz81CBICCGEsBFJghbo9Xp+++23Ggc/nTodo2fPAzg6Flu1v5+fHy1atFC/MSxaCnDgwI1FSkNDJfgRQghhc9ID1MiVlJTwyy+/cOrUqRqdR6PRM3z4Rvz80tFoFA4c6F3J/hp69eqlrimmKGoxw9JefVUWKRVCCFFrpAeoESssLOSnn37i9OnTNT5Xhw4n8fNLJy/PjSNHulp1TLdu3dQv1q2DQ4dMn9y//0YvkBBCCGFjEgA1Unl5efzwww/Ex8dXa6p7WYap7wcO9Ka42LnCfbVaLR06dMDDw0Pt/XnttfI9PQ4O6nYbtE0IIYQoSwKgRigrK4tvv/2WK1eu2CT4ad48gbCwBEpKHNi7t2+l+5skPxtyf8q2Q6dTt0svkBBCiFogAVAjk5qayjfffENGRoZNgh+40ftz9GgXcnK8Kt3fw8ODtm3bWu79MdBqpRdICCFErZAk6EbkypUrLFiwoEqLmlbGxyedjh1PALB7d/9K9zcsfKrVaqGwEOLjLQc4ej0kJEBREbi42KS9QgghBEgA1GhUd1HTyjg5lXD2bFu0Wj3JyUGV7m+y8KmLizrMlZSk1v/JyYEBZapHN2smwY8QQgibkwCoEThz5gxLlixBr9fbNPgBSEkJYNGiqTg4lFS6r0ajITQ0FD8/vxsbQ0PVR9/Kc4eEEEIIW6l2DlBGRgbffPMNL7/8MmlpaQAcPHiQy5cv26xxouaOHj3KokWL0Ol0Ng9+StPpKo+lK134VAghhKgj1QqAjhw5Qvv27Xnvvff44IMPyMjIAGD58uW8XLagnbCbffv2VXtF98potTqGDNmCl1em1cc4OzvTsWNH043r1qmrvcfE2LiFQgghhGXVCoCeeeYZpk+fzpkzZ3B1dTVuHz16NNu2bbNZ40T1KIrCtm3bWL16da1dIzLyOMOGbeHhh79Fo6k8oVqr1dK1a1ecnJxMn1i5Er7+GmqxrUIIIURZ1coB2rdvH/PmzSu3vXnz5ly5cqXGjRLVpygKa9euZc+ePbV5FQYM+BOAgwd7oShVXPi0tM2b1X+HDbNlA4UQQogKVasHyNXVlaysrHLbT506RUBAQI0bJapv27ZttRz8QHh4HCEhSRQXO7J/f8VrfhkEBAQQHBxsuvHKFXX2l0YDt95aCy0VQgghzKtWADR+/HjefPNNiovVVb81Gg3x8fG89NJL3HPPPTZtoKiaEydO1Mp5NRo94eEX6Nz5KMOHbwTg8OFu5OVZt1K7ceHT0rZsUf/t1g1KzwwTQgghalm1hsA++OADxowZQ7NmzcjPz+fWW2/lypUrDBgwgHfeecfWbRRWKikp4dq1azY/b0TEcaKiYvDxMe31S05uZtXxWq2WLl26lH/CEADJ8JcQQog6Vq0AyNvbmx07drBp0yYOHjyIXq+nZ8+ejBgxwtbtE1WQnJxsswrPBhERx4mOXlpuu6LA6NFryM724uTJSIvHa7VaIiIicHd3L/+kIf9n6FAbtVYIIYSwTpUDoJKSElxdXYmNjWX48OEMHz68NtolqiExMdGm59No9ERFxVz/uuxzahAUFRXDqVMRFhOhDcFxOTk56r9aLQwZYstmCyGEEJWqcg6Qo6Mj4eHh6HS62miPqIGkpCR1jS0bCQuLw8cny+JapRoN+PhkERYWZ/EcXl5etG7duvwTnp5w6pSaCO3ra5sGCyGEEFaq1t3y1VdfNakALeqHS5cu2XQIzMsrp0b7aTQaevbsWT75uTSZNSiEEMIOqpUDNGfOHM6ePUtISAjh4eF4eJjOBDp48KBNGiesVxsJ0NnZnjXaT1EUunfvbv6gkhJwlKXohBBC2Ee17kATJkywcTNETV29etXmS17Ex4eTmemFt3e22WEwRYGsLG/i48PLPafRaGjZsiW+5oa3EhKgY0c1+fm339Q8ICGEEKIOVSsA+ve//23rdogasnUCNICiaElKCsHH5xSKYpoIbYi1YmKizCZAK4piPvkZ1Nlfublw7ZoEP0IIIeyiRmMQBw4c4MSJE2g0GiIjI80vdSDqRGJiIlqt1qY5QJGRx4iIOAVAfr4b7u75xueysryJiYmyOAXexcWFiIgI8yc21P+R6e9CCCHspFoBUHJyMlOmTGHLli34+vqiKAqZmZkMGzaMxYsXy3IYdmDrBOiAgKuMH78SgJ07B7Fx422EhcXh5ZVDdrYn8fHhFqe+azQaunXrhqOlHB9Z/0sIIYSdVWv84cknnyQrK4u//vqLtLQ00tPTOXbsGFlZWcyaNcvWbRSVKC4uJjU11Wbnc3HJZ/LkJTg7F3P+fCs2bhyOomiJi2vFsWNdiItrVeECqBUOf128qD4cHeGWW2zWZiGEEKIqqhUAxcTEMHfuXDp27GjcFhkZyeeff86aNWuqdK4vvviCVq1a4erqSq9evdi+fbvFfadPn45Goyn36NSpk8l+y5YtIzIyEhcXFyIjI1m+fHnVXmADY+sEaL1ey5UrQWRk+PDLLxNRFIcqHR8UFERgYKD5Jw29P336qLWAhBBCCDuoVgCk1+txcnIqt93JyalKwzBLlixh9uzZvPLKKxw6dIjBgwczevRo4uPjze7/ySefkJSUZHwkJCTg5+fHpEmTjPvs2rWLyZMnM23aNA4fPsy0adOIjo6u9RXS7cnWCdDFxS788sskvv32EfLzrVvstLRevXpZflLW/xJCCFEPaJRqdB2MHz+ejIwMFi1aREhICACXL19m6tSpNGnSxOoel379+tGzZ0/mzp1r3NaxY0cmTJjAu+++W+nxK1as4O677+bChQuEh6tTsSdPnkxWVpZJT1RUVBRNmjRh0aJFVrUrKysLHx8fMjMz8fb2tuoYe1q5ciVHjhypcQ5QkyZppKc3ASooXFgJBwcHnnvuOVxdXc3vMG8e/PILvPKKJEELIYSwqarcv6vVA/TZZ5+RnZ1Ny5YtadOmDW3btqVVq1ZkZ2fz6aefWnWOoqIiDhw4wMiRI022jxw5kj///NOqc3z77beMGDHCGPyA2gNU9pyjRo2q8JyFhYVkZWWZPBqShIQEGwQ/qcyY8RUTJ/6Mk1NRtc6h1WqJjIy0HPwA/P3vsH69BD9CCCHsqlqzwEJDQzl48CDr16/n5MmTKIpCZGRklVaDT0lJQafTlcsVCQwM5MqVK5Uen5SUxJo1a/jf//5nsv3KlStVPue7777LG2+8YXXb65OioqIaL0ni5FTE5MlLcHMrwMsrG52uerV5LC58KoQQQtQzNaoDdPvtt3P77bfXqAFl14lSFKXitaOumz9/Pr6+vmarUlf1nC+//DLPPPOM8fusrCxCQ0MrbUN9UPMEaIWxY38jMDCZ7GxPfv55Enp99X4tfHx8THrjytm9G1q1AksJ0kIIIUQdqdZH/VmzZjFnzpxy2z/77DNmz55t1Tn8/f1xcHAo1zOTnJxseQbRdYqi8N133zFt2jScnZ1NngsKCqryOV1cXPD29jZ5NBSJiYlWBYyW9O+/iy5djqHTafn550nk5FTvtWs0Gnr16mW5LYoCd98NQUGwa1e12yuEEELYQrUCoGXLljFo0KBy2wcOHMgvv/xi1TmcnZ3p1asX69evN9m+fv16Bg4cWOGxW7du5ezZszz88MPlnhswYEC5c65bt67SczZUNZkB1rLlBW6/XX2v1q6NIiGhgt6bSiiKQrdu3SzvcOYMJCWBiwtIxXAhhBB2Vq2xjtTUVHx8fMpt9/b2JiUlxerzPPPMM0ybNo3evXszYMAAvvrqK+Lj45k5cyagDk1dvnyZBQsWmBz37bff0q9fPzp37lzunE899RRDhgzhvffeY/z48axcuZINGzawY8eOKr7KhuHSpUvVGgJzcCjmrrt+RatViI3txr59fardBo1GQ5s2bSruOTPU/+nfHypKkhZCCCHqQLV6gNq2bUtMTEy57WvWrKF169ZWn2fy5Ml8/PHHvPnmm3Tv3p1t27axevVqYx5JUlJSuZpAmZmZLFu2zGzvD6i9UIsXL+b777+na9euzJ8/nyVLltCvX78qvMKGoSYJ0DqdE7/8MpHTp9vxxx93UpOp7xVWfjaQ5S+EEELUI9WqA/Tdd9/xxBNP8PzzzzN8+HAANm7cyAcffMAnn3zCjBkzbN7QutRQ6gDFxcUxf/58ezcDNzc3nn32WRwcLFSMVhQIDoarV2HrVhgypG4bKIQQolGoyv27WkNgDz30EIWFhbzzzju89dZbALRq1Yovv/ySBx54oDqnFNVgSICuSgzbqdMxkpMDuHbNNjOxNBoN3bt3txz8AJw8qQY/rq5wE/bECSGEaHiqNQSWn5/P3/72Ny5dusTVq1c5cuQITzzxRKWzt4RtJSUlVWn/5s0TuOuuX3nkkW/w979mkzYoikKPypKaDctfDByoJkELIYQQdlatHqDx48dz9913M3PmTJycnBgxYgROTk6kpKTw4Ycf8thjj9m6ncKMhIQEq3t/PDxyiI5eioODnlOnOpCS4l/j62s0GoKDgwkICKh4x7vvBi8v8PWt8TWFEEIIW6hWD9DBgwcZPHgwAL/88guBgYHExcWxYMECs/WBhO0VFhaSkZFh1b5arY6JE3/G2zuba9f8WblyAjVJejZQFKXihU8NAgPh/vvhzjtrfE0hhBDCFqoVAOXl5eHl5QWoNXbuvvtutFot/fv3Jy4uzqYNFOZVZfjr9tvX07JlHIWFzixZMpmiItsMQzk6OtKpUyebnEsIIYSoS9WeBr9ixQoSEhJYu3atcfHR5OTkej1r6mZibQXozp2P0L//bgBWrLiL1NRKhquspNVq6dy5My6V5fSsWgX/+Y+aCC2EEELUE9UKgP71r3/x3HPP0bJlS/r168eAAQMAtTeo0oRYYRPWVYBW6Nz5GADbtw/m5MmONru+1QuffvcdvPACLF9us2sLIYQQNVWtJOiJEydyyy23kJSUZLL8wW233cZdd91ls8YJy8xVgNZo9ISFxeHllUN2tifx8eEsWTKZHj1iOXTItoGpn58fLVq0qHgnvV6t+wNSAFEIIUS9Uu3V4IOCgggKCjLZ1rdv3xo3SFSuoKCAzMxMk20REceJiorBxyfLuC0z05uYmCgOHrQiUbkKKl341ODIEUhLA09PsCZZWgghhKgj1RoCE/ZVNgE6IuI40dFL8fbOMtnu7Z1FdPRSIiKO27wNXbt2rXwnQ/2fW24BJyebt0EIIYSoLgmAGqDSCdAajZ6oqJjrX5vuZ/g+KioGjUZvk2trNBrat2+Pp6dn5TvL+l9CCCHqKQmAGqDSCdBhYXH4+GSVC34MNBrw8ckiLMw25QkURWHQoEGV76jTSf6PEEKIeqvaOUDCfkonQHt55Vh1jLX7VUSj0dCjRw9CQ0Mr3/n8eSgsBG9vkJmBQggh6hkJgBqY/Px8srJu5PpkZ1sxFFWF/Sri6urKiBEjrNu5XTtIT4ezZ8FRfs2EEELULzIE1sCUrf8THx9OZqY3lpYEUxR1Nlh8fHiNrz169Gjc3NysP8DVFTp3rvF1hRBCCFuTAKiBSUpKMpl+rihaYmKizO5rCIpiYqJQlOr/qLVaLa1ataKzBDNCCCFuEhIANTDmKkCfPBnJb7+NIz/f1WR7VpY3S5dGc/JkZI2ve+edd1q19AYA+/erPT+vvlrj6wohhBC1QZIzGhhzFaABYmN7cvhw93KVoGvS82Nw66234ufnZ/0BmzbBX39BmzY1vrYQQghRGyQAakDy8vLIzs62+LyiaImLa2Wz62k0Gpo0aWLdtPfSpP6PEEKIek6GwBoQSwugOjoW0b37IQICkgEL2dDVoCgK48aNw8HBwfqDioth+3b1awmAhBBC1FMSADUgpStAlxYcnMT48Su5//4fASvzdCphqPkTHl7F2WP790NuLvj5QZcuNmmLEEIIYWsSADUglnqAmje/DMDly81tdi0XFxduv/32qh9oWP/r1ltBK79eQggh6ie5QzUgly9fNpsAbQiAEhNtFwBVueaPgeT/CCGEaAAkAGogcnJyyMkxv5yFLXuANBoNLVu2pEt1h6/at4fwcAmAhBBC1GsSADUQSUlJZre7u+fSpEkGigKJiSE1vo5Go6lazZ+yPvsMLl6ETp1q3BYhhBCitkgA1EBYSoA29P6kpPhTWOha7vmqGjJkCE2bNq3xeSwuTy+EEELUAxIANRC1nQCt0Wjw8/Ores2f0s6cAZ2uRu0QQggh6oIUQmwgLFWA3rOnH5cuNScvz71G5zfU/HGs7srthYXQtau6AOqxY9DcdgnZQgghhK1JANQAZGdnk5eXZ/a5/Hx3zp5tX6PzazQaunXrVvWaP6Xt2QMFBeDjAyE1z0USQgghapMMgTUAlhKgbaXaNX9KM0x/HzpU8n+EEELUexIANQCJiYlozRQVbNnyAsOHbyQsLK5G54+KisLdvWZDaFL/RwghREMiQ2ANwOXLl9Hr9eW2d+hwkv799+DsXEh8fNWHr7RaLaGhoXTt2rVmDczPh1271K8lABJCCNEASA9QPacoCpcvXzb73I0ZYC2qff6xY8dWv+aPwe7dUFQEwcHQrl3NziWEEELUAQmA6rns7Gzy8/PLbddqdQQHq7lB1Z0Cb7OaP6WHvyT/RwghRAMgQ2D1nKX6P4GBV3F01JGf70pamp9xe6tz5xi9Zg1rRo/mQps2Zo/VaDT4+vrWrOZPaRMmqP/262eb8wkhhBC1TAKgei4pKQmtVlsuB8i0AOL1XhdF4baNGwlISeG2jRv5pnVrsz0yNa75U1bPnupDCCGEaCBkCKyes5QAHRJSvgJ0m3PnaH69x6h5YiJtzp0rd5yh5k/Lli1rp8FCCCFEAyABUD1WUQJ0QMA1oFQApCgM27QJ/fUeH71Gw7BNm6BM9WhnZ2dGjhxpu0b+8QesWAEZGbY7pxBCCFHLZAisHsvKyqKgoMDsc99++zB+fmlkZ3sBpr0/AFpFMfYCnWvb1rjdJjV/SnvnHXUK/HffwYMP2u68QgghRC2SHqB6zFICtEpLWpo/xcUu5Xp/DBQgas0aUBS0Wi1hYWF069bNdg3MyYF9+9Svpf6PEEKIBkQCoHrMUgXosgy9P9oyw10awD81ldvXrgVsVPOntJ07oaQEwsNBcoqEEEI0IDIEVo8lJiaaTYCOilqNu3seu3YNJCkxmGGbNqFgnAtWzsDduwlp2xZ/W9T8KU2WvxBCCNFASQ9QPVVRAnTHjifo0uUYTk7FOOh0+GRmWgx+DFr+9BN88IFtGykBkBBCiAZKeoDqqczMTAoLC8tt9/LKwts7G71eQ1JSMDpHR75+9FHGL19O64sX2dWvH0fK5PlMbNqUpmvWwIwZtmtgVhYcOKB+PXSo7c4rhBBC1AEJgOopSwnQhvo/ycnNKC52BiDLxwe/9HQATkVEcCUkBFBr/nTp0oWmd90Fb74JhsKHigJpaVCTIbHdu0Gng9atISys+ucRQggh7ECGwOopSwnQhgrQiYk3CiC65ebim5kJwJXgYON2k5o/pas+f/wxREbemMFVHbffDmfPwvz51T+HEEIIYScSANVTlipAmy6BoQpJUhdFTWnalEJXV+P2UaNG4eHhYXqC4mL46SdITlaHrn77rXoN1GigTRsYPLh6xwshhBB2JAFQPaQoioUhML3ZAKjAxYVjnTpxukMHALRaLaGhoXTv3r38KZycYMsWiIqCvDy46y747DPbvwghhBCiHpMAqB5KT0+nqKio3HZ39zxSU5tSUOBCcnKAcfvl0FCWTZrE+lJLXFRY88fLC37/HR59FPR6ePJJeOYZ9WtrrF+vBk7/+1+VXpcQQghRX0gSdD2UdH1Iq6y8PE++/vrvaLUlKIqD2X20Wi2dOnUiICDA7PNGjo7w5ZfQqhW8/DJ89BEkJMDSpWZXkDexZo26/lezZnDffVa8IiGEEKJ+kR6geqiyCtB6/Y241amoiKYpKcbeG71eT2hoqHUX0mjgpZdg0SJwdlbzeaypFC31f4QQQjRw0gNUD1lKgNZqdej1pj0/YXFx3L9wIYnBwXz9978D0KJFi6pdcMoU6NtXndJuoCjmg6G0NDh8WP1a6v8IIYRooKQHqJ6xlADt4FDMiy/+Px555CtcXG6sEG+cAebvD4CjoyOBgYFVv3Dp4CcjA269FbZvL7/ftm1qcBQRAUFBVb+OEEIIUQ9IAFTPpKWlUVxcXG57UNAVnJ2L8fHJpLDQxbg9+HqwlHS9/k9ISIhVC6hW6I031OBnxAhYssT0ORn+EkIIcROQAKiesVQB2nT6+42hqZDr+ydeD3zCbFGV+Z131FleRUXq8Nh776m9Phs2wLx56j4SAAkhhGjAJAeonjEkQJfNATJXAdo9JwefrCwU1ArQer2+6vk/5ri7w88/w3PPqVWjX3oJzp+H/fuhsFDNDRoypObXEUIIIexEeoDqmcTExCpXgE5t2pQiF3VYzCYBEICDgzo1/pNP1IDnq6/g4EH1OUWB2FjbXEcIIYSwAwmA6hFLCdCurnk0bZoGwOXLIcbtwaWGvwB8fHzKL31RU7Nmwa+/ms4Ic3CA115TAyEhhBCiAZIhsHokNTWVkpKSctubN0+8/rwfBQXuxu3nW7fGUacjMTgYrVZLeHh47TTMzc002NHp1IVU162DUaNq55pCCCFELZIAqB6xlABdWOjCX39FkpXlY7L9cmgolw1FD22V/1OWoqi9PQ4OauBjYOgFGjnSuuKJQgghRD0iAVA9YikB+tKlUH75pfLqzrUSAK1bp/b2lCW9QEIIIRowyQGqRyxVgDbHLzWVtqdP456TA9SgAGJFDL0/luoKabWSCySEEKJBkgContDr9Vy5cqXcdmfnApo0SQNMg4zOx44x9X//Y+S6dYCNCiCWVVQE8fGWV4nX69UFVM2sXC+EEELUZzIEVk9YSoBu2/Yskyb9woULLVmwYLpxe+kK0DYrgFiWi4s6zHXtmuV9mjVT9xNCCCEaELv3AH3xxRe0atUKV1dXevXqxXZz60+VUlhYyCuvvEJ4eDguLi60adOG7777zvj8/Pnz0Wg05R4FBQUVnNX+KqsAnZLib7LdUAE6KSTEdgUQzQkNhZ49LT9q67pCCCFELbJrD9CSJUuYPXs2X3zxBYMGDWLevHmMHj2a48ePW+zRiI6O5urVq3z77be0bduW5OTkcj0n3t7enDp1ymSbq6trrb0OW6isAvTlyzcCDY/sbLyzs1GApOsLktZaACSEEELchOwaAH344Yc8/PDDPPLIIwB8/PHHrF27lrlz5/Luu++W2z8mJoatW7dy/vx5/Pz8AGjZsmW5/TQaDUENbKVycwnQGo2O4OCk68+XrwCd4u9PsYtL7RRAFEIIIW5idhsCKyoq4sCBA4wcOdJk+8iRI/nzzz/NHvPbb7/Ru3dv3n//fZo3b0779u157rnnyM/PN9kvJyeH8PBwWrRowZ133smhQ4cqbEthYSFZWVkmj7pkKQG6WbNrODsXU1joTEpKU+P24DILoNZaAUQhhBDiJmW3HqCUlBR0Ol25qduBgYFmgwGA8+fPs2PHDlxdXVm+fDkpKSn84x//IC0tzZgHFBERwfz58+nSpQtZWVl88sknDBo0iMOHD9OuXTuz53333Xd54403bPsCq+DatWvoShcZvM50/a8bsWrw9R6gWs//EUIIIW5Sdp8FpilTRVhRlHLbDPR6PRqNhoULF+Ljo1ZF/vDDD5k4cSKff/45bm5u9O/fn/79+xuPGTRoED179uTTTz9lzpw5Zs/78ssv88wzzxi/z8rKIjS08sKDtlJZAnTp4S+ADSNGcDIigoTreVISAAkhhBBVY7cAyN/fHwcHh3K9PcnJyRYL+gUHB9O8eXNj8APQsWNHFEXh0qVLZnt4tFotffr04cyZMxbb4uLigosdp3InJSWZTYA+erQzeXlunDlj+rpSAwJIDQgAaqkAohBCCHGTs1sOkLOzM7169WL9+vUm29evX8/AgQPNHjNo0CASExPJuV79GOD06dNotVqLvSCKohAbG0twcLDtGm9jly5dMlsB+uLF1mzceDsJCZZzfGqlAKIQQghxk7PrnfOZZ57hm2++4bvvvuPEiRM8/fTTxMfHM3PmTEAdmnrggQeM+9933300bdqUBx98kOPHj7Nt2zaef/55HnroIdzc3AB44403WLt2LefPnyc2NpaHH36Y2NhY4znrG51Ox9WrV63ev8PJk/TdvZumKSm1VwBRCCGEuMnZNQdo8uTJpKam8uabb5KUlETnzp1ZvXq1cVZTUlIS8fHxxv09PT1Zv349Tz75JL1796Zp06ZER0fz9ttvG/fJyMjg0Ucf5cqVK/j4+NCjRw+2bdtG37596/z1WePatWtme39atEjAxaWQy5ebU1DgZtze4+BBOpw+zZqoKPb6+0v+jxBCCFENGkWRlSzLysrKwsfHh8zMTLy9vWv1WgcPHuT3338vt33ixKV06nScDRtuY+fOwcbtz3zwAV45OXz30EMkhIXx3HPPSQ0gIYQQgqrdvyV5xM4MFaDLMjcDzDM7G6+cHPQaDVeCgqQAohBCCFFNEgDZmbkK0B4eOfj6ZqIokJgYYtxuKICY4u+PztVVCiAKIYQQ1SQBkB3pdDqSk5PLbQ8JUXt/rl0LoKjoxhpmIaUqQEsBRCGEEKL6JACyo+TkZAsJ0OYLIJauAA3UabFGIYQQ4mYiAZAdWaoAbegBSkw0HwAlhoTg6OhIs2bNareBQgghxE3K7kthNGaGBGjTXiDF4hIYnz3xBEFXrpAUFERzKYAohBBCVJsEQHZkqQL0/PnTadHiMlevmvbwFLu4kBAeLgUQhRBCiBqSAMhOSkpKSElJMfOMhuTkIJKTgyweKwnQQgghRM1IAGQnlhKgLRmxfj2OJSXs692b1IAACYCEEEKIGpAAyE4sJUDfeusWcnPd+euvTuTnXy9yqCh0PXwYr5wcjnbuLAUQhRBCiBqSAMhOzCVAa7Ul3HLLdhwddZw719YYAHmVqgB9LSSECCmAKIQQQtSITCOyE3MJ0IGBV3F01JGX50Z6ehPjdkMF6GsBARQ5OMjwlxBCCFFDEgDZQXFxsdkEaNPp7xrj9tIVoEEKIAohhBA1JQGQHVy9ehVFUcptNwRAlgogJkkBRCGEEMImJACyA0sJ0GYLICrKjR6g4GCaN28uBRCFEEKIGpI7qR0kJSWVC2JcXAoICFCHxS5fvrECvGt+Pnqt1pgALcNfQgghRM3JLDA7MJcAHRCQjKJARoYveXmexu0F7u589OyzuOXmUuTgIAGQEEIIYQMSANWx4uJiUlNTy22/dCmM//f/XsLHJ8vscfnX6/40b97c7PNCCCGEsJ4EQHXsypUrZhOgAYqKXLl2zdXisVIAUQghhLANCYDqWFpamvU7Kwoz584l08eHP8aPJ7xr19prmBBCCNGISABUx8z1/nh5ZXHPPb9w6VIoGzaMwFADyDsri8DkZPxTUshzcZECiEIIIYSNyCyweqB580uEh8fTps1ZzBVAvBYQQImTkyRACyGEEDYiAVA9cKP+j2kPj2EJDCmAKIQQQtiWDIHVA2YLIHKjAnRiSIgUQBRW0el0FBcX27sZQghRa5ydnW1yP5QAyM40Gj0hIWpPj6UK0FeaN6elDH+JCiiKwpUrV8jIyLB3U4QQolZptVpatWqFs7Nzjc4jAZCd+fun4OJSRFGRE9euBRi3e2dm4pGXh06rJSkggMESAIkKGIKfZs2a4e7ujkajqfwgIYRoYPR6PYmJiSQlJREWFlajv3USANnZjQVQQ1CUG116roWFxIWFAaBzcpIZYMIinU5nDH6aNm1q7+YIIUStCggIIDExkZKSEpycnKp9HgmA7MzRsZicHI9y+T/JgYHMf+ghAHx9fXF3d7dH80QDYMj5kd8RIURjYBj60ul0EgA1ZPv392X//j44OOjMPq/VagkPD6/jVomGSIa9hBCNga3+1sm0onpBg05XKhZVFJwLCgB1vFPW/xJCCCFsSwIgO9JodED5ytA+mZm8/P/+H499/jno9VIAUdQRHbAFWHT9X/O9krYydOhQZs+eXaVjFEXh0Ucfxc/PD41GQ2xsbKXHbNmyBY1GU+EMufnz5+Pr62vx+YsXL1p9vYakZcuWfPzxx8bvNRoNK1assLi/Ld+Hyq5lS99++y0jR4606TnL/v6WfS/NsdVrtuZaDdVzzz3HrFmz6uRaMgRmR3377mXQoJ3s3duXHTuGGLcbCiDqHBxwdHaWAoiiDvwKPAVcKrWtBfAJcLddWmROTEwM8+fPZ8uWLbRu3Rp/f397N+mmkpSURJMmTW6qaxUWFvKvf/2LxYsX1+p19u3bV2eLVdfltWxh+vTpZGRkmAR/Fy9epFWrVhw6dIju3bsbt7/wwgu0adOGp59+mlatWtVqu6QHyI6aN7+Ml1cOGo1pL1BIqQrQUgBR1L5fgYmYBj8Al69v/7XOW2TJuXPnCA4OZuDAgQQFBeHoWH8/wzXEgpRBQUG4uLjU6jWKiorq7FoAy5Ytw9PTk8GDB9fqdQICAmp9IoLhvauLa9lLs2bNGDlyJF9++WWtX0vurHZUWQXopObNZfhL1EBuBY+C6/voUHt+yg/F3tj2FKbDYZbOWTMxMTH4+PiwYMECs89Pnz6dJ598kvj4eDQaDS1btgTUT/izZs2iWbNmuLq6csstt7Bv374KrzV//nzCwsJwd3fnrrvuIjU1tUpt1ev1zJgxg/bt2xMXFweowxtffvkl48ePx8PDg7fffpu2bdvywQcfmBx77NgxtFot586dA+D1118nLCwMFxcXQkJCTLr/v/jiC9q1a4erqyuBgYFMnDgRgHnz5tG8eXP0er3JuceNG8ff/vY3QA0Wx48fT2BgIJ6envTp04cNGzZU+LrKDtHs3buXHj164OrqSu/evTl06FC5Y44fP86YMWPw9PQkMDCQadOmkZKSYnx+6NChPPHEEzzzzDP4+/tz++23l7uWYWjt119/ZdiwYbi7u9OtWzd27dplcq2vv/6a0NBQ48/tww8/rHDoEmDx4sWMGzfO+P3atWtxdXUtNyQ6a9Ysbr31VgBSU1O59957adGiBe7u7nTp0oVFixZVeJ2yw1JnzpxhyJAhuLq6EhkZyfr168sdc/nyZSZPnkyTJk1o2rQp48eP5+LFi8bnp0+fzoQJE3j33XcJCQmhffv2Zq+l0Wj45ptvuOuuu3B3d6ddu3b89ttvJtf67bffaNeuHW5ubgwbNowffvjBZGg4Li6OsWPH0qRJEzw8POjUqROrV68GID09nalTpxIQEICbmxvt2rXj+++/t+p1vP766/zwww+sXLkSjUaDRqNhy5Ytxt6dHj16oNFoGDp0qPF848aNq/T9tgUJgOzEzS0PP790ABITzVeAvhwUJAGQqAHPCh73XN9nO+V7fkpTrj+/vdS2lhbOWX2LFy8mOjqaBQsW8MADD5jd55NPPuHNN9+kRYsWJCUlGYOcF154gWXLlvHDDz9w8OBB2rZty6hRo0hLSzN7nj179vDQQw/xj3/8g9jYWIYNG8bbb79tdVuLioqIjo5m//797Nixw2SW5r///W/Gjx/P0aNHeeihh3jooYdMbhQA3333HYMHD6ZNmzb88ssvfPTRR8ybN48zZ86wYsUKunTpAsD+/fuZNWsWb775JqdOnSImJoYhQ9Sh8kmTJpGSksLmzZuN501PT2ft2rVMnToVgJycHMaMGcOGDRs4dOgQo0aNYuzYscTHx1v1OnNzc7nzzjvp0KEDBw4c4PXXX+e5554z2ScpKYlbb72V7t27s3//fmJiYrh69SrR0dEm+/3www84Ojqyc+dO5s2bZ/Gar7zyCs899xyxsbG0b9+ee++9l5KSEgB27tzJzJkzeeqpp4iNjeX222/nnXfeqfR1bN++nd69exu/HzFiBL6+vixbtsy4TafTsXTpUuN7V1BQQK9evVi1ahXHjh3j0UcfZdq0aezZs6fyNw41QL777rtxcHBg9+7dfPnll7z44osm++Tl5TFs2DA8PT3Ztm0bO3bswNPTk6ioKGNPD8DGjRs5ceIE69evZ9WqVRav+cYbbxAdHc2RI0cYM2YMU6dONf4fuHjxIhMnTmTChAnExsby97//nVdeecXk+Mcff5zCwkK2bdvG0aNHee+99/D0VP9fv/baaxw/fpw1a9Zw4sQJ5s6daxx+rux1PPfcc0RHRxMVFUVSUhJJSUkMHDiQvXv3ArBhwwaSkpL49dcbPc19+/YlISHB+OGi1iiinMzMTAVQMjMzbX7uQ4cOKa+//rry449TFUVBuXatqfL6668bHx899ZSigFKi1Spvvfqqkpuba/M2iJtLfn6+cvz4cSU/P7/MM1TwGHN9n/9Vsp/h8b9S5/W3sE/V3HrrrcpTTz2lfP7554qPj4+yadOmSo/56KOPlPDwcOP3OTk5ipOTk7Jw4ULjtqKiIiUkJER5//33FUVRlM2bNyuAkp6eriiKotx7771KVFSUyXknT56s+Pj4WLzuhQsXFEDZvn27MmLECGXQoEFKRkaGyT6AMnv2bJNtiYmJioODg7Jnzx5j2wICApT58+criqIo//3vf5X27dsrRUVF5a65bNkyxdvbW8nKyjLbpnHjxikPPfSQ8ft58+YpQUFBSklJicXXERkZqXz66afG78PDw5WPPvrI5DUsX77ceD4/Pz+Tv0Fz585VAOXQoUOKoijKa6+9powcOdLkGgkJCQqgnDp1SlEU9efcvXv3cm0pfS3D+/vNN98Yn//rr78UQDlx4oSiKOrP6I477jA5x9SpUyv8uaWnpyuAsm3bNpPts2bNUoYPH278fu3atYqzs7OSlpZm8VxjxoxRnn32WeP3ht9fg9Lv5dq1axUHBwclISHB+PyaNWtMXvO3336rdOjQQdHr9cZ9CgsLFTc3N2Xt2rWKoijK3/72NyUwMFApLCw0aYu5n9urr75q/D4nJ0fRaDTKmjVrFEVRlBdffFHp3LmzyTleeeUVk/8XXbp0UV5//XWzr33s2LHKgw8+aPY5a1/H+PHjTY4z/MwNv0ulGe7BW7ZsMXtNy3/zqnb/lh4gO2nRwvzwV8j14a+rgYF4+fvftOO8oi7kVPAwfPoNtvJcpfe7aOGcVbds2TJmz57NunXrGDZsmHH79u3b8fT0ND4WLlxo9vhz585RXFzMoEGDjNucnJzo27cvJ06cMHvMiRMnGDBggMm2st9bcu+995KTk8O6devw8fEp93zpngaA4OBg7rjjDr777jsAVq1aRUFBAZMmTQLUnpz8/Hxat27NjBkzWL58ubHH4/bbbyc8PJzWrVszbdo0Fi5cSF5envHcU6dOZdmyZRQWFgKwcOFCpkyZgoODA6D24LzwwgtERkbi6+uLp6cnJ0+etLoH6MSJE3Tr1s3kb1DZ9+nAgQNs3rzZ5GcVEREBYBziM/e+WNK1a1eT9w4gOTkZgFOnTtG3b1+T/ct+X1Z+fj4Arq6uJtunTp3Kli1bSLze275w4ULGjBljTMrW6XS88847dO3alaZNm+Lp6cm6deuq9N6FhYWZVPA3996dPXsWLy8v43vn5+dHQUGByXvXpUsXq9a8Kv3eeXh44OXlZfLe9enTx2T/su/drFmzePvttxk0aBD//ve/OXLkiPG5xx57jMWLF9O9e3deeOEF/vzzzyq/jqpwc3MDMPl9rw0SANlJSIhhCQzTACjLy4tDPXpwKjJSCiCKGvKo4GG4IQxGne1lqbCYBgi9vl9l56267t27ExAQwPfff4+i3MhD6t27N7GxscZH6RyO0gzHlC2MpiiKxWJppa9TVWPGjOHIkSPs3r3b7PPmZuY88sgjLF68mPz8fL7//nsmT55sDCpCQ0M5deoUn3/+OW5ubvzjH/9gyJAhFBcX4+XlxcGDB1m0aBHBwcH861//olu3bsacjbFjx6LX6/njjz9ISEhg+/bt3H///cbrPv/88yxbtox33nmH7du3ExsbS5cuXUyGVypizfuk1+sZO3asyc8qNjbWmP9S0ftiTumqvoafnyHPydzPtLI2Nm3aFI1GQ3p6usn2vn370qZNG+PPZfny5Sbv3X//+18++ugjXnjhBTZt2kRsbCyjRo2q0XtXtu16vZ5evXqVe+9Onz7NfffdZ9yvOu+d4XpVee8eeeQRzp8/z7Rp0zh69Ci9e/fm008/BWD06NHExcUxe/ZsEhMTue2224zDoda+jqowDN0FBARUsmfNSABkJ4mJzYmPDyUhwTTH53JoKL+NH8+2wYNl/S9RBxxQp7pD+SDI8P3H1/ezvTZt2rB582ZWrlzJk08+adzu5uZG27ZtjQ8vLy+zx7dt2xZnZ2d27Nhh3FZcXMz+/fvp2LGj2WMiIyPLBTCWApqyHnvsMf7f//t/jBs3jq1bt1p1zJgxY/Dw8GDu3LmsWbOGh64vcWPg5ubGuHHjmDNnDlu2bGHXrl0cPXoUAEdHR0aMGMH777/PkSNHuHjxIps2bTIed/fdd7Nw4UIWLVpE+/bt6dWrl/G827dvZ/r06dx111106dKFoKAgkwTbykRGRnL48GFjLwqUf5969uzJX3/9RcuWLU1+Xm3btrX5NO2IiAhj3ojB/v37KzzG2dmZyMhIjh8/Xu65++67j4ULF/L777+j1Wq54447jM9t376d8ePHc//999OtWzdat27NmTNnrG5rZGQk8fHxxh4moFxCd8+ePTlz5gzNmjUr996Z612siYiIiHITA8y9d6GhocycOZNff/2VZ599lq+//tr4XEBAANOnT+enn37i448/5quvvrL6dTg7O6PTmdYVK72cRVnHjh3DycmJTp061eyFV0ICIDvZunUo33//MElJIRb3kQBI1I27gV+AshXHW1zfXrt1gNq3b8/mzZuNw2FV4eHhwWOPPcbzzz9PTEwMx48fZ8aMGeTl5fHwww+bPWbWrFnExMTw/vvvc/r0aT777DNiYmKsvuaTTz7J22+/zZ133mkSeFni4ODA9OnTefnll2nbtq3JUMj8+fP59ttvOXbsGOfPn+fHH3/Ezc2N8PBwVq1axZw5c4iNjSUuLo4FCxag1+vp0KGD8fipU6fyxx9/8N1335n0YIAaHP7666/ExsZy+PBh7rvvvnKzxipy3333odVqefjhhzl+/DirV68uN6Pt8ccfJy0tjXvvvZe9e/dy/vx51q1bx0MPPWT2xlYTTz75JKtXr+bDDz/kzJkzzJs3jzVr1lS6LMKoUaPM/pymTp3KwYMHeeedd5g4caLJMFnbtm1Zv349f/75JydOnODvf/87V65csbqtI0aMoEOHDjzwwAMcPnyY7du3l0s6njp1Kv7+/owfP57t27dz4cIFtm7dylNPPcWlSxVNTKi6v//975w8eZIXX3yR06dPs3TpUubPnw/c6JmaPXs2a9eu5cKFCxw8eJBNmzYZP0T861//YuXKlZw9e5a//vqLVatWGZ+z5nW0bNmSI0eOcOrUKVJSUiguLqZZs2a4ubkZE+czMzON7d2+fTuDBw82DoXVFgmA6hGX/HwCk5LQ6nQ4OjpKAURRh+5Gze3ZDPzv+r8XqKsiiB06dGDTpk0sWrSIZ599tkrH/r//9/+45557mDZtGj179uTs2bOsXbvWYpG9/v3788033/Dpp5/SvXt31q1bx6uvvlqla86ePZs33niDMWPGmORDWPLwww9TVFRUrvfH19eXr7/+mkGDBtG1a1c2btzI77//TtOmTfH19eXXX39l+PDhdOzYkS+//JJFixaZfCoePnw4fn5+nDp1qtxww0cffUSTJk0YOHAgY8eOZdSoUfTs2dPq1+jp6cnvv//O8ePH6dGjB6+88grvvfeeyT4hISHs3LkTnU7HqFGj6Ny5M0899RQ+Pj42r182aNAgvvzySz788EO6detGTEwMTz/9dLn8nrJmzJjB6tWrTW6wAO3ataNPnz4cOXLEOPvL4LXXXqNnz56MGjWKoUOHEhQUxIQJE6xuq1arZfny5RQWFtK3b18eeeSRcjPW3N3d2bZtG2FhYdx999107NiRhx56iPz8fLy9va2+ljVatWrFL7/8wq+//krXrl2ZO3euMSAz1GLS6XQ8/vjjdOzYkaioKDp06MAXX3wBqL01L7/8Ml27dmXIkCE4ODgYC0ta8zpmzJhBhw4d6N27NwEBAezcuRNHR0fmzJnDvHnzCAkJYfz48cb2Llq0iBkzZtj0PTBHo9RkQPwmlZWVhY+PD5mZmTb/RYyNjWXr1u/JzfWguNi0CFjHv/4i+uefudS8ORvefpvp06fb9Nri5lRQUMCFCxdo1apVpTcDYR87d+5k6NChXLp0icDAQHs356YxY8YMTp48yfbt2yvcLzo6mh49evDyyy/XUcvqv3feeYcvv/yShIQEezfFxB9//MHzzz/PkSNHLBY6rehvXlXu3/W3jOpN7K67ltOixSWWLo3m1KkbeQrGGWDBwVL/R4ibQGFhIQkJCbz22mtER0dL8FNDH3zwAbfffjseHh6sWbOGH374wdhLUZH//Oc/5QoDNjZffPEFffr0oWnTpuzcuZP//Oc/PPHEE/ZuVjm5ubl8//33dVLlXQKgOldMcHASWq1CSopphrthDbDEoCA6SAAkRIO3aNEiHn74Ybp3786PP/5o7+Y0eHv37uX9998nOzub1q1bM2fOHB555JFKjwsPDzdJsm+Mzpw5w9tvv01aWhphYWE8++yz9bJHrGwRzdokAVCd0uHn9ztOTiUUFTmRluZ746lSFaATQ0K4TRKghWjwpk+fLkPZNrR06VJ7N6HB+uijj/joo4/s3Yx6RZKg68yvQEvCwt4CwNm5mKeemkNEhDo90zc9HbeCAkocHChq314KIAohhBC1SAKgOmF+tW1v7yyio5cSEXHcmP+THBhIaJs2dd9EIYQQohGRAKjWWV5t21C+IioqhpDE65Whg4Ol/o8QQghRyyQHqNZVvNq2RgM+Plmk3daEja63cTkkhJESAAkhhBC1SgKgWpdk1V5FXV05qO0jBRCFEEKIOiBDYLXOutW2s7M9AWjevLnNK6gKIYQQwpTcaWtdxattKwpk7/TAY00uPtnZhIWF1WnrhBB1Y9q0afzf//2f8fuWLVvy8ccfG7/XaDSsWLHC4vFl96+p+fPn4+vra7Pz1QcXL15Eo9EQGxsLwJYtW9BoNGRkZFg8xlbvgzXXqo6JEyfy4Ycf2vScQiUBUK2zvNq2YRGSS++1YNLSX7hl61ZJgBb2tWEDREaq/9YijUZT4aMmtXNsHSgADB06tNxCrVW54R05coQ//vjDpBjfvn37ePTRR23aTmFq4MCBJCUl2Xx19bq81r/+9S/eeecdsrKybHpeIQFQHTG/2nZWljdLl0bjcqwIkBlgws4UBf75TzhxQv23FpcJTEpKMj4+/vhjvL29TbZ98sknlZ+kAfnss8+YNGkSXl5exm0BAQENqt6XoiiUlJTYuxlV4uzsTFBQUKUrxtdUcXFxrV2ra9eutGzZkoULF9r0vEICoDqkrrZ99uzXLFt2D/Pn/41PPpnNyRMdjRWgszt0aFB/EEU9l5tr+VFQUH7f336DffvU7/ftU7/PzYX8fOvOWwVBQUHGh4+PDxqNxmTbtm3b6NWrF66urrRu3Zo33njD5Ob7+uuvExYWhouLCyEhIcyaNQtQe2ri4uJ4+umnjb1JAHFxcYwdO5YmTZrg4eFBp06dWL16tfF8x48fZ8yYMXh6ehIYGMi0adNISUkB1GrOW7du5ZNPPjGe8+LFiwwbNgyAJk2aVNhrpdfr+fnnnxk3bpzJ9sp6qt58800CAwONwzllff/99/j4+LB+/fpKX4M1UlNT6du3L+PGjaOgoMDYw7V27Vp69+6Ni4sLP/74I1qtlv3795sc++mnnxIeHo6iKKSnpzN16lQCAgJwc3OjXbt2fP/99wAUFRXxxBNPEBwcjKurKy1btuTdd98F4N5772XKlCkm5y0uLsbf3994fExMDLfccgu+vr40bdqUO++8k3Pnzll8TeZ66ebPn09YWBju7u7cddddpKamljvu999/r/D3T6PR8OWXXzJ+/Hg8PDx4++23y13LMLS2du1aOnbsiKenJ1FRUSQl3ZgYU1JSwqxZs4yv58UXX+Rvf/tbuZXnx40bx6JFiyy+TlE9EgDVKQdycnpz7FgX4uJaoShamqSl4VpYSImjI579+tm7geJm4ulp+XHPPab7BgRAmT+6TJig7jt6tOn2li3Nn9NG1q5dy/3338+sWbM4fvw48+bNY/78+bzzzjsA/PLLL3z00UfMmzePM2fOsGLFCrp06QLAr7/+SosWLXjzzTeNvUkAjz/+OIWFhWzbto2jR4/y3nvv4Xm9zUlJSdx66610796d/fv3ExMTw9WrV41rEn3yyScMGDCAGTNmGM8ZGhrKsmXLADh16lSFvVZHjhwhIyOD3r17W/X6FUXhqaee4ttvv2XHjh1079693D4ffPABzz33HGvXruX222+v9DVU5tKlSwwePJiIiAh+/fVXkxW2X3jhBd59911OnDjBuHHjGDFihDEgMfj++++ZPn06Go2G1157jePHj7NmzRpOnDjB3Llz8ff3B2DOnDn89ttvLF26lFOnTvHTTz/RsmVLAKZOncpvv/1GTk6O8bxr164lNzeXe67/vubm5vLMM8+wb98+Nm7ciFar5a677kKv11v1Ovfs2cNDDz3EP/7xD2JjYxk2bBhvv/22yT6V/f4Z/Pvf/2b8+PEcPXqUhx56yOz18vLy+OCDD/jxxx/Ztm0b8fHxPPfcc8bn33vvPRYuXMj333/Pzp07ycrKMpsH1rdvX/bu3UthYaFVr1NYSRHlZGZmKoCSmZlp83MfOnRIef31142PnydOVBRQLoWEKPv27bP59cTNLz8/Xzl+/LiSn59v+oQ6iGX+MWaM6b4uLpb3vfVW0339/c3vV03ff/+94uPjY/x+8ODByv/93/+Z7PPjjz8qwcHBiqIoyn//+1+lffv2SlFRkdnzhYeHKx999JHJti5duiivv/662f1fe+01ZeTIkSbbEhISFEA5deqUoiiKcuuttypPPfWUyT6bN29WACU9Pb3C17d8+XLFwcFB0ev1FbYTUH7++Wfl/vvvVyIiIpSEhASz+7/00ktKcHCwcuTIkSq9hrIM7/upU6eUsLAw5cknnzRpo+H1rVixwuS4JUuWKE2aNFEKCgoURVGU2NhYRaPRKBcuXFAURVHGjh2rPPjgg2av+eSTTyrDhw8v914oiqIUFRUp/v7+yoIFC4zb7r33XmXSpElmz6UoipKcnKwAytGjRxVFUZQLFy4ogHLo0CGT12D4Gd17771KVFSUyTkmT55cpd8/RVF/VrNnzzbZp+y1vv/+ewVQzp49a9zn888/VwIDA43fBwYGKv/5z3+M35eUlChhYWHK+PHjTc59+PBhBVAuXrxo8b1oTCz+zVOqdv+WHiA7Cy61AKrk/wibysmx/LjeewGo4UunTuDgYHq8gwP07AmlhooAuHjR/Dlt5MCBA7z55pt4enoaH4bel7y8PCZNmkR+fj6tW7dmxowZLF++vNLclFmzZvH2228zaNAg/v3vf3PkyBGT623evNnkehEREQAVDq9YKz8/HxcXF6tyQ55++ml27drF9u3bzf49+O9//8u8efPYsWOHsderJq8hPz+fW265hQkTJjBnzhyzbSzbczVhwgQcHR1Zvnw5AN999x3Dhg0z9uQ89thjLF68mO7du/PCCy/w559/Go+dPn06sbGxdOjQgVmzZrFu3Trjc05OTkyaNMmY65Kbm8vKlSuZOnWqcZ9z585x33330bp1a7y9vWnVqhUA8fHxFl9jaSdOnGDAgAEm28p+X9nvn6X3xRx3d3falFraKDg4mOTkZAAyMzO5evUqffv2NT7v4OBAr169yp3Hzc0NwOT6ouYkALIzQ/7P1RYtpACisC0PD8uPUkMcrFsHBw+CTmd6vE6nbt++3brz2oher+eNN94gNjbW+Dh69ChnzpzB1dWV0NBQTp06xeeff46bmxv/+Mc/GDJkCMXFxRbP+cgjj3D+/HmmTZvG0aNH6d27N59++qnxemPHjjW5XmxsLGfOnGHIkCE1fj3+/v7k5eVRVFRU6b633347ly9fZu3atWafHzx4MDqdrtyq6NV9DS4uLowYMYI//viDS5fMV6z3KPOzdXZ2Ztq0aXz//fcUFRXxv//9z2QIaPTo0cTFxTF79mwSExO57bbbjMM+PXv25MKFC7z11lvk5+cTHR3NxIkTjcdOnTqVDRs2kJyczIoVK3B1dWV0qSHYsWPHkpqaytdff82ePXvYs2cPgFXvLajDi5Wp7PfP0vtijpOTk8n3Go2mXBvKBp3m2piWlgaoifPCdqQStJ39Nn48IZcvo/TvLwUQRd1TFHjtNdBqwVwehVarPj9y5I3F62pZz549OXXqFG3btrW4j5ubG+PGjWPcuHE8/vjjREREcPToUXr27ImzszO6ssEcEBoaysyZM5k5cyYvv/wyX3/9NU8++SQ9e/Zk2bJltGzZEkdH838SzZ3T2dkZwOy1SjPk8Bw/ftxsPk9p48aNY+zYsdx33304ODiUSwru27cvTz75JKNGjcLBwYHnn38ewKrXYI5Wq+XHH3/kvvvuY/jw4WzZsoWQkJBKj3vkkUfo3LkzX3zxBcXFxdx9990mzwcEBDB9+nSmT5/O4MGDef755/nggw8A8Pb2ZvLkyUyePJmJEycSFRVFWloafn5+DBw4kNDQUJYsWcKaNWuYNGmS8X1OTU3lxIkTzJs3j8GDBwOwY8cOq18rQGRkJLt37zbZVvZ7a37/bMHHx4fAwED27t1rfD06nY5Dhw6V+z05duwYLVq0MOZSCduQAMjOMpo0IatpUwb16GHvpojGqKgI4uPNBz+gbk9IUPdzcamTJv3rX//izjvvJDQ0lEmTJqHVajly5AhHjx7l7bffZv78+eh0Ovr164e7uzs//vgjbm5uhIeHA+rsqm3btjFlyhRcXFzw9/dn9uzZjB49mvbt25Oens6mTZvo2LEjoCZIf/3119x77708//zz+Pv7c/bsWRYvXszXX3+Ng4MDLVu2ZM+ePVy8eBFPT0/8/PwIDw9Ho9GwatUqxowZg5ubmzGxurSAgAB69uxpMaG5rLvuuosff/yRadOm4ejoaNJDAuqQzZo1a4iKisLR0ZGnn37aqtdgiYODAwsXLuTee+81BkFBQUEVtrFjx47079+fF198kYceesg4RGP4+fXq1YtOnTpRWFjIqlWrjO/1Rx99RHBwMN27d0er1fLzzz8TFBRkLESo0Wi47777+PLLLzl9+jSbN282nrdJkyY0bdqUr776iuDgYOLj43nppZcqfT9LmzVrFgMHDuT9999nwoQJrFu3jpiYGJN9Kvv9s6Unn3ySd999l7Zt2xIREcGnn35Kenp6uV6h7du3M3LkSJteW8gQWL2g1+sl/0fYh4uLOuX9wAHLj3376iz4ARg1ahSrVq1i/fr19OnTh/79+/Phhx8aAxxfX1++/vprBg0aRNeuXdm4cSO///47TZs2BdTp4xcvXqRNmzbGIQOdTsfjjz9Ox44diYqKokOHDnzxxRcAhISEsHPnTnQ6HaNGjaJz58489dRT+Pj4GHtln3vuORwcHIiMjCQgIID4+HiaN2/OG2+8wUsvvURgYCBPPPGExdf06KOPVqmOy8SJE/nhhx+YNm0av/76a7nnBw0axB9//MFrr73GnDlzrHoNFXF0dGTRokV06tSJ4cOHG/NUKvLwww9TVFRUbgaUs7MzL7/8Ml27dmXIkCE4ODiwePFiADw9PXnvvffo3bs3ffr04eLFi6xevdqkjVOnTuX48eM0b96cQYMGGbdrtVoWL17MgQMH6Ny5M08//TT/+c9/Km1naf379+ebb77h008/pXv37qxbt45XX33VZJ/Kfv9s6cUXX+Tee+/lgQceYMCAAXh6ejJq1CiTobaCggKWL1/OjBkzbH79xk6jWDMo2shkZWXh4+NDZmYm3t7eNj13bGwsK1euBKDL4cN4Z2VxqkMHHvzPf6QGkKiWgoICLly4QKtWrUz+cIr6o6CggA4dOrB48eJySbcN1TvvvMPixYs5evSovZty09Dr9XTs2JHo6GjeeustAD7//HNWrlxpkjDe2FX0N68q928ZArOjHocO0eriRZQGVhFWCFE1rq6uLFiwoEqFCeurnJwcTpw4waeffmq8SYvqiYuLY926ddx6660UFhby2WefceHCBe677z7jPk5OTsaEfWFbdh8C++KLL4xRXK9evdhedsZJGYWFhbzyyiuEh4fj4uJCmzZt+O6770z2WbZsGZGRkbi4uBAZGWmcrlmv6PUEXy/S5lBqGqQQ4uZ06623MnbsWHs3o8aeeOIJbrnlFm699VaLBQCFdbRaLfPnz6dPnz4MGjSIo0ePsmHDBmPOFKjDpx06dLBjK29edu0BWrJkCbNnz+aLL75g0KBBzJs3j9GjR3P8+HGLq6JHR0dz9epVvv32W9q2bUtycrJJDZBdu3YxefJk3nrrLe666y6WL19OdHQ0O3bsoF89qrTsl56Oa2EhxY6O+NwkXeJCiJvf/PnzmT9/vr2bcVMIDQ1l586d9m5Go2XXHKB+/frRs2dP5s6da9zWsWNHJkyYYFwfprSYmBimTJnC+fPn8fPzM3vOyZMnk5WVxZo1a4zboqKiaNKkidVrqdRFDlDno0e5Z9kyLjVvjuP+/ZXOuhDCEskBEkI0JrbKAbLbEFhRUREHDhwoN7Vv5MiRJpVDS/vtt9/o3bs377//Ps2bN6d9+/Y899xz5JdarHHXrl3lzjlq1CiL5wR1WC0rK8vkUdsMFaCvSAFEIYQQos7ZbQgsJSUFnU5HYGCgyfbAwECuXLli9pjz58+zY8cOXF1dWb58OSkpKfzjH/8gLS3NmAd05cqVKp0T4N133+WNN96o4SuqGkP+T0GnTlIAUQghhKhjdr/zmisDbmnNHL1ej0ajYeHChfTt25cxY8bw4YcfMn/+fJNeoKqcE+Dll18mMzPT+EhISKjBK7KCotDsep0Np/79a/daQgghhCjHbj1A/v7+/P/27j0u5nz/A/hrmukyXYztPiPdRCFS2qVYYSmXLXdhJWLPcRTF0XFtheO6Wy7H0dImYR3WurO6sOSyP7qQjXpUq3YLkcWG2nSZz++PtllT0bTzbZJ5Px+PHur7/c7r+zGPPvN99/lePnw+v8HITElJSYMRnDpisRgdOnSASCSSLevatSsYY7h79y46d+4Mc3PzZmUCtfPhaKvwQW/g8RC5YAFMS0ow6JUHfRFCCCFENVptBEhLSwu9e/dGUlKS3PKkpCS4u7s3+pp+/frh/v37ePHKzNO5ubnQ0NCQPUnZzc2tQWZiYuJrM1uLVCDAA4kEFn/MoEwIIYQQ1WnV2+AXLFgAPz8/uLq6ws3NDTt37kRhYSFmz54NoPbU1L1797Bnzx4AwJQpU7B69WrMmDEDK1euxK+//orQ0FC5uWiCg4MxYMAAbNiwAaNGjcLx48dx9uzZZk+apwrt27enByCSFlNaWory8nKV7U9XV1dudLYlWVtbIyQkBCEhIQBqT3sfPXoUo0ePVmh7Lly4cAGDBg3C06dPZXNZEULajlYtgHx9ffH48WOsWrUKxcXFcHR0xHfffSebc6W4uBiFhYWy7fX19ZGUlIS5c+fC1dUVRkZGmDhxotwEde7u7jhw4ACWL1+OsLAwdOrUCQcPHnyrngH08YkT4PF4eDJjRms3hbyjSktLsW3bNrlnZLU0gUCAoKAglRRBqamp0NPTa/H9qEJTxRshpGW0+lQYc+bMwZw5cxpd19jDthwcHBqc4qpv/PjxDWZQfmtIpXC8dQvalZW4vXBha7eGvKPKy8tVWvwAQHV1NcrLy1VSANVNctpaKisrW3X/hBDltfpdYOpGu7AQ2pWVqBIIYPzhh63dHEJajbW1NTZv3iy3rFevXggPDwcAhIeHw9LSEtra2pBIJJg3b94bX/uqVatWwczMDBkZGY2uj42NhUgkkv0xlZWVhREjRkBfXx9mZmbw8/OTm7dr4MCBCAoKwoIFC2BsbIyhQ4fK1qWnp8PV1RW6urpwd3dHTk6O3L5OnjyJ3r17Q0dHB7a2tli5cqWsOLX+4xrAMWPGgMfjyX4mhLQ8KoBUTDcrCwDwUCKBiVjcyq0h5O307bffYtOmTdixYwfy8vJw7Ngx9OjRo8nXMcYQHByMmJgYXL58Gb169WqwzRdffIGFCxciISEBQ4cORXFxMTw8PNCrVy+kpaUhPj4eDx8+xMSJE+VeFxcXB4FAgCtXrmDHjh2y5cuWLUNERATS0tIgEAjk5sdKSEjA1KlTMW/ePGRlZWHHjh3YvXs31qxZA6D2VB5QW5AVFxfLfiaEtLxWPwWmbtqdOwcAqDEyogcgEvIahYWFMDc3x5AhQ6CpqQlLS0t80MSkwdXV1Zg2bRrS0tJw5coV2Z2hr1qyZAni4uJw4cIFWUEVFRUFFxcXrF27Vrbdrl270LFjR+Tm5qJLly4AADs7O2zcuFG2Td3jNtasWQMPDw8AwOLFizFy5EhUVFRAR0cHa9asweLFi+Hv7w8AsLW1xerVq/Gvf/0LK1askJ3Ka9++PU2HQ4iKUQGkSoyh3f/9HwDAtLgYYAx4wwMaCVFXEyZMwObNm2Fra4thw4ZhxIgR8Pb2hkDw+o+s+fPnQ1tbG1evXoWxsXGD9RERESgrK0NaWhpsbW1ly9PT03H+/Hno6+s3eM2dO3dkBZCrq2uj++3Zs6fse/Efo7olJSWwtLREeno6UlNTZSM+AFBTU4OKigqUl5fTXaCEtCIaglClM2fA/+OJ1cIHD4DExFZuECGtR0NDA/XnYq6qqgJQO0t2Tk4O/vvf/0IoFGLOnDkYMGCAbH1jhg4dinv37iEhIaHR9R9++CFqamrwzTffyC2XSqXw9vZGRkaG3FdeXh4GDBgg2+51d51pamrKvq974rxUKpX9u3LlSrnczMxM5OXl0cS1hLQyGgFSFcaApUvBAPAAMD4fvLAwwNOTRoGIWjIxMUHxH3PiAbWzOBcUFMh+FgqF8PHxgY+PDwIDA+Hg4IDMzEy4uLg0mufj4wNvb29MmTIFfD4fkyZNklv/wQcfYO7cufDy8gKfz0doaCgAwMXFBYcPH4a1tfUbR5j+ChcXF+Tk5MDOzu6122hqaqKmpobT/RJCmkYjQKqSmAjcvIm6UodXUwOkptIoEFFbgwcPxt69e3Hp0iXcunUL/v7+4PP5AGofgRETE4Nbt24hPz8fe/fuhVAolD0j7HXGjBmDvXv3YsaMGfj2228brHdzc8OZM2ewatUqbNq0CQAQGBiIJ0+eYPLkyUhJSUF+fj4SExMREBCgdGHy2WefYc+ePQgPD8ft27eRnZ2NgwcPYvny5bJtrK2tce7cOTx48ABPnz5Van+EEMVRAaQKjAFhYcAfH+4yfH7t8nqnAQhRlq6uLuejGU0RCATNuqZlyZIlGDBgAD7++GOMGDECo0ePRqdOnQDUXhQcHR2Nfv36oWfPnjh37hxOnjwJIyOjJnPHjx+PuLg4+Pn54ciRIw3W9+vXD6dPn0ZYWBi2bt0KiUSCK1euoKamBl5eXnB0dERwcDBEIpHSNyp4eXnh1KlTSEpKwvvvv4++ffsiMjJSrpCLiIhAUlISOnbsCGdnZ6X2RwhRHI/VPwlP8OzZM4hEIpSWlqJdu3bKByYkAMOGvX59fDzg5aX8fohaqqioQEFBAWxsbOSuK3mXp8IghKiv133mAc07ftM1QC2tbvRHQwP448JIORoatevpWiDCMZFIRAUJIYS8Bp0Ca2mVlUBhYePFD1C7vKiodjtCCCGEqASNALU0be3ai50fPXr9NqamtdsRQgghRCWoAFKFjh1rvwghhBDyVqBTYIS8I+h+BkKIOuDqs44KIELauLonEavyji9CCGktlX9cM8uv/2iZZqJTYIS0cXw+H+3bt0dJSQmA2tvReXRHISHkHSSVSvHo0SNOnnVGBRAh74C6mcTriiBCCHlXaWhowNLSUuk/9KgAIuQdwOPxIBaLYWpq+sYJQwkhpK3T0tJS+intABVAhLxT+Hy+0ufFCSFEHdBF0IQQQghRO1QAEUIIIUTtUAFECCGEELVD1wA1ou4hS8+ePWvllhBCCCFEUXXHbUUelkgFUCOeP38OAOhI01cQQgghbc7z588hEoneuA2P0fPzG5BKpbh//z4MDAw4f6Dcs2fP0LFjRxQVFaFdu3acZrd0fltue0vnt+W2t/X8ttz2tp7fltve1vPbcttbMp8xhufPn0MikTR5qzyNADVCQ0MDFhYWLbqPdu3atcgvlSry23LbWzq/Lbe9ree35ba39fy23Pa2nt+W295S+U2N/NShi6AJIYQQonaoACKEEEKI2qECSMW0tbWxYsUKaGtrt7n8ttz2ls5vy21v6/ltue1tPb8tt72t57fltqsiXxF0ETQhhBBC1A6NABFCCCFE7VABRAghhBC1QwUQIYQQQtQOFUCEEEIIUTtUAKnIxYsX4e3tDYlEAh6Ph2PHjnGWvW7dOrz//vswMDCAqakpRo8ejZycHM7yo6Ki0LNnT9kDq9zc3HDmzBnO8l+1bt068Hg8hISEcJIXHh4OHo8n92Vubs5Jdp179+5h6tSpMDIygq6uLnr16oX09HROsq2trRu0n8fjITAwkJP86upqLF++HDY2NhAKhbC1tcWqVasglUo5yX/+/DlCQkJgZWUFoVAId3d3pKam/qWspvoQYwzh4eGQSCQQCoUYOHAgbt++zVn+kSNH4OXlBWNjY/B4PGRkZHDW/qqqKixatAg9evSAnp4eJBIJpk2bhvv373PS9vDwcDg4OEBPTw/vvfcehgwZgmvXrnHS9vr+/ve/g8fjYfPmzZzlT58+vUEf6Nu3L2dtz87Oho+PD0QiEQwMDNC3b18UFhZykt9Y/+XxePj88885yX/x4gWCgoJgYWEBoVCIrl27IioqSqFsRfIfPnyI6dOnQyKRQFdXF8OGDUNeXp5C2Yocm5Ttt8qgAkhFysrK4OTkhG3btnGenZycjMDAQFy9ehVJSUmorq6Gp6cnysrKOMm3sLDA+vXrkZaWhrS0NAwePBijRo3i/Jc0NTUVO3fuRM+ePTnN7d69O4qLi2VfmZmZnGU/ffoU/fr1g6amJs6cOYOsrCxERESgffv2nOSnpqbKtT0pKQkAMGHCBE7yN2zYgC+//BLbtm1DdnY2Nm7ciM8//xz/+c9/OMmfNWsWkpKSsHfvXmRmZsLT0xNDhgzBvXv3mp3VVB/auHEjIiMjsW3bNqSmpsLc3BxDhw6Vze2nbH5ZWRn69euH9evXN7vtTeWXl5fj+vXrCAsLw/Xr13HkyBHk5ubCx8eHk7Z36dIF27ZtQ2ZmJi5fvgxra2t4enri0aNHnOTXOXbsGK5duwaJRKJQbnPyhw0bJtcXvvvuO06y79y5g/79+8PBwQEXLlzAzZs3ERYWBh0dHU7yX21zcXExdu3aBR6Ph3HjxnGSP3/+fMTHx2Pfvn3Izs7G/PnzMXfuXBw/flzpfMYYRo8ejfz8fBw/fhw3btyAlZUVhgwZotDxRZFjk7L9VimMqBwAdvTo0RbLLykpYQBYcnJyi+3jvffeY1999RVnec+fP2edO3dmSUlJzMPDgwUHB3OSu2LFCubk5MRJVmMWLVrE+vfv32L59QUHB7NOnToxqVTKSd7IkSNZQECA3LKxY8eyqVOnKp1dXl7O+Hw+O3XqlNxyJycntmzZMqWy6/chqVTKzM3N2fr162XLKioqmEgkYl9++aXS+a8qKChgANiNGzeanatIfp2UlBQGgP3yyy+cZ5eWljIA7OzZs83KflP+3bt3WYcOHditW7eYlZUV27RpU7OzX5fv7+/PRo0a9Zfymsr29fXl5Pf9dfn1jRo1ig0ePJiz/O7du7NVq1bJLXNxcWHLly9XOj8nJ4cBYLdu3ZItq66uZoaGhiw6OrrZ+fWPTVz32+aiEaB3UGlpKQDA0NCQ8+yamhocOHAAZWVlcHNz4yw3MDAQI0eOxJAhQzjLrJOXlweJRAIbGxtMmjQJ+fn5nGWfOHECrq6umDBhAkxNTeHs7Izo6GjO8l9VWVmJffv2ISAggLNJevv3749z584hNzcXAHDz5k1cvnwZI0aMUDq7uroaNTU1Df6SFgqFuHz5stL5ryooKMCDBw/g6ekpW6atrQ0PDw/88MMPnO5LVUpLS8Hj8TgbTaxTWVmJnTt3QiQSwcnJiZNMqVQKPz8/hIaGonv37pxk1nfhwgWYmpqiS5cu+PTTT1FSUqJ0plQqxenTp9GlSxd4eXnB1NQUffr04fQShVc9fPgQp0+fxsyZMznL7N+/P06cOIF79+6BMYbz588jNzcXXl5eSme/fPkSAOT6MJ/Ph5aW1l/qw/WPTa3db6kAescwxrBgwQL0798fjo6OnOVmZmZCX18f2tramD17No4ePYpu3bpxkn3gwAFcv34d69at4yTvVX369MGePXuQkJCA6OhoPHjwAO7u7nj8+DEn+fn5+YiKikLnzp2RkJCA2bNnY968edizZw8n+a86duwYfvvtN0yfPp2zzEWLFmHy5MlwcHCApqYmnJ2dERISgsmTJyudbWBgADc3N6xevRr3799HTU0N9u3bh2vXrqG4uJiD1v/pwYMHAAAzMzO55WZmZrJ1bUlFRQUWL16MKVOmcDZR5KlTp6Cvrw8dHR1s2rQJSUlJMDY25iR7w4YNEAgEmDdvHid59Q0fPhxff/01vv/+e0RERCA1NRWDBw+WHaD/qpKSErx48QLr16/HsGHDkJiYiDFjxmDs2LFITk7mqPV/iouLg4GBAcaOHctZ5tatW9GtWzdYWFhAS0sLw4YNw/bt29G/f3+lsx0cHGBlZYUlS5bg6dOnqKysxPr16/HgwYNm9+HGjk2t3W9pNvh3TFBQEH788UfO/8K2t7dHRkYGfvvtNxw+fBj+/v5ITk5WuggqKipCcHAwEhMTFT7n3hzDhw+Xfd+jRw+4ubmhU6dOiIuLw4IFC5TOl0qlcHV1xdq1awEAzs7OuH37NqKiojBt2jSl818VExOD4cOHN/v6ijc5ePAg9u3bh/3796N79+7IyMhASEgIJBIJ/P39lc7fu3cvAgIC0KFDB/D5fLi4uGDKlCm4fv06B61vqP7IGGOMs9EyVamqqsKkSZMglUqxfft2znIHDRqEjIwM/Prrr4iOjsbEiRNx7do1mJqaKpWbnp6OLVu24Pr16y32Xvv6+sq+d3R0hKurK6ysrHD69Gmliom6i/1HjRqF+fPnAwB69eqFH374AV9++SU8PDyUa3g9u3btwieffMLpZ93WrVtx9epVnDhxAlZWVrh48SLmzJkDsVis9Ii6pqYmDh8+jJkzZ8LQ0BB8Ph9DhgyR+1xV1JuOTa3Vb2kE6B0yd+5cnDhxAufPn4eFhQWn2VpaWrCzs4OrqyvWrVsHJycnbNmyRenc9PR0lJSUoHfv3hAIBBAIBEhOTsbWrVshEAhQU1PDQev/pKenhx49eih8F0NTxGJxgyKwa9euCt9BoqhffvkFZ8+exaxZszjNDQ0NxeLFizFp0iT06NEDfn5+mD9/PmejcZ06dUJycjJevHiBoqIipKSkoKqqCjY2Npzk16m7s6/+X40lJSUN/rp8m1VVVWHixIkoKChAUlISZ6M/QO3vvp2dHfr27YuYmBgIBALExMQonXvp0iWUlJTA0tJS1od/+eUX/POf/4S1tbXyDW+EWCyGlZWV0v3Y2NgYAoFAJX340qVLyMnJ4bQP//7771i6dCkiIyPh7e2Nnj17IigoCL6+vvjiiy842Ufv3r1lf/wWFxcjPj4ejx8/blYfft2xqbX7LRVA7wDGGIKCgnDkyBF8//33nB9cXrdPZYefAeCjjz5CZmYmMjIyZF+urq745JNPkJGRAT6fz0Fr//Ty5UtkZ2dDLBZzktevX78Gt3Xm5ubCysqKk/w6sbGxMDU1xciRIznNLS8vh4aG/McAn8/n7Db4Onp6ehCLxXj69CkSEhIwatQoTvNtbGxgbm4uu0sOqL3WJTk5Ge7u7pzuq6XUFT95eXk4e/YsjIyMWnR/XPVhPz8//Pjjj3J9WCKRIDQ0FAkJCRy0tKHHjx+jqKhI6X6spaWF999/XyV9OCYmBr179+bsuiug9nemqqpKJX1YJBLBxMQEeXl5SEtLU6gPN3Vsau1+S6fAVOTFixf46aefZD8XFBQgIyMDhoaGsLS0VCo7MDAQ+/fvx/Hjx2FgYCCrpkUiEYRCoVLZALB06VIMHz4cHTt2xPPnz3HgwAFcuHAB8fHxSmcbGBg0uFZJT08PRkZGnFzDtHDhQnh7e8PS0hIlJSX497//jWfPnnFyegeovQXV3d0da9euxcSJE5GSkoKdO3di586dnOQDtcP0sbGx8Pf3h0DAbZf19vbGmjVrYGlpie7du+PGjRuIjIxEQEAAJ/kJCQlgjMHe3h4//fQTQkNDYW9vjxkzZjQ7q6k+FBISgrVr16Jz587o3Lkz1q5dC11dXUyZMoWT/CdPnqCwsFD2bJ66g6a5ublCz5Z6U75EIsH48eNx/fp1nDp1CjU1NbJ+bGhoCC0trb+cbWRkhDVr1sDHxwdisRiPHz/G9u3bcffuXYUfp9DUe1O/WNPU1IS5uTns7e2Vzjc0NER4eDjGjRsHsViMn3/+GUuXLoWxsTHGjBmjdNtDQ0Ph6+uLAQMGYNCgQYiPj8fJkydx4cIFpdte99n+7NkzHDp0CBEREQplNiffw8MDoaGhEAqFsLKyQnJyMvbs2YPIyEhO8g8dOgQTExNYWloiMzMTwcHBGD16tNyFy6/T1LGp7plvyvRbpbT4fWaEMcbY+fPnGYAGX/7+/kpnN5YLgMXGxiqdzRhjAQEBzMrKimlpaTETExP20UcfscTERE6yG8PlbfC+vr5MLBYzTU1NJpFI2NixY9nt27c5ya5z8uRJ5ujoyLS1tZmDgwPbuXMnp/kJCQkMAMvJyeE0lzHGnj17xoKDg5mlpSXT0dFhtra2bNmyZezly5ec5B88eJDZ2toyLS0tZm5uzgIDA9lvv/32l7Ka6kNSqZStWLGCmZubM21tbTZgwACWmZnJWX5sbGyj61esWKF0ft2t9Y19nT9/Xqns33//nY0ZM4ZJJBKmpaXFxGIx8/HxYSkpKZy9N/U19zb4N+WXl5czT09PZmJiwjQ1NZmlpSXz9/dnhYWFnLU9JiaG2dnZMR0dHebk5MSOHTvGSdvr7NixgwmFwr/0u99UfnFxMZs+fTqTSCRMR0eH2dvbs4iICIUfldFU/pYtW5iFhYXsvV++fLnCnw+KHJuU7bfK4P3RSEIIIYQQtUHXABFCCCFE7VABRAghhBC1QwUQIYQQQtQOFUCEEEIIUTtUABFCCCFE7VABRAghhBC1QwUQIYQQQtQOFUCEEEIIUTtUABFC3lk///wzeDweMjIyFH4Nj8fDsWPHWnw/hJDWRQUQIUTleDzeG7+mT5/e2k0khLzjaDJUQojKFRcXy74/ePAgPvvsM7kZuetP4ltVVQVNTU2VtY8Q8u6jESBCiMrVzaBubm4OkUgEHo8n+7miogLt27fHN998g4EDB0JHRwf79u0DAMTGxqJr167Q0dGBg4MDtm/fLpebkpICZ2dn6OjowNXVFTdu3JCtY4zBzs4OX3zxhdxrbt26BQ0NDdy5c0e2rLi4GMOHD4dQKISNjQ0OHTqk8H7qZGVlYcSIEdDX14eZmRn8/Pzw66+/Kv3eEUK4QQUQIeSttGjRIsybNw/Z2dnw8vJCdHQ0li1bhjVr1iA7Oxtr165FWFgY4uLiAABlZWX4+OOPYW9vj/T0dISHh2PhwoWyPB6Ph4CAAMTGxsrtZ9euXfjwww/RqVMn2bKwsDCMGzcON2/exNSpUzF58mRkZ2crtB+gtoDy8PBAr169kJaWhvj4eDx8+BATJ05sqbeLENJcKplznhBCXiM2NpaJRCLZzwUFBQwA27x5s9x2HTt2ZPv375dbtnr1aubm5sYYY2zHjh3M0NCQlZWVydZHRUUxAOzGjRuMMcbu37/P+Hw+u3btGmOMscrKSmZiYsJ2794tew0ANnv2bLn99OnTh/3jH/9QeD9hYWHM09NTLqOoqIgBYDk5OYq+NYSQFkTXABFC3kqurq6y7x89eoSioiLMnDkTn376qWx5dXU1RCIRACA7OxtOTk7Q1dWVrXdzc5PLFIvFGDlyJHbt2oUPPvgAp06dQkVFBSZMmCC3Xf3Xubm5ye7wUmQ/6enpOH/+PPT19Rv8v+7cuYMuXboo8hYQQloQFUCEkLeSnp6e7HupVAoAiI6ORp8+feS24/P5AGqv8VHErFmz4Ofnh02bNiE2Nha+vr5yxczr8Hg8hfcjlUrh7e2NDRs2NFgnFosVaichpGVRAUQIeeuZmZmhQ4cOyM/PxyeffNLoNt26dcPevXvx+++/y+4iu3r1aoPtRowYAT09PURFReHMmTO4ePFig22uXr2KadOmyf3s7Oys8H5cXFxw+PBhWFtbQyCgj1lC3kZ0ETQhpE0IDw/HunXrsGXLFuTm5iIzMxOxsbGIjIwEAEyZMgUaGhqYOXMmsrKy8N133zW44wuoHTGaPn06lixZAjs7uwanrwDg0KFD2LVrF3Jzc7FixQqkpKQgKChI4f0EBgbiyZMnmDx5MlJSUpCfn4/ExEQEBASgpqamBd4dQkhzUQFECGkTZs2aha+++gq7d+9Gjx494OHhgd27d8PGxgYAoK+vj5MnTyIrKwvOzs5YtmxZo6egAGDmzJmorKxEQEBAo+tXrlyJAwcOoGfPnoiLi8PXX3+Nbt26KbwfiUSCK1euoKamBl5eXnB0dERwcDBEIhE0NOhjl5C3AY8peuKcEELeEVeuXMHAgQNx9+5dmJmZtXZzCCGtgAogQojaePnyJYqKivC3v/0NYrEYX3/9dWs3iRDSSmgslhCiNv73v//B3t4epaWl2LhxY2s3hxDSimgEiBBCCCFqh0aACCGEEKJ2qAAihBBCiNqhAogQQgghaocKIEIIIYSoHSqACCGEEKJ2qAAihBBCiNqhAogQQgghaocKIEIIIYSonf8H0ahG2BHsXPUAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plt.plot(treedepth,mean_scores,color='yellow', marker='o', linestyle='dashed',label='k-fold kryssvalidering (valideringssett)')\n",
    "plt.plot(treedepth,mean_scores_test,color='red', marker='^', linestyle='dashed',label='Testsett (ikke kryssvalidering)')\n",
    "plt.fill_between(treedepth, mean_scores-std_scores, mean_scores+std_scores,color='gray',label='usikkerhet')\n",
    "ax.set_xticks(treedepth);\n",
    "ax.set_title('')\n",
    "ax.set_xlabel('Tredybde')\n",
    "ax.set_ylabel('score')\n",
    "plt.legend()\n",
    "#plt.ylim(-0.1, 1.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8a12664",
   "metadata": {},
   "source": [
    "Her ser vi liknende resultat som vi fikk for modellen vi testet på testsettet fra i stad (røde punkter).\n",
    "\n",
    "Vi kan også implementere vår egen kryssvalidering ved å bruke såkalte [kryssvalideringsiteratorer](https://scikit-learn.org/stable/modules/cross_validation.html#cross-validation-iterators) som KFold og LeaveOneOut. De returnerer sett med indekser som definerer trenings- og testsettet for en gitt kryssvalideringsmetode. F.eks. for en k-fold kryssvalidering med 5 splits ser det slik ut: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "12e7bf3b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Doing split 0/5\n",
      "Doing split 1/5\n",
      "Doing split 2/5\n",
      "Doing split 3/5\n",
      "Doing split 4/5\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import KFold, LeaveOneOut\n",
    "\n",
    "# Definer kryssvalideringsiteratoren\n",
    "kf = KFold(n_splits=5)\n",
    "#kf = LeaveOneOut()\n",
    "\n",
    "mse_kf_test = np.zeros((20,kf.get_n_splits(X)))\n",
    "mse_kf_train = np.zeros((20,kf.get_n_splits(X)))\n",
    "\n",
    "scores_kf_train = np.zeros((20,kf.get_n_splits(X)))\n",
    "scores_kf_test = np.zeros((20,kf.get_n_splits(X)))\n",
    "\n",
    "j = 0\n",
    "# Loop over alle \"split\"'ene \n",
    "for train, test in kf.split(X):\n",
    "    print(\"Doing split %i/%i\"%(j,kf.get_n_splits(X)))\n",
    "    \n",
    "    # Hent ut trening og testdatasettet fra indeksene \n",
    "    X_train_kf, X_test_kf, y_train_kf, y_test_kf = X[train], X[test], Y[train], Y[test]\n",
    "    i = 0\n",
    "    # Loop over tredybder\n",
    "    for depth in range(1,21):\n",
    "        clf = tree.DecisionTreeClassifier(max_depth=depth)\n",
    "        clf = clf.fit(X_train_kf,y_train_kf)\n",
    "        \n",
    "        # Bruker den trente på modellen til å predikere \n",
    "        # både på testsettet ...\n",
    "        y_pred_test_kf = clf.predict(X_test_kf)\n",
    "        # ... og på treningssettet\n",
    "        y_pred_train_kf = clf.predict(X_train_kf)\n",
    "    \n",
    "        #Regner ut MSE\n",
    "        mse_kf_test[i][j] = mean_squared_error(y_test_kf,y_pred_test_kf)   \n",
    "        mse_kf_train[i][j] = mean_squared_error(y_train_kf,y_pred_train_kf)   \n",
    "        \n",
    "        # ... og score\n",
    "        scores_kf_train[i][j] = clf.score(X_train_kf, y_train_kf)\n",
    "        scores_kf_test[i][j] = clf.score(X_test_kf, y_test_kf)\n",
    "        \n",
    "        i += 1\n",
    "        \n",
    "    j += 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c9f41f6",
   "metadata": {},
   "source": [
    "Prøv gjerne å kjøre en LeaveOneOut, men det tar litt tid. Uansett kan vi regne ut gjennomsnitt og standardavvik for både MSE og score fra alle splitsene:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "24107c0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "mse_avg_kf_test = np.zeros((20,))\n",
    "mse_avg_kf_train = np.zeros((20,))\n",
    "\n",
    "mse_err_kf_test = np.zeros((20,))\n",
    "mse_err_kf_train = np.zeros((20,))\n",
    "\n",
    "score_avg_kf_test = np.zeros((20,))\n",
    "score_avg_kf_train = np.zeros((20,))\n",
    "\n",
    "score_err_kf_test = np.zeros((20,))\n",
    "score_err_kf_train = np.zeros((20,))\n",
    "\n",
    "\n",
    "std_scores = np.zeros((20,))\n",
    "for i in range(mse_kf_test.shape[0]):\n",
    "    mse_avg_kf_test[i] = mse_kf_test[i].mean()\n",
    "    mse_avg_kf_train[i] = mse_kf_train[i].mean()\n",
    "    \n",
    "    mse_err_kf_test[i] = mse_kf_test[i].std()\n",
    "    mse_err_kf_train[i] = mse_kf_train[i].std()\n",
    "    \n",
    "    score_avg_kf_test[i] = scores_kf_test[i].mean()\n",
    "    score_avg_kf_train[i] = scores_kf_train[i].mean()\n",
    "    \n",
    "    score_err_kf_test[i] = scores_kf_test[i].std()\n",
    "    score_err_kf_train[i] = scores_kf_train[i].std()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "179594df",
   "metadata": {},
   "source": [
    "... og plotte de"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "393c9ee3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fc06088ec90>"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plt.plot(treedepth,mse_avg_kf_test,color='yellow', marker='o', linestyle='dashed',label='5-fold kryssvalidering (validering)')\n",
    "\n",
    "plt.plot(treedepth,mse_avg_kf_train,color='green', marker='o', linestyle='dashed',label='5-fold kryssvalidering (trening)')\n",
    "\n",
    "plt.plot(treedepth,score_avg_kf_test,color='red', marker='o', linestyle='dashed',label='5-fold kryssvalidering (validering)')\n",
    "plt.plot(treedepth,score_avg_kf_train,color='blue', marker='o', linestyle='dashed',label='5-fold kryssvalidering (trening)')\n",
    "\n",
    "plt.fill_between(treedepth, mse_avg_kf_test-mse_err_kf_test, mse_avg_kf_test+mse_err_kf_test,color='gray')\n",
    "plt.fill_between(treedepth, mse_avg_kf_train-mse_err_kf_test, mse_avg_kf_train+mse_err_kf_test,color='gray')\n",
    "plt.fill_between(treedepth, score_avg_kf_train-mse_err_kf_test, score_avg_kf_train+mse_err_kf_test,color='gray')\n",
    "plt.fill_between(treedepth, score_avg_kf_test-mse_err_kf_test, score_avg_kf_test+mse_err_kf_test,label='usikkerhet',color='gray')\n",
    "\n",
    "\n",
    "ax.set_xticks(treedepth);\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "081a450a",
   "metadata": {},
   "source": [
    "Igjen ser vi det samme som vi har sett tidligere. Etter rundt en dybde på 9 ser det ut til at både score og MSE flater mer eller mindre ut når vi tester modellen på testsettet. \n",
    "\n",
    "La oss videre undersøke variansen og hypotesen om at jo flere splits vi kjører i kryssvalideringen jo mer korrelerte blir de ulike treningssettene som vi bruker til å trene modellen i hver split og dermed øker også variansen til modellen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "f800bdd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import KFold, LeaveOneOut\n",
    "\n",
    "# Vi definerer et sett med uike splits for å sjekke variansen\n",
    "splits = [2,5,8,10,20,50,80,100,200]\n",
    "\n",
    "mse_avg_kf_test = np.zeros((len(splits),))\n",
    "mse_avg_kf_train = np.zeros((len(splits),))\n",
    "\n",
    "mse_err_kf_test = np.zeros((len(splits),))\n",
    "mse_err_kf_train = np.zeros((len(splits),))\n",
    "\n",
    "score_avg_kf_test = np.zeros((len(splits),))\n",
    "score_avg_kf_train = np.zeros((len(splits),))\n",
    "\n",
    "score_err_kf_test = np.zeros((len(splits),))\n",
    "score_err_kf_train = np.zeros((len(splits),))\n",
    "\n",
    "i = 0\n",
    "# Looper over splits\n",
    "for split in splits:\n",
    "\n",
    "    # Definer en iterator\n",
    "    kf = KFold(n_splits=split)\n",
    "    \n",
    "    mse_kf_test = np.zeros((kf.get_n_splits(X),))\n",
    "    mse_kf_train = np.zeros((kf.get_n_splits(X),))\n",
    "\n",
    "    scores_kf_train = np.zeros((kf.get_n_splits(X),))\n",
    "    scores_kf_test = np.zeros((kf.get_n_splits(X),))\n",
    "\n",
    "    j = 0\n",
    "    for train, test in kf.split(X):\n",
    "        # For hver split trekk ut test og treningssett\n",
    "        X_train_kf, X_test_kf, y_train_kf, y_test_kf = X[train], X[test], Y[train], Y[test]\n",
    "\n",
    "        clf = tree.DecisionTreeClassifier(max_depth=6)\n",
    "        clf = clf.fit(X_train_kf,y_train_kf)\n",
    "\n",
    "        y_pred_test_kf = clf.predict(X_test_kf)\n",
    "        y_pred_train_kf = clf.predict(X_train_kf)\n",
    "\n",
    "        mse_kf_test[j] = mean_squared_error(y_test_kf,y_pred_test_kf)   \n",
    "        mse_kf_train[j] = mean_squared_error(y_train_kf,y_pred_train_kf)   \n",
    "\n",
    "        scores_kf_train[j] = clf.score(X_train_kf, y_train_kf)\n",
    "        scores_kf_test[j] = clf.score(X_test_kf, y_test_kf)\n",
    "\n",
    "\n",
    "        j += 1\n",
    "    mse_avg_kf_test[i] = mse_kf_test.mean()\n",
    "    mse_avg_kf_train[i] = mse_kf_train.mean()\n",
    "\n",
    "    mse_err_kf_test[i] = mse_kf_test.var()\n",
    "    mse_err_kf_train[i] = mse_kf_train.var()\n",
    "\n",
    "    score_avg_kf_test[i] = scores_kf_test.mean()\n",
    "    score_avg_kf_train[i] = scores_kf_train.mean()\n",
    "\n",
    "    score_err_kf_test[i] = scores_kf_test.var()\n",
    "    score_err_kf_train[i] = scores_kf_train.var()\n",
    "    i += 1\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7b082c8",
   "metadata": {},
   "source": [
    "Plot så variansen som en funksjon av antall splits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "688debbb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'varians')"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "plt.plot(splits,mse_err_kf_test,color='green', marker='o', linestyle='dashed',label='varians')\n",
    "plt.legend()\n",
    "ax.set_title('')\n",
    "ax.set_xlabel('k-fold split')\n",
    "ax.set_ylabel('varians')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ac6baf4f",
   "metadata": {},
   "source": [
    "Som forventet ser vi at variansen øker når antall splits øker fordi korrelasjonen mellom hvert enkelt treningssett øker (de blir likere og likere). "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf2c6184",
   "metadata": {},
   "source": [
    "## Bruk av kryssvalidering i hyperparameteroptimering\n",
    "\n",
    "En vktig del av maskinlæring er optimaliseringen av hyperparameterne til modellen (dvs. parameterne som definerer arkitekturen til modellen - sånn som f.eks. tredybden). F.eks. kan man sette en rekke parametere for beslutningstrær (se [dokumentasjonen](https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn-tree-decisiontreeclassifier])). Ved å bruke kryssvalidering kan man da optimalisere parameterne og finne den modellen som gir best score på valideringssettet."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "ec271244",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 17.3 s, sys: 3.7 s, total: 21 s\n",
      "Wall time: 24.8 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'criterion': 'gini',\n",
       " 'max_depth': 9,\n",
       " 'max_leaf_nodes': 21,\n",
       " 'min_samples_split': 2}"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "from sklearn.model_selection import GridSearchCV, RandomizedSearchCV\n",
    "parameters = {'criterion':['gini','entropy'],\n",
    "              'max_depth':np.arange(1,21).tolist()[0::2],\n",
    "              'min_samples_split':np.arange(2,11).tolist()[0::2],\n",
    "              'max_leaf_nodes':np.arange(3,26).tolist()[0::2]}\n",
    "\n",
    "# create an instance of the grid search object\n",
    "g1 = GridSearchCV(tree.DecisionTreeClassifier(), parameters, cv=5, n_jobs=-1)\n",
    "\n",
    "# conduct grid search over the parameter space\n",
    "g1.fit(X_train,y_train)\n",
    "\n",
    "# show best parameter configuration found for classifier\n",
    "cls_params1 = g1.best_params_\n",
    "cls_params1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "56ffdc1b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy score: 0.82\n",
      "precision score: 0.78\n",
      "recall score: 0.88\n",
      "f1 score: 0.83\n"
     ]
    }
   ],
   "source": [
    "# compute performance on training set\n",
    "model = g1.best_estimator_\n",
    "y_pred = model.predict(X_train)\n",
    "print('accuracy score: %.2f' % accuracy_score(y_train,y_pred))\n",
    "print('precision score: %.2f' % precision_score(y_train,y_pred))\n",
    "print('recall score: %.2f' % recall_score(y_train,y_pred))\n",
    "print('f1 score: %.2f' % f1_score(y_train,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c60967db",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy score: 0.76\n",
      "precision score: 0.71\n",
      "recall score: 0.85\n",
      "f1 score: 0.77\n"
     ]
    }
   ],
   "source": [
    "# compute performance on test set\n",
    "model = g1.best_estimator_\n",
    "y_pred = model.predict(X_test)\n",
    "print('accuracy score: %.2f' % accuracy_score(y_test,y_pred))\n",
    "print('precision score: %.2f' % precision_score(y_test,y_pred))\n",
    "print('recall score: %.2f' % recall_score(y_test,y_pred))\n",
    "print('f1 score: %.2f' % f1_score(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ac02a8d1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f810a4d31c0>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "disp = ConfusionMatrixDisplay(confusion_matrix(y_test, y_pred),display_labels=[\"bad\",\"good\"])\n",
    "disp.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c4a844f8",
   "metadata": {},
   "source": [
    "Se forøvrig [sklearn](https://scikit-learn.org/stable/modules/model_evaluation.html#precision-recall-f-measure-metrics) og [wikipedia](https://en.wikipedia.org/wiki/Precision_and_recall#Precision) for mer informasjon på hvordan de ulike scorene er regnet ut. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "94ad9b26",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "800"
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "288+134+55+323"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "046f33d9",
   "metadata": {},
   "source": [
    "Precision Score:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "cd0c7d90",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7067833698030634"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(323)/(323+134)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "86a8895d",
   "metadata": {},
   "source": [
    "Recall Score:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "b0e75cf7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8544973544973545"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(323)/(323+55)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aeedc1d1",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.11.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}