328 lines
85 KiB
Plaintext
328 lines
85 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "e0323bfc",
|
|
"metadata": {},
|
|
"source": [
|
|
"# Nachträge"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "41f6b0e4",
|
|
"metadata": {},
|
|
"source": [
|
|
"* dieses Dokument kann von Kurstag zu Kurstag ergänzt werden, daher sollte es jeweils ebenfalls jeweils neu heruntergeladen werden"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "0fce9c83",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Extra Themen"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "7673663f",
|
|
"metadata": {},
|
|
"source": [
|
|
"* eine Zusammenstellung von Themen aus der Präsentation, welche **nicht** prüfungsrelevant sind, zusätzlich zu den explizit mit (i) gekennzeichneten Folien\n",
|
|
"* die Zusammenstellung wird fortlaufend ergänzt"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "65be8be0",
|
|
"metadata": {},
|
|
"source": [
|
|
"|Kapitel |Folie(n)\n",
|
|
"|:------------------------------------------------------------|:-------\n",
|
|
"|**1. Feature Engineering** |\n",
|
|
"|1.1.4.5 Begriffe - Homonyme und Synonyme |20-21\n",
|
|
"|1.2.3.1 Numerische Variablen - Quantitativ (Nachbemerkung) |24\n",
|
|
"|1.2.4.5 Lineare Zusammenhänge, etc. |49-55\n",
|
|
"| |\n",
|
|
"|**2. Klassifikation** |\n",
|
|
"| |\n",
|
|
"|**3. Regression** |\n",
|
|
"|3.2.1.1 Multiple Lineare Regression |13\n",
|
|
"|3.3.2.1 DecisionTreeRegressor - Theorie |12\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "6d64e04d-d476-41f6-9bff-2bb4328c420d",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Bessere Parametrisierung bei Tuning Plots"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 1,
|
|
"id": "cb5a4819-967f-4188-b646-4cfade141217",
|
|
"metadata": {
|
|
"tags": []
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"## for scikit-learn 1.4.2, to silence warnings regarding physical cores\n",
|
|
"import os\n",
|
|
"os.environ['LOKY_MAX_CPU_COUNT'] = '4' ## depending on the hardware used"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 2,
|
|
"id": "dff45a61-33a3-45e9-8537-a53b5dd5005e",
|
|
"metadata": {
|
|
"tags": []
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"## suppress Future warning\n",
|
|
"import warnings\n",
|
|
"warnings.simplefilter(action='ignore', category=FutureWarning)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 3,
|
|
"id": "d6872c57-777f-4811-a5d5-c35655146f19",
|
|
"metadata": {
|
|
"tags": []
|
|
},
|
|
"outputs": [
|
|
{
|
|
"name": "stdout",
|
|
"output_type": "stream",
|
|
"text": [
|
|
"5 0.7493154852449042\n",
|
|
"6 0.7505324003650745\n",
|
|
"7 0.7520535442652875\n",
|
|
"8 0.7517493154852449\n",
|
|
"9 0.7554000608457561\n",
|
|
"10 0.7532704593854579\n",
|
|
"11 0.7608761788865227\n",
|
|
"12 0.7532704593854579\n",
|
|
"13 0.7602677213264375\n",
|
|
"14 0.7554000608457561\n",
|
|
"15 0.7578338910860968\n"
|
|
]
|
|
},
|
|
{
|
|
"data": {
|
|
"image/png": "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",
|
|
"text/plain": [
|
|
"<Figure size 640x480 with 1 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"import pandas as pd\n",
|
|
"import numpy as np\n",
|
|
"import matplotlib.pyplot as plt\n",
|
|
"import seaborn as sns; sns.set()\n",
|
|
"%matplotlib inline\n",
|
|
"\n",
|
|
"datapath = '../3_data'\n",
|
|
"from os import chdir; chdir(datapath)\n",
|
|
"\n",
|
|
"## load data\n",
|
|
"data = pd.read_csv('bank_data_prep.csv')\n",
|
|
"data.shape ## check\n",
|
|
"\n",
|
|
"## features - target - split\n",
|
|
"## organize features and target as independent objects\n",
|
|
"X = data.drop('y', axis=1)\n",
|
|
"y = data['y']\n",
|
|
"\n",
|
|
"## test - train - split\n",
|
|
"from sklearn.model_selection import train_test_split\n",
|
|
"X_train, X_test, y_train, y_test = train_test_split(\n",
|
|
" X,\n",
|
|
" y,\n",
|
|
" train_size=2 / 3,\n",
|
|
" random_state=1234)\n",
|
|
"\n",
|
|
"## demo dataset\n",
|
|
"demo_data = pd.read_csv('demo_data_class.csv')\n",
|
|
"X_demo = demo_data.drop('y', axis=1)\n",
|
|
"y_demo = demo_data['y']\n",
|
|
"\n",
|
|
"## import trainer class\n",
|
|
"from sklearn.neighbors import KNeighborsClassifier\n",
|
|
"\n",
|
|
"## instantiate (and parameterize) the model\n",
|
|
"model = KNeighborsClassifier()\n",
|
|
"\n",
|
|
"from sklearn.neighbors import KNeighborsClassifier\n",
|
|
"model = KNeighborsClassifier()\n",
|
|
"\n",
|
|
"params = range(5, 16) ## k values as range from 1 to 20 by 1\n",
|
|
"scores = [] ## empty list for collecting score results by iteration\n",
|
|
"\n",
|
|
"## iterate over params\n",
|
|
"for param in params:\n",
|
|
" model.set_params(n_neighbors=param)\n",
|
|
" model.fit(X_train, y_train)\n",
|
|
" scores.append(model.score(X_test, y_test))\n",
|
|
" print(param, model.score(X_test, y_test)) ## for trace progress only\n",
|
|
" \n",
|
|
"## visualization\n",
|
|
"fig = sns.lineplot(x=params, y=scores)\n",
|
|
"plt.scatter(x=params[scores.index(max(scores))], y=max(scores), color=\"black\")\n",
|
|
"plt.xlabel('k')\n",
|
|
"plt.ylabel('accuracy');"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "ac07bdfa-46f8-47ff-b511-a7f4b3b6ae2d",
|
|
"metadata": {},
|
|
"source": [
|
|
"* in der obigen Darstellung wird die Wahl der ticks (Stützpunkte) auf der X-Achse dem System überlassen\n",
|
|
"* diese kann aber auch wie folgt übersteuert werden"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 4,
|
|
"id": "bfd3cfad-c111-4f49-886c-a16362815143",
|
|
"metadata": {
|
|
"tags": []
|
|
},
|
|
"outputs": [
|
|
{
|
|
"data": {
|
|
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAkwAAAG1CAYAAAALEauPAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuNSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/xnp5ZAAAACXBIWXMAAA9hAAAPYQGoP6dpAABzhUlEQVR4nO3deXhU5fk38O+s2Sf7HrIBSdjCIqAIiiyCxa2AbUVA2aSKSstPAa24FFRawKVCcStLwYpW64tosUUUNyzITiAkISEJCdkXsmfW8/4xOUNC9mRmzszk+7muXMUzM+fcczpJ7jzP/dyPTBAEAURERETULrnUARARERE5OiZMRERERJ1gwkRERETUCSZMRERERJ1gwkRERETUCSZMRERERJ1gwkRERETUCSZMRERERJ1gwkRERETUCaXUAbgKQRBgMtmmabpcLrPZuW2FMdsHY7YPxmwfzhgz4JxxM+Zr55TJZF16LhMmKzGZBFRU1Fn9vEqlHP7+XqiurofBYLL6+W2BMdsHY7YPxmwfzhgz4JxxM+ZrAgK8oFB0LWHilBwRERFRJ5gwEREREXWCCRMRERFRJ5gwEREREXWCCRMRERFRJ5gwEREREXWCCRMRERFRJ5gwEREREXWCCRMRERFRJ9jpm4jIhRiNRhw58iNqa6/C29sPY8aMg0KhkDosIqfHhImIyEV88cU+rFmzCgUFBZZjEREReOmlDbjrrnskjIzI+XFKjojIBXzxxT4sXjy/RbIEAIWFhVi8eD6++GKfRJERuQYmTERETs5oNGLNmlUQhNY7uYvH1qxZDaPRaO/QiFwGEyYiIid35MhPLUaWvPwi0H/0TMjk5qoLQRBQUHAFR478JFWIRE6PNUxERE6uuLioxX8PmbwEIbGjoNfW4XLKgXafR0RdxxEmIiInFxoa1uK//cMSAQB+4QkdPo+Iuo4JExGRk7vpppsREREBmUwGL79wqNy9AAB+YQMBADKZDBERkbjpppulDJPIqTFhIiJycgqFAi+9tAEA4Bc2wHLcJyAKSpU7AOCll/7MfkxEvcCEiYjIBdx11z3Ytm03IuKHW47J5ArEDRqDbdt2sw8TUS+x6JuIyEXcddc9OHc1Ehn5VZDJBAiCDM+/shXTx8ZIHRqR0+MIExGRizCZBOSW1AIAxidHAgAuF9dKGRKRy2DCRETkIgor6qHVGeGmUmDSDf0AANmFNRJHReQamDAREbmInMJqAEBsmA8Sov0BAEUV9WjQGqQMi8glMGEiInIROUXm0aTYCA38fNwQqDGvkMst4igTUW8xYSIichE5ReYRprhwjfl/I3yajjNhIuotJkxERC7AaDJZCrzjwn2a/tecOImJFBH1HBMmIiIXcKW0DnqDCR5uSoQGeAK4ljBlFzJhIuotJkxERC7AUr8U5gO5TAbgWsJUerURtQ16yWIjcgVMmIiIXEDzhEnk5aFCiJ8HABZ+E/UWEyYiIhdgaSnQNKokig0XC785LUfUG0yYiIicnN5gQl5Th+/mI0zm/xbrmDjCRNQbTJiIiJxcfmktjCYB3h4qBPm6t3gsjiNMNpNdWI017x1BSlaZ1KGQHTBhIiJycs3rl2RNBd+i6FAfyABUVGtRVaeTIDrX9fWJfFwursU/v8qQOhSyAyZMRERO7lr9kk+rxzzclAgLNLcZyOUok1VlXakCAJzNKuMqxD6ACRMRkZO7NsKkafNxsa4ph3VMVlNTr0NxZQMAwGQScCqjVOKIyNYkT5hMJhPefPNN3HLLLRgxYgQefvhh5OXltfnczZs3IzExsc2vZ555xvK87OxsLF26FCNHjsT48eOxdu1aNDQ09OiaRESOTKs34kppHYBrfZeuF8sGllaXVdDyXp5IZ8Lk6iRPmLZu3YoPPvgA69atw4cffgiTyYQlS5ZAp2s9175o0SL8+OOPLb4WL14MT09PLFiwAABQWVmJefPmQalU4uOPP8bGjRvx1Vdf4c9//nOPrklE5MjySmphEgT4eqnh561u8zlxYeIWKTUQBMGe4bmsSwXm6ThxGjTlUjm0OqOUIZGNSZow6XQ6bN++HcuXL8dtt92GpKQkvP766ygqKsKBAwdaPd/LywvBwcGWr9LSUuzatQvPP/88EhMTAQDvv/8+lEolXn/9dQwYMAA333wzli9fjrNnz0IQhG5fk4jIkVnql9oo+Bb1C/WGTAZU1elwtZZ/GFpD1hXzfb9tRCTCAj2hN5iQcqlc4qjIliRNmNLS0lBXV4dx48ZZjmk0GgwePBjHjh3r9PVr167F6NGjMXPmTMuxH3/8Ebfffjvc3Nwsx371q1/h008/hUwm6/U1iYgciaV+qZ3pOABwUykQGeRlfj6n5XrNZBJwqek+DojyxU1DwwEAJ1nH5NKUUl68qKgIABAeHt7ieEhIiOWx9hw6dAinTp3C3r17WxzPzs7GlClTsH79evz3v/+FSqXC7bffjt/97ndwc3Pr1TU7o1RaP/9UKOQt/tcZMGb7YMz24egxiwlT/0hfy8+gtmKOi9Agv7QOuSW1GDM41P6BdsLR73Nzl4troNUZ4a5WICZMA7WbGnu/y8KZrDJABigd/D04070WOULMkiZMYiG2Wt1y3t3NzQ1VVVUdvnbHjh2YNGkSBg0a1OJ4bW0t3nvvPdx5553YsmULCgoKsG7dOpSWlmLjxo29umZH5HIZ/P29evz6zmg0HjY7t60wZvtgzPbhiDHXN+pRWG4u+B4xKBT+Pi2bVjaPeeiAYPxwphD5pXU2/VnVW454n693JM08kpQY4w8/P09oNB7w93FDZY0Wl8vqcUOS4yWkbXGGe309KWOWNGFydzd/c+t0Osu/AUCr1cLDo/2bUlBQgKNHj+Ldd99t9ZhSqURcXBxefPFFAMDQoUNhNBrx+9//Hk8//XSPr9kZk0lAdXV9j1/fHoVCDo3GA9XVDTAaTVY/vy0wZvtgzPbhyDGn5VZCEIAAjRtgMKKy0pw8tRVzqK+5TCHjciUqKmrbrXeSiiPf5+udzSgBAMSE+qC6ugEajQdGJQTj6xP5+PZ4HuJDvSWOsGPOdK9FtopZo/Ho8qiVpAmTOC1WUlKC6Ohoy/GSkhJLEXdbDh48iICAAIwfP77VY2FhYRg4cGCLY+J/X7lypcfX7AqDwXYfPKPRZNPz2wJjtg/GbB+OGLPYODE2TNNmbM1jDg/wgkIuQ22DHsXl9Qjyc8zRBUe8z9e7mG++7/HhPpZf3jckmhOmk+klmHd7AuRyx0pI2+IM9/p6UsYs6QRmUlISvL29cfToUcux6upqpKamYsyYMe2+7vjx4xg7diyUytb53pgxYywr4kQZGRlQKBSIiorq8TWJiBxN8y1ROqNSyhEV7N3iddR9tQ16FFeYZxPiI3wtx5Ni/OHppkR1vR6ZV3pe3kGOS9KESa1WY968edi0aRO+/vprpKWlYcWKFQgLC8O0adNgNBpRWlqKxsbGFq9LTU1FUlJSm+dcvHgx8vLy8MILLyA7Oxs//PAD/vznP+Pee+9FQEBAp9ckInIWYiPK9hpWXk/ciJcNLHtO7L8UGuAJbw+V5bhSIcfwAUEA2MTSVUleIr98+XLcd999WLNmDebMmQOFQoFt27ZBpVKhsLAQEyZMwP79+1u8prS0FH5+fm2eLz4+Hrt27cKlS5dw77334umnn8aMGTPwxz/+sUvXJCJyBnWNepQ0bc0R04URJuBa6wGOMPVcZlP/pQERrZPUGxKDAZjbC7BBqOuRtIYJABQKBVauXImVK1e2eiwqKgrp6emtjp85c6bDcyYnJ+P999/v0TWJiJxBblPSE+zn3mKkoyOWPeWKamASBMgdrPDbGYh1Y/0jfVs9NiQuAGqlHOXVjbhcXNvlRJacg+QjTERE1H2dbbjbloggLygVcjRoDSitbOj8BdRC84aV8W2MMLmpFBgWHwgAOMEmli6HCRMRkROybIkS3vVRDKVCjuimJe/ZRaxj6q4rZXXQ6oxwUyssBfTXG5VwbVqOXAsTJiIiJ5RdaB5hiuvGCFPz5+cUso6pu7IKxHYCmnbbBgwfEAiFXIaCsjpLU1FyDUyYiIicTHW9DuXV5tXD3a2TEUekuKdc912rX2o/SfV0V2FQjD8AjjK5GiZMRERORiz4DgvwhIdb99buiIXfucW1MJm4kqs7sq6I9UutC76b47Sca2LCRETkZHpSvyQKD/SCWiWHVm9EYYX1t3NyVbUNehQ13a/+bRR8NzdyYBBkME+bVlQ3dvhcch5MmIiInIxYv9SdFXIiuVyGmFBOy3XXpQLzvQr194CPp7rD5/p6u6F/lHkUiqNMroMJExGRk8kpEjt896zPj9gZnIXfXddR/6W23MBpOZfDhImIyIlU1mhxtVYHmQyIDulZwnStgSVHmLpKXCHX2XScSKxjSs+7ipp6nc3iIvthwkRE5ETEgu+IIC+4qRU9Ooe4RcrlkloYjM61W70UTCbBMiXX1RGmYD8PRId4QxCA0xfLbBke2QkTJiIiJyKOCsX2YtuNEH8PeLgpoDeYUFDGXkGdKSivQ6POCDeVApHBXl1+3ahETsu5EiZMREROxNKwMrz7Bd8iuUxmKRjnRrydE+uX4sJ9oJB3/demOC13PqcCDVqDTWIj+2HCRETkJARBaDbC1POEyfx6rpTrKrH/Ulen40SRQV4I9feAwSgg5VK5LUIjO2LCRETkJCqqtaip10Mhl6FfSNenhtoi1jFlc4SpU9cKvruXMMlkMjaxdCFMmIiInIQ4uhQZ7AWVsmcF3yJxhCm/pBZ6Awu/21PXqEdhublhZXwHW6K0R6xjOpNVDr3BaNXYyL6YMBEROYneNKy8XpCvO7w9VDCaBOSX1vb6fK5KXB0X4u8BTScNK9sSF66Bv48btDojUnMqrR0e2RETJiIiJ9HbhpXNyWQy1jF1gaVhZRf7L11PLpNh5MAgAMAJTss5NSZMREROQBAES2dua4wwAdf2omMdU/u62+G7LWLX79MXy2A0cfrTWTFhIiJyAqVXG1CvNUCpkHerF1BHLK0FuEVKm0yCgEtNo2/dLfhuLiHaD17uStQ26HExr8pa4ZGdMWEiInICYr+kfiHeUCqs86NbnJIrKKuDVs+C5OsVltWhQWtuWBnVi1WJCrkcIzgt5/SYMBEROYHsQuvVL4n8fdzg66WGSRCQV8LC7+tlFVy7591pWNmWGxJCAJjbCwiC0OvYyP6YMBEROQFr1y8BLQu/s1n43UpmU/1SfC+m40RD4vzhplKgskbL7upOigkTEZGDMwkCcoqbEiYrjjCZz8c6pvZc23C390mqSqnAsP6BANjE0lkxYSIicnDFFfXQ6oxQq+QID/S06rktrQWKOMLUXH2j3rIxcW8KvpsbldBUx5TOhMkZMWEiInJw4uhPdGjva2muJ44wFZXXc4PYZiwNK/08oPHqfsPKtgzvHwSlQoaiinpLMkbOgwkTEZGDsxR8W7F+SeTrpUaAxg0CgMvFnJYTWeqXrDAdJ/JwU2JwbAAArpZzRkyYiIgcnFgkbO36JZFYSJ7NOiYLcYWctabjRJbNeDkt53SYMBEROTCjyWQZ+RHrjayNdUwtmQTBMiU3oBcdvtsyYkAQZDIgt7gGZVUNVj032RYTJiIiB1ZYVg+dwQQPNwVCA6xb8C0SR6643N2ssKmeS62U96phZVs0XmoMjPIDAJzMKLPqucm2mDARETkwsX4pJtQHcpnMJtcQp+RKKhtQ16i3yTWcibh/XGy4xupF9sC1veXYXsC5MGEiInJg1+qXrF/wLfL2UCHYz73F9fqyaxvu2uaei3VMF/OuoqpOZ5NrkPUxYSIicmBiXZGt6pdE1zbiZR2TpX7JygXfokBfd8SE+UAAcPoiR5mcBRMmIiIHZTCaLHu82XKEyXx+1jEBQH2jwdIjKd7KBd/NXZuWYx2Ts2DCRETkoK6U1sFgFODlrkSwr7tNr3VthKlvJ0yXCqsgAAjydYevlRpWtkWclkvNqUB9IxuGOgMmTEREDkos+I4N10Bmo4JvkTjlV17diOr6vltXk3XFNu0ErhcR5IXwQE8YTQLOXuIokzNgwkRE5KDsVb8EmLtQhzW1LejLo0xZBWLBt20TJoBNLJ0NEyYiIgclJi6xNtgSpS3X6pj6ZuG3SRBwqWmEyVYr5JoTE6azl8qh0xttfj3qHSZMREQOSKc34kpT8XGcjbZEuV5fr2MqKq9HvdiwMtjb5teLDfNBgMYNOr0J53MqbH496h0mTEREDiivpBZGkwCNpwr+Pm52uWZf3yLF0rAyzAdKhe1/PcpkMowayGk5Z8GEiYjIATVvWGnrgm9RTKgPZDLgaq0OlTVau1zTkVg23LVD/ZLohkRzwnQ6swwGo8lu16XuY8JEROSAxAaS9ij4FrmpFYgIMu+d1hdHmexZ8C0aGOUHH08V6hoNyMi7arfrUvcxYSIickD22BKlLZZpuT5Wx1TfaEBBqblmrH+E/e65XC7DiAFBAIAT3FvOoTFhIiJyMI06AwrKmwq+7TjCBDQr/O5jHb+zC6uvNaz0tk/NmEicljuVUQqTINj12tR1TJiIiBzM5eJaCALg7+Nm91/eceFiwlQNoQ/98pZiOk40KCYA7moFrtbqkF3Q96ZCnQUTJiIiByNF/ZKoX4gXFHIZaur1KK9utPv1pSJ2+LbndJxIpZQjuX8gAOAkp+UcFhMmIiIHI1X9EgColApEBjcVfveROiaTIOCShCNMwLUmlicySvvUyJ4zkTxhMplMePPNN3HLLbdgxIgRePjhh5GXl9fmczdv3ozExMQ2v5555hnL8xYuXNjq8fnz51ser6urwx//+EdMmDABo0ePxsMPP4ysrCybv1cioq7IbkqY7F2/JOprdUzFFfWoazRApZSjX4jtG1a2ZVh8IJQKOUoqGywNS8mxSJ4wbd26FR988AHWrVuHDz/8ECaTCUuWLIFO13rzx0WLFuHHH39s8bV48WJ4enpiwYIFluelp6fjxRdfbPG8zZs3Wx5ft24djh49ijfffBMfffQRFAoFlixZAq227/UdISLHUt9oQHFFPQAgRqqEqY9tkSJOx9mrYWVbPNyUGBoXAIBNLB2VpAmTTqfD9u3bsXz5ctx2221ISkrC66+/jqKiIhw4cKDV8728vBAcHGz5Ki0txa5du/D8888jMTERAFBeXo7y8nIMHz68xXP9/Pws5zl48CDmzJmDUaNGoX///vj973+PgoICZGZm2uutExG1KbcpSQnydYePp1qSGOKabZHSF6aHpCz4bm5kAtsLODJJE6a0tDTU1dVh3LhxlmMajQaDBw/GsWPHOn392rVrMXr0aMycOdNyLD09HTKZDHFxce2+LjAwEPv370d5eTl0Oh0++eQT+Pn5ITo6undviIiol6SsXxJFBntBqZCjXmtAydUGyeKwF3FLlP4R0iZMIwYEQS6TIa+ktk/cd2ejlPLiRUVFAIDw8PAWx0NCQiyPtefQoUM4deoU9u7d2+J4RkYGfHx8sHbtWhw+fBienp644447sGzZMqjV5r/WXn75ZaxatQo333wzFAoFPD09sX37dvj49G74W6m0fv6paBoeVkg0TNwTjNk+GLN92Dvm3GJzwtQ/QtPjnym9jVmplCM61BuXCqqRV1KLSDtsRCvVZ6NBa8CVpoaVidF+3b7n1ozbX+OOxGg/XMitxJnMMvzipphen7Mt/D7sGUkTpoYGcwYtJjIiNzc3VFVVdfjaHTt2YNKkSRg0aFCL4xkZGdBqtUhOTsbChQtx4cIFbNiwAQUFBdiwYQMA8yhUv3798NJLL8HT0xPvvfceHn/8cXz00Uetkreukstl8Pf36tFru0Kj8bDZuW2FMdsHY7YPe8WcW1wLABiWENLrnym9iTkpNgCXCqpRUNFg059t17P3ZyM3owQCgBB/D8RFB/T4PNaK+9ZRUbiQW4nTmeV44BeDrXLO9vD7sHskTZjc3d0BmGuZxH8DgFarhYdH+zeloKAAR48exbvvvtvqsbVr12L16tXw9TUPrSYkJEClUmHFihVYtWoV8vPzsW7dOnzzzTeIiIgAALzxxhv4xS9+ge3bt+PZZ5/t0XsxmQRUV9f36LUdUSjk0Gg8UF3dAKOTbMzImO2DMduHPWOuqddZCr6DvNWorOzZailrxBwZ6AkASMsu73Ec3SHVZ+N0WjEAID5c06P3ae24B/Uz/+5Ky6lA9uUK+PlYv3Epvw+v0Wg8ujxqJWnCJI7mlJSUtKgfKikpsRRxt+XgwYMICAjA+PHjWz2mVCotyZJo4MCBAMxTgCdOnEBgYKAlWQIAlUqFwYMHIzc3t1fvx2Cw3QfPaDTZ9Py2wJjtgzHbhz1izso3j6yHBnhCrZT3+nq9iVlcXp9dVAOdzgi5XNarWLrK3p+Ni033PC5C06vrWitujacaceEaZBdW41haCSaNjOz1OdvD78PukXQCMykpCd7e3jh69KjlWHV1NVJTUzFmzJh2X3f8+HGMHTsWSmXrfG/+/PktejIBQEpKClQqFWJjYxEWFobKykqUlJRYHjeZTMjMzERsbGzv3xQRUQ9J3X+pufBAT6hVcmh1RhRVWH/03BEIguAwBd/NiXvLsev3NbUNehhN0q7YlDRhUqvVmDdvHjZt2oSvv/4aaWlpWLFiBcLCwjBt2jQYjUaUlpaisbFle/7U1FQkJSW1ec7p06fjs88+w549e5CXl4f9+/djw4YNWLx4Mby9vTFp0iT069cPy5cvx5kzZ5CVlYXnnnsOhYWFePDBB+3xtomI2iTllijXU8jliA517X5MRc0aVkaHStOwsi1i1++03ErUNeoljkZ6Z7PK8cTr32PbvnOSxiF5ifzy5ctx3333Yc2aNZgzZw4UCgW2bdsGlUqFwsJCTJgwAfv372/xmtLS0hZ9lZqbN28enn32WezevRszZszApk2bsGDBAvzud78DAHh6emLXrl2IjIzEY489hvvvvx+FhYXYs2cPoqKibP12iYja5QgtBZoTEzdX3SLlUtNGtzESNqxsS1iAJyKDvGA0CTibWS51OJJq0Brw9/+kwWgS4OOhkjQWSWuYAEChUGDlypVYuXJlq8eioqKQnp7e6viZM2c6POfcuXMxd+7cdh8PDQ3Fq6++2v1giYhspKpWi8oaLWQyOMxoR1y4a2+RIk7HDXCg6TjRqIRgXCmrw4mMUowbGiZ1OJL59LtLqKzRIsTPAzMnDUBDnXQ7cjhOSk1E1IeJ9UsRgV5wV0v+tyyAayNMl4trYDQ5V3FwV2Q2bYkSH+EYI3rNidNy5y6VQ6s3ShyNNDKvVOGbk/kAgAUzkiT/vmDCRETkABypfkkUGuAJd7UCOoMJBWWuVfjdoDXgSpm555XUW6K0JTrUG0G+7tAZTDh3qULqcOzOYDTh71+mQQAwfmgYhsYHSh0SEyYiIkfgaPVLACCXyZrVMblW4XdOYTUEAQjUuMHfBr2Oeksmk1lGmU5mlHTybNfz5ZFcXCmrg4+nCr+ZMlDqcAAwYSIikpwgCNdGmMIdZ4QJuJbAuVodU2ZTwbcjji6JxITpTGY5DE7SYNIaCsvr8PlPOQCAOVMGwlviYm8REyYiIolV1mhRXa+HQi5DPzvs29Yd4ghTtouNMIkF3/EOWPAtGhDpC42XGvVaA9IuV0odjl2YBAF//zINBqOAYfGBuHFwqNQhWTBhIiKSWHbTsv3IIC+oVQqJo2lJHGHKK6mF3sm6QrdHEARLS4H+kY4zBXo9uVyGkQODAAAnM8okjsY+vj9TgIz8KripFJg/PQEymX06zHcFEyYiIomJjSEdbToOAIJ93eHlroTRJFiKpJ1dSWUDahv0UCrkiAl1vHve3A0J17p+myTudG1rV2u1+PhQFgBg5q3xCPJ1rM2BmTAREUnMUvAd5nijHbIWhd+uUceU2TQdF+tgDSvbkhTjDw83JarrdMgqqJI6HJv6x1cZaNAaEBfug6k3OF4jacf+pBARubjmBd9xDrRCrrlrhd+uUceU5QTTcSKlQo7hA8xL6l15b7mTGaU4kV4KuUyGh+5Isttmz93BhImISEKlVY2oazRAqZAhMthL6nDaJI58ZbvICJMjbrjbEXFa7kR6KQTB9abl6hsNeP+AeVePX9wUbdnD0NEwYSIikpA4utQvxNthp4fimmqrrpTWQefkXacbdQbklzpuw8q2DI0LhEopR1lVI/JKXKOOrLl/fZeFq7U6hPh74O6bY6UOp12O+d1JRNRHOHL9ksjfxw0aTxVMguD0v7CzC2sgCECAgzasbIubWoGhcQEAXG9a7mL+VRw6dQUA8NAdSQ63SrQ5JkxERBJy1IaVzclkMpdpYOls03EisYnlCRdKmPQGE3Z+mQYAmJAcjkEx/hJH1DEmTEREEjEJAnKLzQlInAOPMAFwmS1SriVMjn2/rzd8QBAUchmulNahuMI19vX79/9yUFheD42nCr+eNEDqcDrFhImISCLFFfVo0BqhVsoRHuQpdTgdEkeYsp14hEkQhGYr5JxrhMnbQ4WkaD8ArjEtd6WsDv/+Xy4A4IHbExxm+5OOMGEiIpKIOL0VHeoDhdyxfxyLI0yFZXVo1BkkjqZnSq6KDStlDrsSqyOjmjWxdGbi9idGk4Dh/QMxJilE6pC6xLG/Q4mIXJjYCFJMRhyZn7e5SFoAcLnYOQu/xem4mDAfqJTO9+tvZEIwZDD3kaqs0UodTo99d+oKMq9UwU2twPzpiQ61/UlHnO8TQ0TkIrKLHLth5fWcvY4p60rTdJyTFXyL/LzdEN/UbNNZR5kqa7T4+Fvz9iezb41HgMZd4oi6jgkTEZEEjCYTLjcVfDvyCrnmnL2OyVLw7WT1S83dkGCevnLGhEkQBLx/IB2NOiPiIzSYPMrxtj/pCBMmIiIJFJbXQ6c3wU2tQGiAYxd8i+KceIRJqzMiT2xY6WQr5JoblRAEAEi/fBW1DXqJo+meE+mlOHWxDAq5DAscdPuTjjBhIiKSgKV+KdQHciep4YhpSpiKKxtQ3+hcv6yzC6shCOYmnM40DXS9EH9PRAV7wyQIOJNZJnU4XVbfqMc/vsoAAPziphhEhXhLHFH3MWEiIpKAuJGts0zHAYCPpxpBvuZkI9fJpuWyCpyz/1Jbbki8trecs/j42yxU1ekQFuCJu2+OkTqcHmHCREQkAXEjW2cp+BY5a8dvS8G3E9cvicT2AueyK5yixUP65Up8d7oAAPDQHYlQKR13+5OOMGEiIrIzg9Fk2ZPNGVoKNCfWMWU7UR2TuWGl8xd8i6KCvRDi5wGD0YRzlyqkDqdDeoMRf/9POgDg1uERSIx27O1POsKEiYjIzq6U1sFgNMHTTYlgPw+pw+kWS2sBJxphKr3agJp6c8PKGCdsWHk9mUzmNHvLff5TLooq6uHrpcavJ/WXOpxeYcJERGRnzeuXnKVpn0gs/C6rakRNvU7iaLpGnI6LDnXOhpVtGdVUx3Q2qwx6g0niaNqWX1qLL4+Ytz+Ze3sCPN0df/uTjrjGJ4eIyIk4a/0SAHi6qyxtEJyl8DvTUvDt/NNxovgIDXy91WjQGnEht1LqcFoxma5tfzJyYJClUN2ZMWEiIrIzywiTk9UviSx1TE6SMF2yFHw7X4LaHrlMhlEDHXdvuUOnriCroBruagXmTXOe7U86woSJiMiO9AYjrpTWAQBiw5zzF7gzbZGi1RktBfYDXKDguzlxWu7UxVKYTILE0VxTUd2IT74zb39y32394e/jJnFE1sGEiYjIjvJK6mA0CfDxVCFA45y/SJyptUBOUTVMggA/b7XL/OIWJfbzg5e7EjX1elzMvyp1OADMKxJ3/zcdWp0RAyJ9cdvISKlDshomTEREdnRtOk7jtNMU0aHekMnMG6lerdVKHU6HMpvtH+es97s9SoUcwweYt0o5meEYXb+PpZXgTFY5FHIZHvpFktN0se8KJkxERHYk9i+Kc6IO39dzVysREegFwPFHmS4VNNUvuVDBd3M3JIh1TCUQBGmn5eoa9fjg4EUAwJ3jYhAZ5CVpPNbGhImIyI7EBMNZ65dEzlDHJAgCsppGmFytfkk0JC4AapUc5dVaXC6ulTSWf36Tieo6HcIDPXHnuFhJY7EFJkxERHai1RlRUGYu+I5x0hVyImeoYyqtakR1vR4KuQwxYc632WtXqFUKDIsPBACcyCiRLI603Er8cLYQAPDQHUku0++qOdd7R0REDupySQ0EAS5RgNx8hEnqqaD2iKNL5oaVzrl/WVdYun5LtBmvTm/E3/+TBgC4bWQkEvr5SRKHrTFhIiKykxwnblh5vX4h3lDIZaiu16OyxjELv12x/1JbhvcPhEIuQ2F5PQrL6+x+/c9/ykFxZQP8vNW4b6Jzb3/SESZMRER2ku3kDSubU6sUlqJesXO5oxE7fLtq/ZLI012FQbHmTW3t3cQyr6QW/zl6GQAw9/ZEeLor7Xp9e2LCRERkJ+IIU6wLjDAB5r3wgGutEhyJVm9EflPDSlddIdfcqAT7d/02mQTs/PICjCYBNyQEu8T2Jx1hwkREZAcNWgOKKuoBOH/Bt0hc6eeIK+VyCqthNAnw9VY7bYPQ7hg5MBgymEf7yqsa7XLNr0/kI7uwBh5uSjxwe4JdriklJkxERHYgblQbqHGHxlMtcTTWcW2EqcbhCr+zmvovDYhwvYaVbfH1UmNAlHkk7eRF248ylVU14NPvLwEAfuVC2590hAkTEZEdiPVLztyw8npRwd5QKmSoazSg1E6jGl2V1azDd18hNrE8ZeNpOfP2JxnQ6o1IiPLFrSMibHo9R8GEiYjIDlytfgkwb83RL8Tc38iRpuUEQbCMMLn6CrnmxDqm9LyrqK7X2ew6Ry8UI+VSOZQK19v+pCNMmIiI7CDHhVbINXetjslxVsqVVTWiuk5nblgZ6lr3uyNBfh6IDvWGIABnLtpmb7naBj32NG1/cte4WIQHutb2Jx1hwkREZGO1DXqUXjVPWblewuR4K+WuNaz0hlrlug0r2yJOy52w0bTcR99cRE29HpFBXpgxLsYm13BUTJiIiGxMLPgO9feAp7tK4misK67ZFikmByn8znLxDXc7Ik7LpeZUoEFrsOq5z+dU4HBKEWQAHvpFEpSKvpVC9K13S0Qkgeym+h5Xql8ShQd5Qq2Uo1FnRHFT2wSp9cWCb1FEkBdCAzxhMApIuVRutfNq9Ubsatr+ZNKoSJdvBtqWHiVMxcXFVgvAZDLhzTffxC233IIRI0bg4YcfRl5eXpvP3bx5MxITE9v8euaZZyzPW7hwYavH58+f3+Jc27Ztw5QpU5CcnIxZs2bhyJEjVntPRETNiRvUutp0HAAo5HJEh4r7yklfx6TTG5FnaVjpeglqZ2Qy2bVpOSvuLbfvx2yUXm2Ev48bZrvw9icd6VHCNGnSJCxZsgT79++HTte7SvytW7figw8+wLp16/Dhhx/CZDJhyZIlbZ530aJF+PHHH1t8LV68GJ6enliwYIHleenp6XjxxRdbPG/z5s0trrllyxY8+eST2LdvH0aMGIFHH3203USNiKg3XLXgWyS+r2wHqGPKKaoxN6z0UiPQ113qcCQhTsudvVQOvcHY6/PlFtXgvz+bfz/Om5YADzfX3f6kIz1KmNavXw+TyYSnnnoKEyZMwB//+EekpKR0+zw6nQ7bt2/H8uXLcdtttyEpKQmvv/46ioqKcODAgVbP9/LyQnBwsOWrtLQUu3btwvPPP4/ExEQAQHl5OcrLyzF8+PAWz/Xz8wMA1NfX47333sNTTz2FGTNmIDY2Fs8++yz69euHEydO9OR2EBG1q6pOh4pqLWSAZSTG1TRvYCm1rIJr03F9oWFlW2LDfeDv4watzojzOZW9OpfRZMLO/6TBJAgYnRSCkQNde/uTjvQoYbr33nuxfft2HDp0CIsWLcKRI0fwq1/9CnfddRe2b9+OsrKuLWdMS0tDXV0dxo0bZzmm0WgwePBgHDt2rNPXr127FqNHj8bMmTMtx9LT0yGTyRAXF9fma06cOIGGhgbceeedlmMKhQL79u3DL3/5yy7FTUTUVWJ/ovAgL5f9y1ws/L5cXAOjySRpLFlX+l7/pevJZTKMGmidveUOHs9HblENPN2UmDt1oDXCc1q9+u4NDQ3FI488gkceeQTnz5/Hn/70J2zcuBGvvfaaZdpu+PDh7b6+qKgIABAeHt7ieEhIiOWx9hw6dAinTp3C3r17WxzPyMiAj48P1q5di8OHD8PT0xN33HEHli1bBrVajezsbPj6+iI9PR1vvPEGcnJyMGDAAKxYsQKjRo3q2Y1oolRav4Ze0bQKQeFEqxEYs30wZvvobcyXm+pp4sI1NvkZ0RZ73+fIEG+4qxVo1BlRerURUU3NLLvDGjGbG1aaR5gS+vnZ5X476md6zOAQfH0yH6cvlkEmN9eaiboac2llA/7fD+btT+6fOhCBfh62C7gTjnCfe/3nzvHjx/HZZ5/hq6++QnV1NcaPH4/bbrsN3377LebMmYNVq1a1qC9qrqGhAQCgVrfcV8nNzQ1VVVUdXnfHjh2YNGkSBg0a1OJ4RkYGtFotkpOTsXDhQly4cAEbNmxAQUEBNmzYgNraWjQ2NuL555/Hk08+iYiICHz00Ud46KGHsHfvXvTv37NiNrlcBn9/2zXw0mik+6D2FGO2D8ZsHz2NOb+sDgAwpH+QTX9GtMWe93lAPz+cyypHcVUjhiWG9vg8vYm5uKIeVbXmhpUjB4fDzY49mBztM32TxgM+nudQU69DQWUjkge0nkrrKGZBEPD6P89ApzdhaP9A/HLSQIeY4pTyPvcoYcrNzcVnn32Gffv24cqVK4iMjMT8+fMxa9Ysy2jRvHnz8NRTT+Gtt95qN2FydzcX5Ol0Osu/AUCr1cLDo/2bUlBQgKNHj+Ldd99t9djatWuxevVq+PqalzwmJCRApVJhxYoVWLVqFZRKJRobG/GHP/wBEydOBAAMGTIEp06dwvvvv48XXnihJ7cEJpOA6mrrL6lVKOTQaDxQXd0Ao1Haoe6uYsz2wZjtozcxC4KAjMvmGpJQPzdUVtbZIsRWpLjP/YK9cC6rHOcyy3DDwKBuv94aMZ9MNc9MRIf6oL62EfZocuDIn+kRAwPxw5lCfHs8D/0CPS3HuxLz4ZRCnMoohUohx/xpCbh6VdqWEba6zxqNR5dHrXqUME2fPh1ubm6YOnUq1q1b16IGqbn4+Hjk5OS0ex4xuSopKUF0dLTleElJiaWIuy0HDx5EQEAAxo8f3+oxpVJpSZZEAwea512LiooQFhYGAC3OL5PJ0L9/f+Tn57d7za4wGGz3zWI0mmx6fltgzPbBmO2jJzFX1mhRVauDXCZDRKCX3d+zPe9zdIi58PtSQXWvrtmbmDMuXwUAxEdoXPped9XIgcH44UwhjqeV4P7JA1qNELUXc029Dv84kAEAuGt8LIJ9PRzmvUl5n3s0Gfjcc8/hxx9/xKuvvtpusgQAy5YtwyeffNLu40lJSfD29sbRo0ctx6qrq5GamooxY8a0+7rjx49j7NixUCpb53vz589v0ZMJAFJSUqBSqRAbG4vRo0dDJpPh9OnTlscFQUBmZiZiYvpWm3cisi2xYWVksJddp4ekENe0Ui6vpBYGiUZaLCvk+mD/pbYMifWHm1qByhptt1Ywfvh1Jmob9IgK9sIvbozu/AV9RI8Sprlz5+KHH37A888/bzl28uRJ3Hffffjmm2+6fB61Wo158+Zh06ZN+Prrr5GWloYVK1YgLCwM06ZNg9FoRGlpKRobG1u8LjU1FUlJSW2ec/r06fjss8+wZ88e5OXlYf/+/diwYQMWL14Mb29vREREYPbs2XjppZfw3XffITs7G+vWrUN+fj4eeOCBntwOIqI2uXr/peaC/Tzg6aaEwWjClVL7TD02p9Mbcbm4qWFlH+xC3RaVUoHk+EAAXW9ieS67HP8733e3P+lIj+7E3r178X//93+4evWq5Zifnx+Cg4Px+OOP4+DBg10+1/Lly3HfffdhzZo1mDNnDhQKBbZt2waVSoXCwkJMmDAB+/fvb/Ga0tJSS1+l682bNw/PPvssdu/ejRkzZmDTpk1YsGABfve731me8+KLL2L27NlYs2YN7r33XqSmpmL79u2Ij4/v1n0gIuqI2PnaFbdEuZ5MJrP0Y5KigWVusblhpcZLjaA+2rCyLaOabcYrdLLXn1ZnxK7/pAMAptwQ1Sf34utIj2qYtm3bhoULF2L16tWWY/Hx8Xjrrbfw5z//GVu3bsXUqVO7dC6FQoGVK1di5cqVrR6LiopCenp6q+Nnzpzp8Jxz587F3Llz231cLAJfsWJFl2IkIuouQRBcekuUtsSGaZCaU2lOFEfY99qW/ksRGodYzeUokvsHQqmQobiiHgXl9YgMan+l5t4fL6GsqhEBGjfMvJUDCNfr0QjT5cuXLSvMrnfrrbfi0qVLvQqKiMjZlVU1orZBD6VChqjg7vclckZxlo7f9h9hEjfc7YubwnbEw02JwbEBAICT6SXtPi+nqBoHjpm3P5k/LdFlm6z2Ro8SpuDgYJw9e7bNx9LS0uDv79+roIiInJ04uhQV7A2VnRpWSi02zDz1eKW0zip7mHWVIAjIbCr4jmfBdyvitNzJjLZ34TCaTNj5ZRoEARg7KATDB3S/LURf0KPv4rvuugtvvfUW3n//fRQXF0Ov16O4uBgffvghNm/ejHvuucfacRIRORVxS5S+UL8kCtC4wcdTBaNJsHQ4t4eKaq2lYWVfut9dNWJgEGQyc51X2dWGVo8fOJaHy8W18HJXYs7UBAkidA49GnN77LHHcOnSJbz00kt4+eWXLccFQcAdd9yBJ554wmoBEhE5o75WvwQ0FX6HaZByqRw5hTV2KxoW2wlEhXi7fPuGntB4qpEQ5Yf0vKs4mVGKGc3qmEoq67H3h2wAwK8nD4Cvl7q90/R5PUqYVCoV3nzzTWRkZODEiROoqqqCj48PbrjhhnaX+xMR9RWmPljwLYoN8zEnTHasY8oU65e4qqtdoxKDryVMN8cCMA9y/P0/6dAbTBgU448Jw8I7Pkkf16uqroSEBCQktB6+q62thbd33yhyJCK6XkllAxq0BqiUckQG23f/OKnFNU2JdadRYm+JK+TiIzkd154bEoKx5+BFXMyvQlWtFv7+XvjxbCEu5FZCpZTjwTsSubqwEz1KmHQ6Hf7+97/j559/hk6ns/R2EAQB9fX1yMzM7HTpPxGRqxLrl6JDvVvsEt8XxDSNqBWU1UGrM8JNbdspMr3BiMvF5uSMDSvbF6BxR2yYD3KKanAyowx+fl744OBFAMA942MR6u/ZyRmoRwnThg0b8P777yMhIQEVFRVwc3NDQEAAMjIyoNfr8fjjj1s7TiIip3FtOq7vjXj4+7jBz1uNq7U65BbXIKGfn02vl1tUa25Y6alCMBtWduiGxGDkFNXgRHoJLhXWoK5Bj34h3pg+ltufdEWP/vQ5cOAAFi5ciH379mHevHkYOnQoPv74Yxw4cACRkZEwmRxjkz4iIilYVsj1sfolkZgo2mNaTqxf6h/pyymlTojtBVIuleO7U/mQyYAF3P6ky3p0lyoqKnDrrbcCMNcxpaSkAABCQ0OxdOnSVluZEBEB5mmaE2nFUodhUyaTgNymPc3i+ugSd3s2sMxi/6UuCw/0QnigJ8QdUqaNje6zn9Ge6FHC5OPjA51OBwCIiYlBYWEhamvNPyBiY2NRWFhovQiJyCXoDSb86f0TePG9I/jxbIHU4dhMYXkdtHpz7U5YQN+sCxF7IYl76dmKIAjs8N1NNySaR5lC/D0weyK3P+mOHiVMo0ePxu7du9HQ0ICYmBh4eHhYNtw9deoUV8gRUSvH00twtdb8h9bfv0xHUUW9xBHZhjgNFRPqA7m8b04RiYXfRRX1qG802Ow6lTVaXK3VQd7U/4k6d8fYaEwf2w9rFt0IdzW3P+mOHiVMjz32GE6fPo2lS5dCqVTigQcewHPPPYdZs2bhL3/5C6ZPn27tOInIyX19Ih8A4K5WQKs34p3PzkNvcL16R3FUpa/WLwHmRomBGnMBdm6x7UaZxPqlfiHeNl+N5yo83VWYOy0RcexZ1W09SpiSkpLw5Zdf4pFHHgEAPPnkk3jssccQFBSERx99FKtWrbJqkETk3LILq3GpoBpKhQx/emwCvD1UyC2uwb++y5I6NKsT63Ziw/tuwgRce/+2rGNi/yWypx6Nxz333HO47777MH78eADmdvhi8kREdL2Dx82jS2MHhaJ/lB8evnswXv/nGRw4lofBsQFI7h8ocYTWYTCaLHuo9fVi2rhwDU6kl9q0jkks+GaHb7KHHo0w7du3D3V1ddaOhYhcUHWdDseaVsbdPqYfAGBkQjCm3BAFANj271RcrdVKFp81FZTVQW8wwcNNiRA/D6nDkZQ4JWmrESa9wdSsYWXfTk7JPnqUMI0cORJHjx61dixE5IK+O1MAg1FAXLimRSfmX0/qj34h3qip1+NvX6TCJK51dmLN94/r6z2BxMLv0quNqG3QW/38ucU1MBgF+HiqENzHk1Oyjx5NySUmJmLbtm34z3/+g6SkJHh6tlw6K5PJ8Morr1glQCJyXgajCd+eugIAmNo0oiRSKRV45N4h+OPOY0jNqcR/jl7GjJtipAjTaiwNK/t4/RIAeLmrEOLvgZLKBuQUVWNonHWnXcV2Av0j2LCS7KNHCdNXX32FkJAQ6PV6S9PK5vjhJSIAOHWxDJU1Wmg8VRidFNLq8fBAL8ydmoAdX6bh/31/CYnRfujvxPUo2U0jTHFc4g7AXMdUUtmAnMIa2yVMnI4jO+lRwvTNN99YOw4ickFfH88DANw6IhIqZdsVABOSw3E+pwI/XyjBO5+dx4sLx8LT3fn6w+gNJuQ3FXxzhMksNswHR1OLbbJFSlaBeTTPmRNsci7cQIaIbOJycQ0y8qugkMswaWRku8+TyWR4cHoSgnzdUVbViF3/TYPghPVM+aXmTWC9PVSWHkR9na0KvyuqG1FZo4VMxtWIZD89+jPuwQcf7PQ5u3bt6smpichFfHPS3EpgVEIw/H3cOnyup7sSv71nCNa/fxI/XyjBkLgA3JIcYY8wraZ5/RLLEsyiQ30gA1BRrUVVnQ6+XmqrnFccXeoXzIaVZD89GmESBKHVV11dHc6ePYvMzEzEx3N/GqK+rLZBjyPnza0EplxX7N2e/pG+mHlrHADgH19loLDcuVqXZFtWyHHEQ+ThpkRYoHlRkJhQWsO1+iVOx5H99GiEaffu3W0er6qqwsMPP8yEiaiP++FsAXQGE/qFeGNgVNd/qf3iphik5lTiQm4l3v7sPNY8eANUSucYQRATgjjWL7UQF65BYXk9copqMHxAkFXOKTasZME32ZNVa5h8fX2xdOlS7Ny505qnJSInYjIJOHTS3Epgyg1R3ZqekstkePjuwfD2UCGvpBYfH3KOrVO0eiOulJlHxDjC1JKljslKI0x6gwm5RWLDSo4wkf3YpOi7vLzcFqclIidwJqsMZVWN8HJX4qbBod1+vZ+3G5bcNQgAcPBEPk5fLLN2iFaXV1wLQQB8vdWd1mv1NbFNRdnZRTVWKea/3NSw0ttD1ee7qZN99WhK7tixY62OGY1GFBUVYevWrRgyZEivAyMi5/T1CXOx963DI6BW9Ww6Lbl/EKaN6YcDx/Kwff8F/HHRWIdORLKbVoGx/1Jr/UK8IZfJUF2nQ2WNFgG9XEF4rWGlhsX1ZFc9Spjmz58PmUwGQRAsH1jxL4fw8HD84Q9/sF6EROQ0CsrqkJpTCZkMHbYS6IrZE/sj7XIlLhfX4r3Pz+Op+0dCLnfMX5DiBrPsv9Sam0qByGAv5JXUIqeopvcJk9h/idNxZGc9Spjaahkgk8ng7e2NxMREyOVs70TUF33d1EpgxIAgBPVyukSllOORe4fijzuOIe3yVfz7SC7uvjnWClFan9hniPVLbYsN82lKmKoxKiG4V+e6VvDNhInsq0eZzdixY5GUlITGxkaMHTsWY8eORXh4OE6ePIm6OudaCkxE1lHfaMBPKUUAut5KoDNhAZ6YNy0BAPDZD9nIzK+yynmtqUFrQFF5PYBrBc7UkljHJI7E9VRljRYV1WLDSt5rsq8eJUxZWVm488478eKLL1qO5eXlYf369Zg9ezYKCgqsFR8ROYnD5wqh1RsREeSFQTH+VjvvzUPDcNPgUJgEAe/sO4f6Rr3Vzm0Nl4trIAAI1LhBY6XGjK5GTCSzC6t7Vfgt1i9FBXvDXe182+eQc+tRwrRx40aEhoZiz549lmPjxo3Dd999Bz8/P2zYsMFqARKR4zMJAr5pKvaeMirSqsW4MpkM86cnItjPHeXVWuz80rG2TskuZMPKzkQFe0Mhl6Gu0YCyqsYenyeTDStJQj1KmE6ePIknnngCoaEtlwwHBgbikUcewZEjR6wSHBE5h/PZFSiubICHmwLjhoZZ/fwebkr89p6hUMhlOJ5eiu/POM4otqV+iVNE7VIp5egX4g0AvdqI95Jlw10mp2R/PUqYZDIZGhoa2nzMYDBAr3esIXMisi2xlcD4YeE2myqJj9Bg1kTzLgJ7Dl60NIqU2rUVcvwl3pFrdUw9a2BpMJosydYAjjCRBHqUMI0ZMwZ//etfUVFR0eL41atX8fbbb2Ps2LFWCY6IHF9xZT1SsszNaqeMsk6xd3umj43GkLgA6AwmvPPZOej0RpterzN1jXqUXDX/8ciC745ZOn73cIQpt7gGBqPJ3LDSnw0ryf569Kfgk08+iV//+teYMmUKRowYgYCAAFRWVuL06dNQq9V49dVXrR0nETmoQyevQAAwLD4QoQGeNr2WXCbDkjsH4YXtPyO/tA4fHcrE/GmJNr1mR8Rf/iF+HvByV0kWhzO4ljBVwyQIkHezzi3rinlkKp4NK0kiPRphiouLwxdffIH7778f9fX1OHfuHKqrq/HrX/8ae/fuRVxcnLXjJCIH1Kgz4IezhQCAKTf0rlFlV/l6u2HJXYMBmJO1kxmldrluW8TpJdYvdS4iyAsqpRwNWiNKKtsu6ejIJfZfIon1uNggNDQUDz/8MAICAgAAVVVVKC0tRViY9Qs+icgx/e98MRq0BoT4e2BofKDdrjs0PhB3jI3Gf36+jB37LyA2zKfXHaR7Iocr5LpMqZAjOtQbWVeqkVNYjbBujkaKLQUGsOCbJNKjEaaamhosWbIEc+fOtRw7c+YM7rrrLixfvhyNjT1fNkpEzkFo1kpg8qiobk+x9NasifGIDfNBXaMB7+47D5PJ/q0GxBVybKLYNWJi2d06psoaLcqbGlayuJ6k0qOEadOmTbhw4QKeeOIJy7GbbroJmzdvxsmTJ7F582arBUhEjint8lVcKauDm0qBCcPC7X59pUKO3947BG5qBTLyq/D5Tzl2vX51nc78SxxAdCgTpq5o3sCyO8TRpcggb3i4sWElSaNHCdM333yD1atXY8aMGZZjarUat99+O/7v//4P+/fvt1qAROSYxFYCNw8Ng6e7NL/EQv098WBT0fe+w9nIyLtqt2uLoyRhgZ78Jd5F4uhQbnFNt0YExf5LAyI5ukTS6VHCVFtbC1/ftgvvgoODW7UbICLXUlbVgFMXzcXWk0fZp9i7PeOGhuHmoWEQBODdz8+jtsE+feC44W73hQd4wk2tgE5vQmF51/toZbLgmxxAjxKmpKQk/Otf/2rzsb179yIxUbplvkRke4dOXYEgAINi/BEZ7C11OJh7ewJC/D1QYcetU641rOR0XFfJ5TLEhHavH5PBaLLc63gWfJOEejSO/Mgjj+CRRx7BrFmzcPvttyMwMBAVFRU4dOgQUlJS8NZbb1k7TiJyEDq9ET+cEVsJ2LZRZVd5uCnxyL1D8PKuEziZUYpvT13BJBs30cwWC745wtQtsWE+yMi7ipzCGozvQu3b5eJaGIwmeLkru72yjsiaejTCNHHiRGzduhUA8Oabb+L555/HX/7yF+j1emzduhUTJ060apBE5DiOXihGbYMegRp3jBgQJHU4FrFhGtx3W38AwJ6vM5FfUmuza1XWaFFVq4NMBvQLlX6EzZmII3JiwtmZrGbTcWxYSVLqcaXipEmTkJycDK1Wi6KiImg0Gri7u6OhoQF79uzBnDlzrBknETkAQRAsxd6TR0VCLnesX2C3j+mH1JxKpFwqx9v7zuO5h0bDTaWw+nXE+qXIIC+bnN+ViSNy4siRUtHx3+3iCjluuEtS61HClJaWhqeeegpZWVltPi6TybqcMJlMJmzZsgUff/wxampqMGbMGDz//PPo169fq+du3rwZW7ZsafM8s2bNwvr16wEACxcuxE8//dTi8bFjx2L37t2tXnf8+HHMnz8fO3fuxI033tilmIn6qswrVbhcXAuVUo5bhkdIHU4rcpkMi5u2Tikoq8NHX1/Eg3ckWf062dxwt8dC/D3g4aZEg9aAgrK6TlsyWLZEYcE3SaxHCdOGDRtQVVWF1atX49ChQ1Cr1Zg0aRK+//57fP/999i1a1eXz7V161Z88MEH+NOf/oSwsDBs3LgRS5Ysweeffw61Wt3iuYsWLcL999/f4tiOHTuwZ88eLFiwwHIsPT0dL774IqZOnWo5plK13ueppqYGq1atgslk6nK8RH2ZOLp04+BQeHs45t5pGi81ltw9GK99eBrfni7A4NgAjE4Kseo1LA0rueFut8lkMsSG+eBCbiVyimo6TJiu1mpRXt0IGYB4JqcksR7VMJ05cwa/+93vsGDBAsyYMQMNDQ144IEH8Pbbb2Pq1KltjuS0RafTYfv27Vi+fDluu+02JCUl4fXXX0dRUREOHDjQ6vleXl4IDg62fJWWlmLXrl14/vnnLSvzysvLUV5ejuHDh7d4rp+fX6vzvfjii22OZBFRa5U1WpxIN7cSmOogxd7tGRIbgDtuigYA7PwyDWVV3d+7rD2CIDRbIcdf4j1hqWPqpIGlOLoUGezFXlckuR4lTDqdDrGxsQCA2NhYpKWlWR6bNWsWTp8+3aXzpKWloa6uDuPGjbMc02g0GDx4MI4dO9bp69euXYvRo0dj5syZlmPp6emQyWSdbgD82Wef4dSpU/jDH/7QpViJ+rrvTl+B0SRgYJSvU3S2nnlLPOLCNajXGvDu56kwWmkkuby6EbUNeijkMkQ5QEsFZyTWMYmJZ3uy2H+JHEiPUvaIiAjk5eVh9OjRiI2NRW1tLfLz8xEVFQW1Wo2qqqounaeoqAgAEB7ecmlpSEiI5bH2HDp0CKdOncLevXtbHM/IyICPjw/Wrl2Lw4cPw9PTE3fccQeWLVtmmeLLz8/Hyy+/jK1bt8LLy6uL77pzSmWP8s8OKZoKIhWdFEY6EsZsH/aM2WA04bvTBQCAaWOje/xZt2fMSqUcj80aiuf+dhSZ+VX44qdczG5aRdcd18ec17T6rl+oNzwk6nDeGUf/PA+IMidA+aW1MEGAWqloM2axw3dCPz+b/Hy1Bke/121hzD3To+/2adOm4dVXX4WnpyemT5+O+Ph4vPHGG3j44Yexffv2Lk9zNTSYh8mvr1Vyc3PrNOnasWMHJk2ahEGDBrU4npGRAa1Wi+TkZCxcuBAXLlzAhg0bUFBQgA0bNsBoNGLlypX4zW9+g9GjRyM/P78b77x9crkM/v7WS76up9F42OzctsKY7cMeMX97Ig9VdToEaNwx9abYTlc2dcZe99nf3wuP3TcCm/5xAvsOZ+PGYREY1sNWCGLMBRXmn1tJsYE2/Z63Bkf9PPv5eULjpUZ1nQ5VDUYkRF+b2hRj1htMyGmashs5KIz32gYYc/f0KGF6/PHHkZubi08++QTTp0/HM888g8cffxz//ve/oVAo8Nprr3XpPO7u7gDMU3zivwFAq9XCw6P9m1JQUICjR4/i3XffbfXY2rVrsXr1asvWLQkJCVCpVFixYgVWrVqFjz76CA0NDS02DrYGk0lAdXW9Vc8JmLNpjcYD1dUNMBqdozidMduHPWPe+10mAOC2kRGoqe55PZAU9zk5zh+3DA/HD2cKsfH943jp4Rvh46nu/IVNro/5QrZ566eIAA9UVnZ9ew97cobPc2yYD85mleNsRgmCfdStYr5UUA2dwdyw0lMl4722IsZ8jUbj0eVRqx4lTG5ubnjzzTeh15v3bLrlllvwxRdf4Ny5cxgyZAiio6O7dB5xKq6kpKTFa0pKSjrcXuXgwYMICAjA+PHjWz2mVCpb7XM3cOBAAOYpwH/9618oKSmxtBAQt1B4+OGH8ctf/hJr167tUuxtMRhs98EzGk02Pb8tMGb7sHXM2YXVyLpSDaVChluSI6xyLXvf5wemJOBiXhWKKurx3r5UPDF7WLebIBqNJuj1RkuhcnSIt8N/Vhz58xwdak6YsvKrMLFZiwox5ozLlQCA+AhfmIwCTLD9dje94cj3uj2MuXt6NQHffKl+v379ur3iLCkpCd7e3jh69KglYaqurkZqairmzZvX7uuOHz+OsWPHQqlsHf78+fMRFRVl6ckEACkpKVCpVIiNjcXu3bthMBgsjxUXF2P+/Pl46aWX2kzAiPo6sZXAmKQQ+Hp1fWTGkbipFXjk3iF4addxnM4swzcnr/RoW5eSqw1o0BqgUsoREeTYU0SOTmzJkNNOx++spvql/pFciUiOQdKKRbVajXnz5mHTpk0ICAhAZGQkNm7ciLCwMEybNg1GoxEVFRXw8fFpMWWXmpqK2bNnt3nO6dOn45VXXkFycjImTJiAlJQUbNiwAYsXL4a3tze8vVuualEozF16Q0NDERgYaLs3S+SEqut0+PlCMQBgsoO3EuhMdKgPfjVpAPYcvIiPvsns0Wo/cVVXdIh3r+u4+jqxJcOVsjpo9cZWRd3XOnxzhRw5Bsm/45cvX4777rsPa9aswZw5c6BQKLBt2zaoVCoUFhZiwoQJ2L9/f4vXlJaWttlXCQDmzZuHZ599Frt378aMGTOwadMmLFiwAL/73e/s8G6IXMt3ZwpgMAqIC/dxiV9cU2+IwvD+gTAYTXhn33lodcZuvV6cjovlhru95u/jBl9vNQQByCtuue9fVa0WZVVNDSu5JQo5CMnXxCoUCqxcuRIrV65s9VhUVBTS09NbHT9z5kyH55w7dy7mzp3bpeu3dw2ivs5oMuHbU1cAoEfTV45IJpNh0Z2D8Pz2n1FYXo8PDmZg4YxBnb+wSU6R2LDS8ftQOYO4MA1OZ5Yhu6gaSbH+luPidFwEG1aSA5F8hImIHNOpjDJU1mih8VRhTFKo1OFYjY+nGkvvHgIZgB/OFlqmHDtjMgnILW5KmLglilWI9zHnuo7fnI4jR8SEiYjadLCp2PvWEZFQOWjTwJ4aFOOPO2+OAQD8/T9pKL3aeauEwop6aHVGuKkUCA9kwbc1iHVM4sid6FrCxOk4chyu9VOQiKwir6QWGXlXIZfJcNuIiM5f4ITuGR+H/pEaNGiNeGffeRg66e2S3bRNR0yoN+Ty7rUkoLaJI0xF5fVo0JpXLxuMJksCxS1RyJEwYSKiVr4+kQcAGJUYjACNeyfPdk5KhRy/vXsIPNyUuFRQjc9+zO7w+dnccNfqNF5qBGrcIADIbUqS8kpqoTOY4OmmRFigp7QBEjXDhImIWqht0OPIeXNdz1QXKfZuT5CfBxb8IgkAsP9/uUjNqWj3uddWyLF+yZrEFYfivnGZ+eaRvPhIDeTdbC5KZEtMmIiohR/PFkJnMKFfiDcGRrn+lMiYpBDcOjwCAoD3Pk9Fdb2u1XOMRpNlBIQjTNYlrjgUE1IxYWLBNzkaJkxEZGEyCfjmpLnYe8oNUd3ePsRZzZk6EOGBnqiq02H7vy/AJLTchuNycQ30BhM83JQI8Xe+DUsdmZiAWhImseCbHb7JwTBhIiKLM1llKKtqhJe7EjcOdp1WAp1xUynwyL1DoVTIcTarHAeP57d4/GLeVQDm6ThOE1mXOMVZUtmAvOIalF5tMDesDOcIEzkWJkxEZCHuG3fL8Ai4qRQSR2Nf/UK88ZvJAwAAHx/KtEzBAUBms4SJrMvLXYUQP/Oo3f7D5sL7iCAveLqzYSU5FiZMRAQAKCirQ2pOJWQyYPLISKnDkcTkUZEYOTAIRpOAtz87h0adean7xfyrAFi/ZCtiHdPXxy8D4HYo5JiYMBERAFhql0YMCEKQX9+s05HJZFg4YxD8fdxQXNmAfxzIgN5gQk5TDyaOMNmGuFKuQWve24/9l8gRMWEiIjRoDTh8rgiA6+wb11PeHiosvXswZDLg8Lki/PNQJgxGAd4eKgT5umZPKqnFXbc3HxMmckRMmIgIP6YUQqszIjzQE4Ni/Dt/gYtLjPbH3TfHAgD+e9Q8TRQXrukzqwbtLTrUB+Kd9XRTIpwNK8kBMWEi6uNMgoBvTl4B0LdaCXTm7vGxLfpQXT8KQtbj0ayrNxtWkqNiwkTUx6VmV6C4oh4ebgrcPDRM6nAchkIux9K7h8CrabXWgCg/aQNycQOb7m9SNEc4yTFx3SZRH3ewqZXA+GHhcFfzR0Jzgb7ueGb+Dcgvq0fygECYjELnL6Ie+dWk/kiMC8CNicFSh0LUJo4wEfVhJZX1SMkqBwBMHtW3i73bEx3qgzsnxHOayMZ8vd1wzy39oe5j/b/IeTBhIurDvjl5BQKAofEBCAtgoS0RUXuYMBH1UVqdET+cLQQATO3jrQSIiDrDhImoj/rf+SI0aA0I8ffA0PhAqcMhInJoTJiI+iBBECz7xk0eFcX6HCKiTjBhIuqD0i5fxZWyOqhVckwYxlYCRESdYcJE1Ad90zS6dPPQcHi6qySOhojI8TFhIupjyqsacfJiKQBgyqhIiaMhInIOTJiI+phDp65AEIBBMf6IDPaWOhwiIqfAhImoD9Hpjfj+TAEA875xRETUNUyYiPqQoxeKUdugR6DGDcMHsJUAEVFXMWEi6iOatxKYNCoKCjm//YmIuoo/MYn6iKwr1bhcXAuVUo5bh0dIHQ4RkVNhwkTURxw8kQcAuHFwKLw92EqAiKg7mDAR9QGVNVqcSBdbCbDYm4iou5gwEfUB352+AqNJwIAoX8SE+UgdDhGR02HCROTiDEYTvj1tbiUwla0EiIh6hAkTkYs7nlaC6jod/LzVGJUQLHU4REROiQkTkYsTWwncNjISSgW/5YmIeoI/PYlcWHZhNbIKqqGQyzCRrQSIiHqMCRORC/umaXRpzKAQ+Hq7SRwNEZHzYsJE5KKq63U4eqEEAPeNIyLqLSZMRC7q+9MFMBhNiAv3Qf8IX6nDISJyakyYiFyQ0WTCoVNXAHB0iYjIGpgwEbmgUxllqKzRwsdThTFJoVKHQ0Tk9JgwEbkgsZXAxBERUCn5bU5E1Fv8SUrkYvJKapGedxVymQy3jYiUOhwiIpfAhInIxYijS6MSgxGgcZc4GiIi18CEiciF1DboceR8EQDuG0dEZE1MmIhcyI9nC6EzmBAV7I2BUWwlQERkLUqpAzCZTNiyZQs+/vhj1NTUYMyYMXj++efRr1+/Vs/dvHkztmzZ0uZ5Zs2ahfXr1wMAFi5ciJ9++qnF42PHjsXu3bsBAIWFhdi4cSOOHj0KnU6H5ORkPP300xg4cKCV3x2R/ZhMAr45aZ6Omzo6CjKZTOKIiIhch+QJ09atW/HBBx/gT3/6E8LCwrBx40YsWbIEn3/+OdRqdYvnLlq0CPfff3+LYzt27MCePXuwYMECy7H09HS8+OKLmDp1quWYSqUCAOh0OixduhR+fn54++234e7ujs2bN+Ohhx7CF198gYCAANu9WSIbOptVjrKqRni5K3HjYLYSICKyJkmn5HQ6HbZv347ly5fjtttuQ1JSEl5//XUUFRXhwIEDrZ7v5eWF4OBgy1dpaSl27dqF559/HomJiQCA8vJylJeXY/jw4S2e6+fnBwA4fvw4MjIysGnTJgwbNgwDBw7Exo0bUV9fj2+++caeb58cSH2jAVdrtFKH0Stfn8gDANwyPAJuKoXE0RARuRZJR5jS0tJQV1eHcePGWY5pNBoMHjwYx44dw1133dXh69euXYvRo0dj5syZlmPp6emQyWSIi4tr8zUDBw7Eu+++i9DQa3+By+XmvLG6uro3b4ecVHZhNV775xnUNejh7+OG6BBvxIT5mL9CfeDv4+bw01uF5XU4n1MJGYBJI9lKgIjI2iRNmIqKzKt5wsPDWxwPCQmxPNaeQ4cO4dSpU9i7d2+L4xkZGfDx8cHatWtx+PBheHp64o477sCyZcugVqsRHByMiRMntnjN7t270djYiPHjx/fq/Sht0CBQoZC3+F9n4EwxZ+ZXYdOHp9CgNQIAKmu0qKzR4kxWueU5Pp4qxIZpEBvmg5hwH8SFaRDk5y55EtX8PovboIxICEJ4kJeUYXXImT4bIsZsH84YM+CccTPmnpE0YWpoaACAVrVKbm5uqKqq6vC1O3bswKRJkzBo0KAWxzMyMqDVapGcnIyFCxfiwoUL2LBhAwoKCrBhw4ZW5/nqq6/w6quvYsGCBZZpvZ6Qy2Xw97fdLyqNxsNm57YVR485NbscG/eYk6Uh8YFYNX80isrrkJVfhawrV5GVX4XLxTWoqdcj5VI5Ui5dS6K8PFToH+mL/lF+GBBl/t/wQC/I5fZPopRqJQ6nFAIAZk0aaNPPobU4+mejLYzZPpwxZsA542bM3SNpwuTubm6qp9PpLP8GAK1WCw+P9m9KQUEBjh49infffbfVY2vXrsXq1avh62teUp2QkACVSoUVK1Zg1apVCAoKsjx3z549WLduHe655x6sWrWqV+/FZBJQXV3fq3O0RaGQQ6PxQHV1A4xGk9XPbwvOEHNabiVe/fA0tHojBsX44/9+MxwBGncoISDczx0ThpqnbHV6I/JKapFbVIOcohrkFFYjr6QWdQ16nM0sw9nMMss53dUKxISaR6Fiw3wQG65BeKAnFHLb/EUk3ud//3AJDVojIoK8EB3kicrKOptczxqc4bNxPcZsH84YM+CccTPmazQajy6PWkmaMIlTcSUlJYiOjrYcLykp6XC05+DBgwgICGhzCk2pVFqSJZHYLqCoqMiSMG3cuBF/+9vfsHDhQqxevdoq0ysGg+0+eEajyabntwVHjflCTgX+8slZ6AwmDIn1x+Ozk6Fq+oa5Pma5TGZOgkJ9cOtw8zGD0YQrpXXILa4xfxXVIK+kFo06I9LzriI976rl9WqlHP1CvBEd5oPYUHNdVESQF5RWGlY2mQQc+PkyAGDyqEgYjQIAwSrntiVH/Wx0hDHbhzPGDDhn3Iy5eyRNmJKSkuDt7Y2jR49aEqbq6mqkpqZi3rx57b7u+PHjGDt2LJTK1uHPnz8fUVFRlp5MAJCSkgKVSoXY2FgA15Kl1atXY9GiRdZ9U+TQzmWXY/O/UqA3mDA0PgBPzBoGlbJ7K8qUCrmlKFxkNJlQWF6P3CJzEnW5qAa5JbXQ6ozIKqhGVkF1s9fLEBnsjZhQ80hUTJgPooK9uh0HAJzOKEVRRT3c1QqMGxLW7dcTEVHXSJowqdVqzJs3D5s2bUJAQAAiIyOxceNGhIWFYdq0aTAajaioqICPj0+LKbvU1FTMnj27zXNOnz4dr7zyCpKTkzFhwgSkpKRgw4YNWLx4sSU5+9vf/ob58+fj7rvvRmlpqeW1np6e8PJy/PoP6pmzWWXY8uk5GIwmDO8fiGUzh0FlpUJ9hVyOqGBvRAV7Y/ww88ipSRBQXFHflEDVIqeoGrnFtWjQGsyJVVENvj9jfr1cJkNEkBdiwsRESoN+Id5wU3ecRH1x+BIAYMKwcHi4Sd5WjYjIZUn+E3b58uUwGAxYs2YNGhsbMWbMGGzbtg0qlQr5+fmYMmUK1q9fj1mzZlleU1paaumrdL158+ZBJpNh9+7deOWVVxAcHIwFCxZg6dKlAIAvvvgCgHllnNj5W/T444/jiSeesM0bJUmduliKrf/vHIwmASMHBuHRXw612rRYe+QyGcIDvRAe6IWbBpuPCYKA0qpG8whU03ReTlENahv0yC+tRX5pLQ6nmFeIygCEBXoiptl0Xr8QH3i6m79tSyrrcfxCMQBgMveNIyKyKZkgCI5f8OAEjEYTKiqsX2yrVMrh7++Fyso6p5lrdrSYT6SX4O3PzsNoEjA6MRhL7xnSKlmSMmZBEFBZo7VM5+UW1SCnuAZVtbo2nx/i74GYUB/Uaw04n12B5P6B+P2vhts15p5ytM9GVzBm+3DGmAHnjJsxXxMQ4OUcRd9EtvbzhWK8uy8VJkHA2EEhePjuwTZbtdZTMpkMARp3BGjcMTIh2HL8aq0Wl5sSqNxi80q98upGlFQ2oKSywfK828e03neRiIisiwkTuawj54vw3hepEARg3JAwLLozyeGSpY74ebvBz9sNyf2vtcKoqdfhcnGtuS6qpBYRwd4Y1j8QJiMHiomIbIkJE7mkwymF2L7/AgTBXBC94BdJkjSVtDYfTzWGxAVgSFxAiyFqkxO0EiAicmZMmMjl/HCmADu/TIMAYOKICMyfngi5g+8FR0REjo0JE7mUb09dwa7/pgMAJo2KxNzbE5gsERFRrzFhIpfx9Yl8/OOrDADA1NFRmDNloOQb5BIRkWtgwkQu4cCxPHz49UUAwB1jo/GrSf2ZLBERkdUwYSKn9+XRXHx8KAsAcOe4GMy6NZ7JEhERWRUTJnJqX/yUg0+/N28Pcs/4WNw7IY7JEhERWR0TJnJKgiBg3+EcfPZjNgDgl7fE4Z7xcRJHRURErooJEzkdQRDw/364hC9+ygUAzJ4YjzvHxUobFBERuTQmTORUBEHAJ99m4cujlwEAv5k8ANPHRkscFRERuTomTOQ0BEHAR99k4sCxPADAnKkDcfto7qNGRES2x4SJnIIgCPjg4EV8fSIfADB/WgImjYqSOCoiIuormDCRwzMJAt4/kIFvT12BDMCDdyRi4ohIqcMiIqI+hAkTOTSTIGDXf9Lw/ZlCyAAsnDEIE5LDpQ6LiIj6GCZM5LBMJgE79l/A4XNFkMmAJXcOxrihYVKHRUREfRATJnJIRpMJ2/59AUfOF0Muk+HhuwfjxsGhUodFRER9FBMmcjgGowl/+yIVP18ogUIuw2/vGYLRSSFSh0VERH0YEyZyKAajCe/sO48T6aVQyGV49JdDMSohWOqwiIioj2PCRA5DbzDhrb3ncDqzDEqFDMtmDsOIAUFSh0VERMSEiRyD3mDEX//fOZzNKodSIccTs4dhWHyg1GEREREBYMJEDkCnN2Lzpyk4n10BtVKOJ+5LxpDYAKnDIiIismDCRJLS6o1485OzuJBbCbVKjt/fNxxJMf5Sh0VERNQCEyaSTKPOgL98fBbpeVfhplZgxa+GI6Gfn9RhERERtcKEiSTRoDXgjY/P4GJ+FdzVCvzfr0dgQJSv1GERERG1iQkT2V19owGv//M0sgqq4eGmxJO/GYH4CI3UYREREbWLCRPZVV2jHq99dBrZhTXwclfiyftHIDaMyRIRETk2JkxkN7UNerz64WnkFtfA20OFp+4fgehQH6nDIiIi6hQTJrKL6nodXv3wNPJKauHjqcLK+0ciKsRb6rCIiIi6hAkT2VxVnQ6b9pzClbI6aLzUWDlnJCKDvKQOi4iIqMuYMJFNXa3VYuOeUygsr4eftzlZCg9kskRERM6FCRPZTEV1I/78wSkUV9TD38cNqx4YiVB/T6nDIiIi6jYmTGQTJZX1eGX3CZRUNiBQ44aVD4xCiJ+H1GERERH1CBMmsrrSqw348wenUFLZgCBfd6x6YCSCfJksERGR82LCRFaVW1SDzZ+eRUW1FqH+Hlg5ZyQCNO5Sh0VERNQrTJjIan48W4jdB9KhN5gQGeyNlXNGQOOpljosIiKiXmPCRL2mN5iw5+uL+PbUFQDAiIFBWP3gGOi1ehgMJomjIyIi6j0mTNQrFdWN2Lr3HC4VVEMG4N4JcfjlxHh4e6pRqdVLHR4REZFVMGGiHruQW4m3PzuHmno9vNyVePjuIUjuHwi5TCZ1aERERFbFhIm6TRAE/PfnPHz8bSYEAYgO8cayWcPYNoCIiFwWEybqlgatATu+TMPxtBIAwM1DwzB/eiLcVAqJIyMiIrIdJkzUZYXlddjyaQoKy+uhkMvwwNSBuG1kJGScgiMiIhfHhIm65ER6Cbb9+wIadUb4eauxbOYwDIj0lTosIiIiu2DCRB0ymkz49PtL+PLIZQBAYj8/PPLLofD1Yn8lIiLqO5gwUbuq63V457PzuJBbCQCYNqYf7rutP5QKucSRERER2ZfkCZPJZMKWLVvw8ccfo6amBmPGjMHzzz+Pfv36tXru5s2bsWXLljbPM2vWLKxfvx4AsHDhQvz0008tHh87dix2794NANBqtfjTn/6E//znP2hsbMTkyZPx7LPPIiAgwMrvznllF1bjr/8vBRXVWripFFg4IwljB4VKHRYREZEkJE+Ytm7dig8++AB/+tOfEBYWho0bN2LJkiX4/PPPoVa3nPZZtGgR7r///hbHduzYgT179mDBggWWY+np6XjxxRcxdepUyzGVSmX594svvojjx49j8+bNUKvVeOGFF7B8+XK8//77tnmTTua701fwj68yYDAKCA3wxOMzhyIy2FvqsIiIiCQjacKk0+mwfft2PPXUU7jtttsAAK+//jpuueUWHDhwAHfddVeL53t5ecHLy8vy36mpqdi1axfWrVuHxMREAEB5eTnKy8sxfPhwBAcHt7pmcXEx9u7di7fffhujR48GALz22mu44447cOrUKYwcOdJG79bx6Q1GvH8gAz+cLQQAjBwYhMV3Doanu+R5NRERkaQkLUZJS0tDXV0dxo0bZzmm0WgwePBgHDt2rNPXr127FqNHj8bMmTMtx9LT0yGTyRAXF9fma06cOAEAuOmmmyzH4uLiEBoa2qVruqqyqgasf/8kfjhbCJkMmD0xHo/NGsZkiYiICBKPMBUVFQEAwsPDWxwPCQmxPNaeQ4cO4dSpU9i7d2+L4xkZGfDx8cHatWtx+PBheHp64o477sCyZcugVqtRXFwMf39/uLm5dfuarup8dgXe2XcetQ16eHuo8Nt7hmBIHOu5iIiIRJImTA0NDQDQqlbJzc0NVVVVHb52x44dmDRpEgYNGtTieEZGBrRaLZKTk7Fw4UJcuHABGzZsQEFBATZs2ICGhoZW1xOvqdVqe/V+lErrD9gpmlakKWywMk0QBHzxUw4++TYLggDEhvtg+exkBPVyixNbxmwrjNk+GLN9MGb7cca4GXPPSJowubu7AzDXMon/Bsyr2Dw82v+lXVBQgKNHj+Ldd99t9djatWuxevVq+PqamyomJCRApVJhxYoVWLVqFdzd3aHT6Vq9rrNrdkYul8Hf36vzJ/aQRmPdfdrqGvR448OTOHLOPKp2+9hoPDIrGWorbnFi7ZjtgTHbB2O2D8ZsP84YN2PuHkkTJnEqrqSkBNHR0ZbjJSUlliLuthw8eBABAQEYP358q8eUSqUlWRINHDgQgHkKMCwsDFevXoVOp2sx0lRSUoLQ0J4vmzeZBFRX1/f49e1RKOTQaDxQXd0Ao9FklXPml9bizY/PoqiiHkqFDPOnJ2HSqEjU1Taizgrnt0XMtsaY7YMx2wdjth9njJsxX6PReHR51ErShCkpKQne3t44evSoJWGqrq5Gamoq5s2b1+7rjh8/jrFjx0KpbB3+/PnzERUVZenJBAApKSlQqVSIjY1FcHAwTCYTTpw4YSk2z87ORnFxMcaMGdOr92Mw2O6DZzSarHL+ny8UY8f+NGj1RgRo3LDsl8MQH6GxSezWitmeGLN9MGb7YMz244xxM+bukTRhUqvVmDdvHjZt2oSAgABERkZi48aNCAsLw7Rp02A0GlFRUQEfH58WU3apqamYPXt2m+ecPn06XnnlFSQnJ2PChAlISUnBhg0bsHjxYnh7e8Pb2xt33nkn1qxZg1deeQUeHh544YUXMHbsWIwYMcJO79z+jCYTPj6UhQPH8gAAg2L88dt7h0DjyS1OiIiIOiP5mvHly5fDYDBgzZo1aGxsxJgxY7Bt2zaoVCrk5+djypQpWL9+PWbNmmV5TWlpKfz8/No837x58yCTybB792688sorCA4OxoIFC7B06VLLc9atW4dXXnkFjz/+OADg1ltvxZo1a2z6PqVUVafD23vPIT3vKgDgFzdFY9at8VDInafgj4iISEoyQRAEqYNwBUajCRUV1qgAakmplMPf3wuVlXU9GobMvFKFrf8vBVdrdXBTK7DkzkG4ITHE6nE219uYpcCY7YMx2wdjth9njJsxXxMQ4OUcNUxkO4Ig4NCpK9hz8CKMJgHhgZ54fNYwhAfabiUfERGRq2LC5IJ0eiN2/TcdPzW1DBidGIyFMwbBw43/dxMREfUEf4O6mJKrDdj6aQoul9RCJgN+ddsATB/bDzKZTOrQiIiInBYTJhdyNqsc731+HnWNBvh4qvDIvUMxKMZf6rCIiIicHhMmF2ASBHxxOAef/ZgNAUBcuAaPzRyKAI17p68lIiKizjFhcnJ1jXq893kqzmaVAwBuGxmJOVMGQmWDfe2IiIj6KiZMTiyvpBZ//TQFJVcboFTI8eD0RExIDpc6LCIiIpfDhMlJ/e98Ef7+ZRp0BhMCNe54fNYwxIT5SB0WERGRS2LC5GQMRhM++iYTX5/IBwAMjQvA0nuGwNtDJXFkRERErosJkxO5WqvF1r3nkJlfBQC46+ZY/HJCHORytgwgIiKyJSZMTiL9ciW2/CsFVXU6eLgpsOSuwRg5MFjqsIiIiPoEJkwOThAE7Ps+C9s/Pw+jSUBksBcenzkMoQGeUodGRETUZzBhcmCCIODdfak4nGLe4uTGwaFYcEcS3NQKiSMjIiLqW5gwObDiygYcTimCQi7D/VMHYvLISG5xQkREJAEmTA4s1N8Dy2YORUJsIAK8VDAYTFKHRERE1CexHbQDk8lkuGlIGPpH+UkdChERUZ/GhImIiIioE0yYiIiIiDrBhImIiIioE0yYiIiIiDrBhImIiIioE0yYiIiIiDrBhImIiIioE0yYiIiIiDrBhImIiIioE0yYiIiIiDrBhImIiIioE0yYiIiIiDrBhImIiIioEzJBEASpg3AFgiDAZLLNrVQo5DAaTTY5t60wZvtgzPbBmO3DGWMGnDNuxmwml8sgk8m69FwmTERERESd4JQcERERUSeYMBERERF1ggkTERERUSeYMBERERF1ggkTERERUSeYMBERERF1ggkTERERUSeYMBERERF1ggkTERERUSeYMBERERF1ggkTERERUSeYMBERERF1ggkTERERUSeYMDmo4uJiJCYmtvr69NNPpQ6tQ3v37sWMGTMwbNgw3Hnnnfjyyy+lDqldR48ebfMeJyYmYsqUKVKH1y6DwYC//OUvmDRpEkaOHIm5c+fi9OnTUofVqdraWrzwwguYMGECxo4di6eeegrl5eVSh9Wmd955B/Pnz29x7MKFC5g3bx5GjBiByZMnY9euXRJF17624gaA3NxcjBgxAvn5+RJE1bG2Yv7mm28we/ZsjBw5EpMnT8af//xnNDY2ShRha23FvH//ftx9991ITk7G1KlT8d5770EQBIkibK29z4ZozZo1mDx5sh0j6lxbMa9Zs6bVz2u7xS2QQ/r222+FYcOGCcXFxUJJSYnlq6GhQerQ2rV3715h8ODBwvvvvy/k5uYKW7duFZKSkoSTJ09KHVqbtFpti3tbUlIiHDhwQEhMTBQ++eQTqcNr15tvvimMHz9e+OGHH4ScnBzh2WefFW644QahuLhY6tA6tGjRImHixInCt99+K2RkZAjLli0TZsyYIWi1WqlDa+H9998XkpKShHnz5lmOVVRUCDfeeKPwzDPPCJmZmcInn3wiDBs2zKE+J23FLQiCkJmZKUyePFlISEgQ8vLyJIqubW3FfOzYMWHQoEHCW2+9JWRnZwvffvutcOuttwpPP/20hJFe01bM33//vTBo0CBh165dwuXLl4X//ve/wogRI4SdO3dKGOk17X02RF999ZWQkJAgTJo0yc6Rta+9mO+77z7htddea/Fzu7y83C4xcYTJQWVkZCA2NhYhISEIDg62fLm7u0sdWpsEQcBf/vIXPPjgg5g7dy6io6Px6KOP4uabb8bPP/8sdXhtUqvVLe6tl5cX1q9fj5kzZ2L27NlSh9eugwcP4q677sKECRMQExODp59+GjU1NQ49ynThwgX8+OOPWLt2LSZOnIiBAwdiw4YNKCkpwb///W+pwwNgHtV95JFHsGnTJsTGxrZ47J///CdUKhXWrl2L/v37Y/bs2ViwYAHeffddaYJtpqO433nnHdx3333w9fWVJrh2dBTzhx9+iBtvvBGPPPIIYmNjMXHiRKxYsQKff/45dDqdNAGj45hLS0uxdOlSzJ8/H/369cO0adNw88034/Dhw9IE26SjmEUlJSV47rnnMHbsWPsG146OYhYEAZmZmRg6dGiLn90BAQF2iY0Jk4NKT09H//79pQ6jy7Kzs3HlyhXcfffdLY5v27YNv/3tbyWKqnvefvttNDQ0YPXq1VKH0qHAwEAcOnQI+fn5MBqN+Oijj6BWq5GUlCR1aO3KyckBAIwePdpyzMvLCzExMQ6TUJ8/fx4qlQr79u3D8OHDWzx2/PhxjB07Fkql0nLspptuQk5ODsrKyuwdagsdxX3w4EGsX7/e4T7THcW8aNGiVvHK5XLo9XrU1tbaM8wWOop51qxZ+P3vfw8AMJlM+Omnn3Ds2DGMHz9egkiv6ShmwJyAPP3007j33nsdJmHqKObLly+jvr4e8fHxksSm7PwpJIWMjAz4+/tj7ty5yM7ORkxMDB599FHceuutUofWpuzsbABAfX09Fi9ejNTUVERFReHRRx91uHnxtlRUVGDnzp148skn4efnJ3U4HXr22Wfxu9/9DlOmTIFCoYBcLsfmzZsRHR0tdWjtCgkJAQAUFhZa/hAwGo0oKipCYGCglKFZTJ48ud3PalFRERISEloca/6egoKCbB5fezqK++OPPwZgrtdzJB3FPHjw4Bb/rdfrsXPnTgwdOtRuIwlt6ShmUUFBAW6//XYYDAZMmDABc+bMsVN0bess5p07d6K0tBRvv/023nnnHTtG1r6OYs7IyAAA7N69G99//z3kcjluvfVWrFixAj4+PjaPjSNMDshgMODSpUuoqqrCE088gXfffRcjRozA0qVL8b///U/q8Nok/uW3evVq3HXXXdi+fTvGjx+PZcuWOWzMzX3wwQfw8fHBb37zG6lD6VRmZiZ8fHzw17/+FR999BFmzZqFp556ChcuXJA6tHYNGzYM8fHxeOGFF1BcXIzGxka8+uqrqKyshF6vlzq8TjU2NkKtVrc45ubmBgDQarVShNQnGAwGrFq1ChcvXsQLL7wgdTid0mg0+Pjjj/HGG28gLS0Nq1atkjqkdqWlpWHLli3YuHFjq8+2o8rIyIBcLkdISAjefvttPP300/jxxx+xbNkymEwmm1+fI0wOSKlU4ujRo1AoFJaapaFDh+LixYvYtm0bxo0bJ3GEralUKgDA4sWLMXPmTADAoEGDkJqaih07djhkzM3t3bsXv/zlLx22RkxUWFiIJ598Ejt37rRMbw0bNgyZmZnYvHkztm7dKnGEbVOr1diyZQtWrVqFW2+9FSqVCnfffTcmTZoEudzx/25zd3dvVT8jJkqenp5ShOTyamtr8fvf/x4///wztmzZguTkZKlD6pS3tzcGDx6MwYMHw2g04sknn8TKlSsRGRkpdWgtaLVaPPXUU3j00Ucdeir/eo8++igeeOAB+Pv7AwASEhIQHByMX//610hJSWlz2tGaHP8nVR/l5eXV6pf3wIEDUVxcLFFEHQsNDQWAVtMWAwYMcMilzM2lpaUhLy+vVf2VIzpz5gz0ej2GDRvW4vjw4cORm5srUVRd079/f/zrX//C0aNHceTIEaxfvx5FRUUOPZUoCgsLQ0lJSYtj4n+Ln32ynpKSEku7jG3btmHixIlSh9Sh48eP4+zZsy2OJSYmAkCrz40jOHPmDC5evIgtW7Zg5MiRGDlyJN555x0UFBRg5MiROH78uNQhtkkul1uSJdHAgQMBmKfNbX59m1+Buu3ixYsYNWpUq7qDc+fOYcCAARJF1bEhQ4bAy8sLZ86caXE8IyPD4X8hHj9+HIGBgU7xl1ZYWBgA86KA5sRVlY6qtrYW8+bNQ1paGvz8/ODt7Y38/HykpqZKXhjbFWPGjMGJEydgNBotx44cOYK4uDiHqcFyFVVVVXjooYdQUVGBf/zjHxgzZozUIXVq165deOWVV1ocO3PmDJRKpUN+XyYnJ+PAgQP47LPPsHfvXuzduxf3338/QkJCsHfvXgwdOlTqENu0atUqLFiwoMWxlJQUALDL70YmTA6of//+iI+Px9q1a3H8+HFkZWVh/fr1OH36NB599FGpw2uTu7s7lixZgr/+9a/44osvcPnyZbz11ls4fPgwFi5cKHV4HUpNTbX8NejokpOTccMNN2D16tU4cuQIcnJy8MYbb+B///sfli5dKnV47fL29oYgCHj55Zdx8eJFpKSk4NFHH8VNN93k8NO1ADB79mzU1tbi2WefRWZmJj799FPs3LnTaVaAOpP169cjLy8PGzduREBAAEpLSy1fzRNWR7JgwQKcPXsWr7/+OnJzc/Hll19i48aNePDBB1uNiDgCd3d3xMTEtPjy9fWFUqlETEyMw5YmTJ8+Hf/73/+wZcsWXL58Gd999x3+8Ic/4K677rLLqnLWMDkguVyOt99+G6+++ip+//vfo7q6GoMHD8aOHTtaTXk5kmXLlsHDwwOvv/46iouL0b9/f2zevBk33nij1KF1qLS01OFXxonkcjneeustvPHGG3jmmWdQVVWFhIQE7Ny50+bz97312muvYd26dZgzZw7UajWmTZuGlStXSh1WlwQGBuJvf/sbXn75ZcycORPBwcFYtWqVpV6PrMNoNGL//v3Q6/V46KGHWj3+9ddfIyoqSoLIOjZq1Ci88847eOONN7Bz504EBARg0aJFePjhh6UOzaVMmTIFb7zxBt59912899578PHxwd13321p6WBrMkFwoN7tRERERA6IU3JEREREnWDCRERERNQJJkxEREREnWDCRERERNQJJkxEREREnWDCRERERNQJJkxEREREnWDCRETUhsmTJ+Ppp5+WOgwichBMmIiIiIg6wYSJiIiIqBNMmIiIuuCTTz5BUlIS/vrXv0odChFJgAkTEVEn9u/fj+eeew7Lli3DY489JnU4RCQBJkxERB04dOgQVq1ahaVLl2L58uVSh0NEEpEJgiBIHQQRkaOZPHkyvLy8kJubi6CgIBw8eBByOf/GJOqr+N1PRNSOjIwMjBs3DleuXME//vEPqcMhIgkxYSIiasctt9yCd955BzNmzMBrr72GwsJCqUMiIokwYSIiakdQUBAA4JlnnoFCocCLL74obUBEJBkmTEREnQgJCcGKFSvw7bff4osvvpA6HCKSABMmIqIumDNnDpKTk/Hyyy+jsrJS6nCIyM64So6IiIioExxhIiIiIuoEEyYiIiKiTjBhIiIiIuoEEyYiIiKiTjBhIiIiIuoEEyYiIiKiTjBhIiIiIuoEEyYiIiKiTjBhIiIiIuoEEyYiIiKiTjBhIiIiIuoEEyYiIiKiTvx/yuj0nOMf8NUAAAAASUVORK5CYII=",
|
|
"text/plain": [
|
|
"<Figure size 640x480 with 1 Axes>"
|
|
]
|
|
},
|
|
"metadata": {},
|
|
"output_type": "display_data"
|
|
}
|
|
],
|
|
"source": [
|
|
"## $$ andere Grafik darstellung mit kontrollierten Werten auf der X-Achse\n",
|
|
"fig = sns.lineplot(x=params, y=scores)\n",
|
|
"plt.scatter(x=params[scores.index(max(scores))], y=max(scores), color=\"black\")\n",
|
|
"plt.xlabel('k')\n",
|
|
"plt.xticks(params) ## new\n",
|
|
"plt.ylabel('accuracy');"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "4188ae6b-70ac-4b99-a967-d38eabfeb6d2",
|
|
"metadata": {},
|
|
"source": [
|
|
"## Splitting Kriterium bei DecisionTreeRegressor"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"id": "e327940d-6b01-4d14-b6b6-90f524a89845",
|
|
"metadata": {},
|
|
"source": [
|
|
"vgl. Folie 12 in Kapitel 3.3:\n",
|
|
"* in der Folie steht für MSE folgende Formel:\n",
|
|
"\n",
|
|
"$$MSE=\\frac{1}{n}\\sum\\limits_{i=1}^{n}{(y_i-\\hat{y}_i)^2}$$\n",
|
|
"\n",
|
|
"* wobei\n",
|
|
" * $n$: Anzahl Beobachtungen der Testdaten\n",
|
|
" * $y_i$: wahrer Targetwert der i-ten Beobachtung der Testdaten\n",
|
|
" * $\\hat{y}_i$: Mittelwert aller Targetwerte der Testdaten\n",
|
|
"* das ist exakt dieselbe Formulierung für Mean squared error, wie sie auch in der Zusammenstellung der Performance Metriken Regression in Kap. 4.4.2.2. geschrieben ist\n",
|
|
"* als Splitting Kriterium wird bei diesem Regressor per Default `'squared_error'` eingesetzt\n",
|
|
"* das ist tatsächlich eine (leichte) Modifikation der oben genannten Metrik:\n",
|
|
" * $n$: Anzahl der Beobachtung im jeweiligen Knoten\n",
|
|
" * $y_i$: wahrer Targetwert der i-ten Beobachtung im jeweiligen Knoten\n",
|
|
" * $\\hat{y}_i$: Prediction des jeweiligen Knotens\n",
|
|
"* da aber die Prediction jedes Knotens aus dem Mittelwert der darin vorliegenden Targetwerte besteht, kann obige Formel auch so geschrieben werden:\n",
|
|
"\n",
|
|
"$$MSE=\\frac{1}{n}\\sum\\limits_{i=1}^{n}{(y_i-\\bar{y})^2}$$\n",
|
|
"\n",
|
|
"* dadurch entspricht die Formulierung jener der *Stichprobenvarianz*, welche aus der Statistik bekann sein dürfte"
|
|
]
|
|
}
|
|
],
|
|
"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": "3.11.7"
|
|
},
|
|
"toc": {
|
|
"base_numbering": "9",
|
|
"nav_menu": {},
|
|
"number_sections": true,
|
|
"sideBar": true,
|
|
"skip_h1_title": false,
|
|
"title_cell": "9 Nachträge Allgemein",
|
|
"title_sidebar": "Contents",
|
|
"toc_cell": true,
|
|
"toc_position": {
|
|
"height": "calc(100% - 180px)",
|
|
"left": "10px",
|
|
"top": "150px",
|
|
"width": "240.667px"
|
|
},
|
|
"toc_section_display": true,
|
|
"toc_window_display": true
|
|
},
|
|
"toc-autonumbering": true,
|
|
"varInspector": {
|
|
"cols": {
|
|
"lenName": 16,
|
|
"lenType": 16,
|
|
"lenVar": 40
|
|
},
|
|
"kernels_config": {
|
|
"python": {
|
|
"delete_cmd_postfix": "",
|
|
"delete_cmd_prefix": "del ",
|
|
"library": "var_list.py",
|
|
"varRefreshCmd": "print(var_dic_list())"
|
|
},
|
|
"r": {
|
|
"delete_cmd_postfix": ") ",
|
|
"delete_cmd_prefix": "rm(",
|
|
"library": "var_list.r",
|
|
"varRefreshCmd": "cat(var_dic_list()) "
|
|
}
|
|
},
|
|
"types_to_exclude": [
|
|
"module",
|
|
"function",
|
|
"builtin_function_or_method",
|
|
"instance",
|
|
"_Feature"
|
|
],
|
|
"window_display": false
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 5
|
|
}
|