From 597c93af3d0c70db7e0476fadf5a189189d5f59f Mon Sep 17 00:00:00 2001 From: Dale Smith Date: Wed, 7 Sep 2016 21:07:36 -0400 Subject: [PATCH] save notebooks in jupyter --- notebooks/Demo-Iris.ipynb | 336 +++++++++++++++++------------ notebooks/skdata_quick_intro.ipynb | 223 ++++++++++--------- 2 files changed, 314 insertions(+), 245 deletions(-) diff --git a/notebooks/Demo-Iris.ipynb b/notebooks/Demo-Iris.ipynb index eb26bbb8..a9283935 100644 --- a/notebooks/Demo-Iris.ipynb +++ b/notebooks/Demo-Iris.ipynb @@ -1,155 +1,209 @@ { - "metadata": { - "name": "Demo-Iris" - }, - "nbformat": 3, - "nbformat_minor": 0, - "worksheets": [ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Hyperopt-Sklearn on Iris\n", + "\n", + "`Iris` is a small data set of 150 examples of flower attributes and types of Iris.\n", + "\n", + "The small size of Iris means that hyperparameter optimization takes just a few seconds.\n", + "On the other hand, Iris is so *easy* that we'll typically see numerous near-perfect models within the first few random guesses; hyperparameter optimization algorithms are hardly necessary at all.\n", + "\n", + "Nevertheless, here is how to use hyperopt-sklearn (`hpsklearn`) to find a good model of the Iris data set. The code walk-through is given in 5 steps:\n", + "\n", + " 1. module imports\n", + " 2. data preparation into training and testing sets for a single fold of cross-validation.\n", + " 3. creation of a hpsklearn `HyperoptEstimator`\n", + " 4. a somewhat spelled-out version of `HyperoptEstimator.fit`\n", + " 5. inspecting and testing the best model" + ] + }, { - "cells": [ + "cell_type": "code", + "execution_count": 1, + "metadata": { + "collapsed": false + }, + "outputs": [ { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "# Hyperopt-Sklearn on Iris\n", - "\n", - "`Iris` is a small data set of 150 examples of flower attributes and types of Iris.\n", - "\n", - "The small size of Iris means that hyperparameter optimization takes just a few seconds.\n", - "On the other hand, Iris is so *easy* that we'll typically see numerous near-perfect models within the first few random guesses; hyperparameter optimization algorithms are hardly necessary at all.\n", - "\n", - "Nevertheless, here is how to use hyperopt-sklearn (`hpsklearn`) to find a good model of the Iris data set. The code walk-through is given in 5 steps:\n", - "\n", - " 1. module imports\n", - " 2. data preparation into training and testing sets for a single fold of cross-validation.\n", - " 3. creation of a hpsklearn `HyperoptEstimator`\n", - " 4. a somewhat spelled-out version of `HyperoptEstimator.fit`\n", - " 5. inspecting and testing the best model" + "name": "stdout", + "output_type": "stream", + "text": [ + "WARN: OMP_NUM_THREADS=None =>\n", + "... If you are using openblas if you are using openblas set OMP_NUM_THREADS=1 or risk subprocess calls hanging indefinitely\n" ] }, { - "cell_type": "code", - "collapsed": false, - "input": [ - "# IMPORTS\n", - "import numpy as np\n", - "import skdata.iris.view\n", - "import hyperopt.tpe\n", - "import hpsklearn\n", - "import hpsklearn.demo_support\n", - "import time" - ], - "language": "python", - "metadata": {}, - "outputs": [], - "prompt_number": 1 - }, + "name": "stderr", + "output_type": "stream", + "text": [ + "/Applications/anaconda/lib/python3.5/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n", + " warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n", + "/Applications/anaconda/lib/python3.5/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n", + " warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n" + ] + } + ], + "source": [ + "# IMPORTS\n", + "import numpy as np\n", + "import skdata.iris.view\n", + "import hyperopt.tpe\n", + "import hpsklearn\n", + "import hpsklearn.demo_support\n", + "import time" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": { + "collapsed": false + }, + "outputs": [ { - "cell_type": "code", - "collapsed": false, - "input": [ - "# PREPARE TRAINING AND TEST DATA\n", - "data_view = skdata.iris.view.KfoldClassification(4)\n", - "attrs = 'petal_length', 'petal_width', 'sepal_length', 'sepal_width'\n", - "labels = 'setosa', 'versicolor', 'virginica'\n", - "X_all = np.asarray([map(d.__getitem__, attrs) for d in data_view.dataset.meta])\n", - "y_all = np.asarray([labels.index(d['name']) for d in data_view.dataset.meta])\n", - "idx_all = np.random.RandomState(1).permutation(len(y_all))\n", - "idx_train = idx_all[:int(.8 * len(y_all))]\n", - "idx_test = idx_all[int(.8 * len(y_all)):]\n", - "\n", - "# TRAIN AND TEST DATA\n", - "X_train = X_all[idx_train]\n", - "y_train = y_all[idx_train]\n", - "X_test = X_all[idx_test]\n", - "y_test = y_all[idx_test]" - ], - "language": "python", - "metadata": {}, - "outputs": [], - "prompt_number": 2 - }, + "ename": "NotADirectoryError", + "evalue": "[Errno 20] Not a directory: '/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/iris/iris.csv'", + "output_type": "error", + "traceback": [ + "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", + "\u001b[0;31mAttributeError\u001b[0m Traceback (most recent call last)", + "\u001b[0;32m/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/toy.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 17\u001b[0m \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 18\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdescr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta_const\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 19\u001b[0m \u001b[0;32mexcept\u001b[0m \u001b[0mAttributeError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;31mAttributeError\u001b[0m: 'Iris' object has no attribute 'meta'", + "\nDuring handling of the above exception, another exception occurred:\n", + "\u001b[0;31mNotADirectoryError\u001b[0m Traceback (most recent call last)", + "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[0;31m# PREPARE TRAINING AND TEST DATA\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mdata_view\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mskdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0miris\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mview\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mKfoldClassification\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m4\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 3\u001b[0m \u001b[0mattrs\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'petal_length'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'petal_width'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'sepal_length'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'sepal_width'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 4\u001b[0m \u001b[0mlabels\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'setosa'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'versicolor'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'virginica'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 5\u001b[0m \u001b[0mX_all\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0masarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mmap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0md\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__getitem__\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mattrs\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0md\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mdata_view\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/iris/view.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self, K, rseed)\u001b[0m\n\u001b[1;32m 25\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0m__init__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mK\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrseed\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 26\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mK\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mK\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 27\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdataset\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mIris\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 28\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrseed\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrseed\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 29\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/toy.py\u001b[0m in \u001b[0;36m__init__\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 18\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdescr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta_const\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 19\u001b[0m \u001b[0;32mexcept\u001b[0m \u001b[0mAttributeError\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 20\u001b[0;31m \u001b[0mmeta\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdescr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmeta_const\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbuild_all\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 21\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmeta\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmeta\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 22\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdescr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdescr\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/toy.py\u001b[0m in \u001b[0;36mbuild_all\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 31\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 32\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mbuild_all\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 33\u001b[0;31m \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbuild_meta\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbuild_descr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbuild_meta_const\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 34\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 35\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mbuild_descr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/iris/dataset.py\u001b[0m in \u001b[0;36mbuild_meta\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m 89\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mbuild_meta\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 90\u001b[0m \u001b[0mmodule_path\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdirname\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m__file__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 91\u001b[0;31m \u001b[0mdata_file\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcsv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mreader\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mopen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mos\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpath\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mjoin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmodule_path\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'iris.csv'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 92\u001b[0m \u001b[0mtemp\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnext\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata_file\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 93\u001b[0m \u001b[0mn_samples\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtemp\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;31mNotADirectoryError\u001b[0m: [Errno 20] Not a directory: '/Applications/anaconda/lib/python3.5/site-packages/skdata-0.0.4-py3.5.egg/skdata/iris/iris.csv'" + ] + } + ], + "source": [ + "# PREPARE TRAINING AND TEST DATA\n", + "data_view = skdata.iris.view.KfoldClassification(4)\n", + "attrs = 'petal_length', 'petal_width', 'sepal_length', 'sepal_width'\n", + "labels = 'setosa', 'versicolor', 'virginica'\n", + "X_all = np.asarray([map(d.__getitem__, attrs) for d in data_view.dataset.meta])\n", + "y_all = np.asarray([labels.index(d['name']) for d in data_view.dataset.meta])\n", + "idx_all = np.random.RandomState(1).permutation(len(y_all))\n", + "idx_train = idx_all[:int(.8 * len(y_all))]\n", + "idx_test = idx_all[int(.8 * len(y_all)):]\n", + "\n", + "# TRAIN AND TEST DATA\n", + "X_train = X_all[idx_train]\n", + "y_train = y_all[idx_train]\n", + "X_test = X_all[idx_test]\n", + "y_test = y_all[idx_test]" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "estimator = hpsklearn.HyperoptEstimator(\n", + " preprocessing=hpsklearn.components.any_preprocessing('pp'),\n", + " classifier=hpsklearn.components.any_classifier('clf'),\n", + " algo=hyperopt.tpe.suggest,\n", + " trial_timeout=15.0, # seconds\n", + " max_evals=15,\n", + " )" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": { + "collapsed": false + }, + "outputs": [ { - "cell_type": "code", - "collapsed": false, - "input": [ - "estimator = hpsklearn.HyperoptEstimator(\n", - " preprocessing=hpsklearn.components.any_preprocessing('pp'),\n", - " classifier=hpsklearn.components.any_classifier('clf'),\n", - " algo=hyperopt.tpe.suggest,\n", - " trial_timeout=15.0, # seconds\n", - " max_evals=15,\n", - " )" - ], - "language": "python", + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEPCAYAAABRHfM8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIYQQTqhgEEII4YQKBiGEEE7+H5u6BE2js1hUAAAAAElFTkSuQmCC\n" + }, "metadata": {}, - "outputs": [], - "prompt_number": 3 - }, + "output_type": "display_data" + } + ], + "source": [ + "# Demo version of estimator.fit()\n", + "fit_iterator = estimator.fit_iter(X_train,y_train)\n", + "fit_iterator.next()\n", + "plot_helper = hpsklearn.demo_support.PlotHelper(estimator,\n", + " mintodate_ylim=(-.01, .05))\n", + "while len(estimator.trials.trials) < estimator.max_evals:\n", + " fit_iterator.send(1) # -- try one more model\n", + " plot_helper.post_iter()\n", + "plot_helper.post_loop()\n", + "\n", + "# -- Model selection was done on a subset of the training data.\n", + "# -- Now that we've picked a model, train on all training data.\n", + "estimator.retrain_best_model_on_full_data(X_train, y_train)" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": { + "collapsed": false + }, + "outputs": [ { - "cell_type": "code", - "collapsed": false, - "input": [ - "# Demo version of estimator.fit()\n", - "fit_iterator = estimator.fit_iter(X_train,y_train)\n", - "fit_iterator.next()\n", - "plot_helper = hpsklearn.demo_support.PlotHelper(estimator,\n", - " mintodate_ylim=(-.01, .05))\n", - "while len(estimator.trials.trials) < estimator.max_evals:\n", - " fit_iterator.send(1) # -- try one more model\n", - " plot_helper.post_iter()\n", - "plot_helper.post_loop()\n", + "name": "stdout", + "output_type": "stream", + "text": [ + "Best preprocessing pipeline:\n", + "PCA(copy=True, n_components=4, whiten=False)\n", "\n", - "# -- Model selection was done on a subset of the training data.\n", - "# -- Now that we've picked a model, train on all training data.\n", - "estimator.retrain_best_model_on_full_data(X_train, y_train)" - ], - "language": "python", - "metadata": {}, - "outputs": [ - { - "output_type": "display_data", - "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEPCAYAAABRHfM8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XlcU1faB/BfICiy1B0RUFFBFoGIsqgVxSqyqGittrgX\nlVo6lrrUV23nnWK1Vmttx2VqrW2x7TtVq7aCiowVpa64gUvFBZmgLIoLArIJCef9gyFjICE36w3h\n+X4++chN7j33CT7hybn3nnMFjDEGQgghRAUzvgMghBDSMlDBIIQQwgkVDEIIIZxQwSCEEMIJFQxC\nCCGcUMEghBDCCa8FIyUlBe7u7nB1dcW6deuavP748WOEhYVhwIAB8PLywo4dOwwfJCFKqMpfAIiL\ni4OrqytEIhEyMzNlzzs7O8PHxwe+vr4ICAgwVMiEaIfxRCKRsL59+zKxWMxqamqYSCRiWVlZcut8\n9NFHbPny5Ywxxh49esQ6derEamtr+QiXEDlc8vfQoUMsPDycMcZYeno6CwwMlL3m7OzMnjx5YtCY\nCdEWbz2M8+fPw8XFBc7OzrCwsEBUVBQSExPl1unevTvKysoAAGVlZejcuTOEQiEf4RIih0v+JiUl\nYfbs2QCAwMBAlJSUoKioSPY6ozGzpIXhrWAUFBSgR48esmUnJycUFBTIrRMTE4Pr16/DwcEBIpEI\nGzduNHSYhCjEJX+bW0cgEGD06NHw8/PD9u3bDRM0IVri7eu6QCBQuc6aNWswYMAApKWlIScnByEh\nIbhy5QpsbW0NECEhynHJX0B5L+LUqVNwcHDAo0ePEBISAnd3dwQFBekyREJ0jreC4ejoiLy8PNly\nXl4enJyc5NY5c+YMPvzwQwBA37590bt3b9y6dQt+fn5y67m4uCAnJ0f/QZNWqW/fvrhz547cc1zy\nt/E6+fn5cHR0BAA4ODgAALp27YpXX30V58+fb1IwKK+JPinKa5X4OnlSW1vL+vTpw8RiMXv+/LnC\nk4aLFi1i8fHxjDHGHjx4wBwdHRWeKNT2bXz00UdabU9tmHYbivKLS/6+eNL77NmzspPeFRUVrKys\njDHGWHl5ORs6dCj717/+xWm/uqCL34mh2m1Jsba0djXJL956GEKhEFu2bEFoaCikUinmzp0LDw8P\nbNu2DQAwf/58fPDBB4iOjoZIJEJdXR0+++wzdOrUia+QCZHhkr8RERFITk6Gi4sLrK2tkZCQAAB4\n8OABJk2aBACQSCSYPn06xowZw9t7IYQrXi85Cg8PR3h4uNxz8+fPl/3cpUsXHDhwwNBhEcKJqvwF\ngC1btjTZrk+fPrh8+bJeYyNEH2ikN4Dg4GBqg9poNfT1O9FHuy0p1pbYrroE/zmW1aIJBAK6pp3o\nDV/5RXlN9EmT/KIeBiGEEE6oYBBCCOGECgYhhBBOqGAQQgjhhAoGIYQQTqhgEEII4YQKBiGEEE6o\nYBBCCOGECgYhhBBOqGAQQgjhhAoGIYQQTqhgEEII4YQKBiGEEE6oYBBCCOGE14KRkpICd3d3uLq6\nYt26dU1e//zzz+Hr6wtfX194e3tDKBSipKSEh0gJIYTwdj8MqVQKNzc3HD16FI6OjvD398fOnTvh\n4eGhcP2DBw/i73//O44ePdrkNbpvANEnuh8GMUUt6n4Y58+fh4uLC5ydnWFhYYGoqCgkJiYqXf/n\nn3/G1KlTDRghIYSQF/FWMAoKCtCjRw/ZspOTEwoKChSuW1lZiX/961947bXXDBUeIYSQRoR87Vgg\nEHBe98CBAxg2bBg6dOigdJ34+HjZz8HBwUZzD1zS8qSlpSEtLY3vMAgxOrwVDEdHR+Tl5cmW8/Ly\n4OTkpHDdXbt2qTwc9WLBIEQbjb9wrFy5kr9gCDEivJ30lkgkcHNzQ2pqKhwcHBAQEKDwpHdpaSn6\n9OmD/Px8tGvXTmFbdHKQ6BOd9CamSJP84q2HIRQKsWXLFoSGhkIqlWLu3Lnw8PDAtm3bAADz588H\nAOzfvx+hoaFKiwUhhBDD4K2HoUv0TYzoE/UwiClqUT2MlqS6uhpCoRBCIf26WqIbN24gNzcXZmZm\n6NWrF9zd3fkOiZAWif4CNqOiogKTJ8/C778fhEBghkWLlmDdulVqXeFF+CEWi/Hll18iOTkZjo6O\ncHBwAGMM9+/fR35+PsaNG4dFixbB2dmZ71C19ve/A599xn39b78FIiL0Fw8xXXRIqhlz5vwFO3c+\nQXX1DwBKYW09Bl99tQSzZs3U+b6Ibr3++uuIiYlBcHAwLCws5F6rra3F8ePH8e233+KXX35R2Zax\nH5J69qz+wcXHHwOursCSJVoGR1q8FjXSuyVISzuN6uqlANoCsENFRQyOHj3Fd1iEg19++QUhISFN\nigUAWFhYYMyYMZyKRXNUzYUGAHFxcXB1dYVIJEJmZqbca1KpFL6+vhg/frxWcdjaAg4O3B49egDF\nxVrtjrRiVDCa4eDQHQLB+f8sMbRtex7Ozg68xkTUU1FRgVWrViEmJgYAkJ2djYMHD2rdrlQqxYIF\nC5CSkoKsrCzs3LkTN27ckFsnOTkZd+7cQXZ2Nr755hvExsbKvb5x40Z4enoa9BBnp05UMIjmqGA0\n4+uv1+Ollz6Gjc1k2NiMQs+eV7BkyUK+wyJqiI6ORps2bXDmzBkAgIODAz788EOt2+UyF1pSUhJm\nz54NAAgMDERJSQmKiooAAPn5+UhOTsa8efMMerirY0cqGERzVDCa4eXlhZs3M/H1169ix46/4PLl\nM2jfvj3fYRE15OTkYNmyZWjTpg0AwNraWiftcpkLrbl1Fi1ahPXr18PMzLAfQephEG3QVVIq2Nvb\nY/r06XyHQTTUtm1bVFVVyZZzcnLQtm1brdvlehipce+BMYaDBw/Czs4Ovr6+Bp+zigoG0QYVDGLS\n4uPjERYWhvz8fEybNg2nT5/Gjh07tG6Xy1xojdfJz8+Ho6Mj9u3bh6SkJCQnJ6O6uhplZWWYNWsW\nfvzxR4XxN9DFpJpUMFovXUyqSZfVEpP3+PFjpKenA6g/l9C1a1e1tleUX1zmQktOTsaWLVuQnJyM\n9PR0LFy4UBZHgz/++AOff/45Dhw4wGm/2iopAXr1AkpLddosaYFopDchjYwaNQqpqakYN25ck+e0\nwWUutIiICCQnJ8PFxQXW1tZISEhQ2JYhr5J66SWgogKorQUUXHFMSLOoh0FMUlVVFSorKzFy5Ei5\nbnhZWRnCwsJw8+ZNzm0Z+8A9dXXpAty4AajZ0SImhnoYhPzHtm3bsHHjRhQWFmLQoEGy521tbbFg\nwQIeI+Nfw3kMKhhEXdTDICZt06ZNiIuL06oNU+thDB4MfPklMGSIzpsmLQj1MAhpJC4uDn/++Sey\nsrJQXV0te37WrFk8RsUvulKKaIoKBjFp8fHx+OOPP3D9+nWMHTsWhw8fxrBhw6hgUMEgGuB1pDeX\nydvS0tLg6+sLLy8vra9BJ63P3r17cfToUXTv3h0JCQm4cuUKSkpK+A6LV1QwiKZ462E0TN529OhR\nODo6wt/fH5GRkXLXsZeUlOAvf/kL/vWvf8HJyQmPHz/mK1zSQrVr1w7m5uYQCoUoLS2FnZ2d3GC6\n1ogKBtEUbz0MLpO3/fzzz3jttddkI2i7dOnCR6ikBfP398fTp08RExMDPz8/+Pr6YujQoXyHxatO\nnYCnT/mOgrREvPUwFE3Mdu7cObl1srOzUVtbi5EjR+LZs2d47733MHMm3byIcPfVV18BAN5++22E\nhoairKwMIpGI56j4RT0MoineCgaX0a21tbXIyMhAamoqKisrMWTIEAwePBiurq5N1tX1nDukZbt0\n6ZLSHMvIyMDAgQOVbquLOXeMGRUMoineCgaXydt69OiBLl26oF27dmjXrh2GDx+OK1euqCwYhCxZ\nsgQCgQBVVVW4dOkSfHx8AABXr16Fn58fzp49q3Tbxl84Vq5cqe9wDYoKBtEUb+cw/Pz8kJ2djdzc\nXNTU1GD37t2IjIyUW2fChAk4deoUpFIpKisrce7cOXh6evIUMWlJ0tLScPz4cTg4OCAjIwOXLl3C\npUuXkJmZCQeH1n3XRCoYRFO89TC4TN7m7u6OsLAw+Pj4wMzMDDExMVQwiFpu3rwJb29v2bKXl1eT\nW6m2NlQwiKZoahBi0qKiomBjY4MZM2aAMYaff/4Z5eXl2LlzJ+c2TG1qEIkEsLQEamoAA9/wjxgR\nTfKLCgYxaVVVVdi6dStOnjwJABg+fDhiY2NhaWnJuQ1TKxgA0L49cPcu0KGDXponLQAVDEKacenS\nJbmZa7kyxYLRuzeQmgr06aOX5kkLoEl+UYeUtBoxMTF8h2A06DwG0QQVDNJqUC/0v6hgEE1QwSCt\nxkcffcR3CEaDCgbRBE1vTkze5cuXZSe9r1y50uqnBgGoYBDNUA+DmLSNGzdixowZePToER4+fIgZ\nM2Zg06ZNfIfFOyoYRBN0lRQxad7e3khPT4e1tTUAoKKiAoMHD8a1a9c4t2GKV0lt2AAUFABffKGX\n5kkLQFdJEaKA2Quj08xopBoA6mEQzag8h1FRUYEvvvgC9+7dw/bt25GdnY1bt25h3LhxhoiPEK1E\nR0cjMDAQkyZNAmMM+/fvx5w5c/gOi3dUMIgmVB6Sev311zFo0CD8+OOPuH79OioqKjB06FBcuXLF\nUDGqRIekSHMuXbqEU6dOQSAQICgoCL6+vmptb4qHpE6eBFasAE6d0kvzpAXQJL9U9jBycnLwyy+/\nYNeuXQAgOxZMSEswc+ZM/PTTT3IjvBuea82oh0E0ofKAbtu2bVFVVSVbzsnJQdu2bfUaFCG68uef\nf8otSyQSXLp0iadojAcVDKIJlQUjPj4eYWFhyM/Px7Rp0/DKK69g3bp1hoiNEI2tWbMGtra2uHbt\nGmxtbWUPOzu7JvddaY06dqwvGHQkl6iD02W1jx8/Rnp6OgAgMDAQXbt21Xtg6qBzGESZ5cuXY+3a\ntVq1YYrnMADA2hp4+LD+X9L66GW22lGjRiE1NVXlc3yigkH0yVQLRo8ewJkz9f+S1ken4zCqqqrw\n5MkTPHr0CMXFxbJHbm4uCgoKtA4WAFJSUuDu7g5XV1eFh7nS0tLQvn17+Pr6wtfXF6tXr9bJfgnR\nBVX5CwBxcXFwdXWFSCRCZmYmAKC6uhqBgYEYMGAAPD09sWLFCkOGLUPnMYjamBJffvklc3Z2Zm3a\ntGHOzs6yh7e3N9u8ebOyzTiTSCSsb9++TCwWs5qaGiYSiVhWVpbcOsePH2fjx49X2VYzb4MQrSnK\nLy75e+jQIRYeHs4YYyw9PZ0FBgbKXquoqGCMMVZbW8sCAwPZyZMnOe1Xl4KDGTt2TK+7IEZMk/xS\n2sNYuHAhxGIx1q9fD7FYLHtcvXoVCxYs0LpQnT9/Hi4uLnB2doaFhQWioqKQmJioqKBpvS/Sup08\neRIJCQkAgEePHkEsFmvdJpf8TUpKwuzZswHUn/srKSlBUVERAMDKygoAUFNTA6lUik6dOmkdk7qo\nh0HUpXIcRlxcHP78809kZWWhurpa9vysWbO02nFBQQF6vHDw1MnJCefOnZNbRyAQ4MyZMxCJRHB0\ndMTnn38OT09PrfZLWpf4+HhcunQJt27dQnR0NGpqajBjxgycPn1aq3a55K+idfLz89GtWzdIpVIM\nGjQIOTk5iI2N5SWvqWAQdaksGPHx8fjjjz9w/fp1jB07FocPH8awYcO0LhgCgUDlOgMHDkReXh6s\nrKxw+PBhTJw4Ebdv31YaZ4Pg4GAEBwdrFR8xDb/99hsyMzNlA/ccHR3x7NmzZrdJS0tDWlpas+tw\nyV+gaQ+5YTtzc3NcvnwZpaWlCA0NRVpamsKc1WdeU8FoXbjktSoqC8bevXtx5coVDBw4EAkJCSgq\nKsL06dO12ilQ/8HNy8uTLefl5cHJyUluHVtbW9nP4eHheOedd1BcXKyw+/7iB4uQBm3btpWbcLCi\nokLlNo3/MK9cubLJOlzyt/E6+fn5cHR0lFunffv2GDt2LC5evKiyYOhaw1gM0jpwyWtVVA7ca9eu\nHczNzSEUClFaWgo7Ozu5D4Gm/Pz8kJ2djdzcXNTU1GD37t1NBlQVFRXJvqGdP38ejDFejvWSlmvK\nlCmYP38+SkpK8M0332DUqFGYN2+e1u1yyd/IyEj8+OOPAID09HR06NAB3bp1w+PHj1FSUgKg/mrE\n33//Xe35rXSBehhEXSp7GP7+/nj69CliYmLg5+cHa2trDB06VPsdC4XYsmULQkNDIZVKMXfuXHh4\neGDbtm0AgPnz52Pv3r3YunUrhEIhrKysZPNZEcLV0qVLceTIEdja2uL27dtYtWoVQkJCtG6XS/5G\nREQgOTkZLi4usLa2lp14v3//PmbPno26ujrU1dVh5syZGDVqlNYxqYsKBlFXswP3GGPIy8tDz549\nAQBisRhlZWVGd4tLGrhHlFm2bFmTMRKKnmuOqQ7cO3YMWLUKOH5cb7sgRkwvN1CKiIiQ/dy7d2+j\nKxaENOfIkSNNnktOTuYhEuNDPQyirmYLhkAgwKBBg3D+/HlDxUOITmzduhXe3t64desWvL29ZQ9n\nZ2f4+PjwHZ5RoIJB1KVyLik3NzfcuXMHvXr1kt0LQyAQ4OrVqwYJkAs6JEUaKy0txdOnT7F8+XKs\nW7dOlh+2trbo3LmzWm2Z6iGp8nKgWzeAw4VjxATpZfLB3Nxchc87OzurtSN9ooJBVHn48KHcwNOG\n83JcmGrBYAxo2xYoKwMsLfW2G2Kk9HLHPWMqDISoKykpCUuWLEFhYSHs7Oxw9+5deHh44Pr163yH\nxjuBoP6w1NOnQPfufEdDWgKVJ70Jacn++te/4uzZs+jXrx/EYjFSU1MRGBjId1hGg85jEHVQwSAm\nzcLCAl26dEFdXR2kUilGjhyJixcv8h2W0aCCQdTR7CEpiUSCkJAQHKcLtUkL1bFjRzx79gxBQUGY\nPn067OzsYGNjw3dYRoMKBlFHsz0MoVAIMzMz2TQGhLQ0iYmJsLKywpdffomwsDC4uLjgwIEDfIdl\nNKhgEHWoPOltbW0Nb29vhISEyF1Wu2nTJr0HR4g2JBIJxo0bh+PHj8Pc3Bxvvvkm3yEZnYaT3oRw\nobJgTJo0CZMmTZJNy8wY4zy1MyF8erGH3KFDB77DMUrUwyDqUFkw3nzzTTx//lx2Hwp3d3dYWFjo\nPTBCdKGhhzxmzBjZXe6oh/xfnToBf/7JdxSkpVBZMNLS0jB79mz06tULAHDv3j388MMPGDFihN6D\nI8aHMYbKykpYWVm1iJ4m9ZCbRz0Mog6VBWPx4sU4cuQI3NzcAAC3b99GVFQUMjIy9B4cMS4XL17E\nuHGv4/HjQrRv3wX79+9EUFAQ32E1i85bNI8KBlGHynEYEolEViwAoF+/fpBIJHoNihifyspKjBkz\nAUVFn0EqrUJx8bcYO3YyiumvTYtGBYOoQ2XBGDRoEObNm4e0tDQcP34c8+bNg5+fn052npKSAnd3\nd7i6ujZ7f4ILFy5AKBTi119/1cl+ifpycnIgkbQHMBmAAEAYzMx648aNGzxHRrRBBYOoQ2XB+Prr\nr+Hh4YFNmzZh8+bN6N+/P7Zu3ar1jqVSKRYsWICUlBRkZWVh586dCv/4SKVSLFu2DGFhYTTBII/s\n7OxQU1MIoOA/zzxGTc2/0d2IJyGSSqV4//33+Q7DqNF9vYk6VI70FolEuHnzJpYsWaLTHZ8/fx4u\nLi6yyQ2joqKQmJgIDw8PufU2b96MyZMn48KFCzrdP1FPt27d8NFHf8Xq1YNhZhYMxk5hwYJY9OnT\nh+/QlDI3N8epU6foRHcz2revn+ZcIgGEKs9oktau2RQRCoVwc3PD3bt3ZVdJ6UpBQQF69OghW3Zy\ncsK5c+earJOYmIhjx47hwoUL9KHn2YoV72P06BG4fv06+vWL1cm93fVtwIABmDBhAqZMmSJ3We2k\nSZN4jsw4mJnVF42SEqBLF76jIcZO5XeK4uJi9O/fHwEBAXIjvZOSkrTaMZc//gsXLsTatWtl87bT\nISn++fv7w9/fn+8wOKuurkanTp1w7NgxueepYPxXw3kMKhhEFZUFY/Xq1U3+UOvim76joyPy8vJk\ny3l5eXBycpJb59KlS4iKigIAPH78GIcPH4aFhQUiIyObtBcfHy/7OTg4GMHBwVrHSFq+HTt2qL1N\nWloa0tLSdB6LsaIT34SrZu+4J5FI0L9/f9y6dUvnO264XDc1NRUODg4ICAjAzp07m5zDaBAdHY3x\n48cr/GZId9wjyuTl5SEuLg6nTp0CAAwfPhwbN25s8uWkOaZ6x70G4eHAu+8CERF63xUxIprkl8rZ\nat3d3XH37l2tAlPW9pYtWxAaGgpPT0+88cYb8PDwwLZt27Bt2zad74+0TtHR0YiMjERhYSEKCwsx\nfvx4REdH8x2WUaEeBuFK5T29g4KCkJmZqfNzGLpEPQyijEgkwpUrV1Q+1xxT72G8+y7g6grExel9\nV8SI6OWe3qtWrVK4I0Jags6dO+Onn37CtGnTwBjDrl270IXO7sqhHgbhSuXAveDgYDg7O0MikSA4\nOBgBAQHw9fU1RGyEaC0hIQG//PIL7O3t0b17d+zZswcJCQl8h2VUqGAQrlT2ML755hts374dxcXF\nyMnJQX5+PmJjY5GammqI+AjRmEQiwQcffEB32FOhUyeAxsUSLlT2MP7xj3/g1KlTeOmllwDUTz74\n8OFDvQdGiLaEQiHu3r2L58+f8x2KUaMeBuFKZcFo27Yt2rZtK1uWSCR0DoO0GL1798awYcOwatUq\nbNiwARs2bMAXX3yhk7a5TJ4ZFxcHV1dXiEQiZGZmAqi/1HfkyJHo378/vLy8eL+ZExUMwpXKQ1Ij\nRozAJ598gsrKSvz+++/46quvMH78eEPERojWXFxc0LdvX9TV1aG8vFxn7TZMnnn06FE4OjrC398f\nkZGRcuOIkpOTcefOHWRnZ+PcuXOIjY1Feno6LCws8OWXX2LAgAEoLy/HoEGDEBISonQMkr7Rfb0J\nVyoLxtq1a/Hdd9/B29sb27ZtQ0REBObNm2eI2AjRikQiwa1bt/Dzzz/rvG0uk2cmJSVh9uzZAIDA\nwECUlJSgqKgI9vb2sLe3BwDY2NjAw8MDhYWFvBYM6mEQLlQWDHNzc7z11lt46623DBEPITojFApx\n7949PH/+XO6wqi5wnTyz8Tr5+fno1q2b7Lnc3FxkZmYiMDBQp/Gpo2PH+h5GXV39ZISEKEMTGhOT\n1nAOIzIyUm622sWLF2vVLtfzeM3Nw1ZeXo7Jkydj48aNsLGxUbi9IeZIEwoBa2vg2bP6mWuJadLF\nHGlUMIhMeXk5Dh8+jNraWoSEhKBr1668xPHgwQOkpqbC0tIS4eHhsj/0mujbt69ezmFwmTyz8Tr5\n+flwdHQEANTW1uK1117DjBkzMHHiRKX7ebFg6FPDYSkqGKar8ReOlStXqt8IMwEm8jZ49fjxY+bs\n7MlsbEKYjc2rrGNHB3br1i2Dx3Ht2jXWvr09s7F5jdnYjGSuriJWUlKidbvl5eUab6sov2pra1mf\nPn2YWCxmz58/ZyKRiGVlZcmtc+jQIRYeHs4YY+zs2bMsMDCQMcZYXV0dmzlzJlu4cKHa+9WXgQMZ\nu3jRYLsjRkCT/FJ5xPLWrVuIiYlBSEgIRo4ciZEjR+KVV15RvzIRo7ZmzXoUFgahvPwIyst/RWnp\nEixYsNzgcbz99lKUlf0N5eV7UV6einv3RNiw4e8at3fmzBl4enrC3d0dAHDlyhW88847WsfJZfLM\niIgI9OnTBy4uLpg/fz6++uorAMDp06fxf//3fzh+/Dh8fX3h6+uLlJQUrWPSBp34JlyoPCQ1ZcoU\nxMbGYt68eTA3NwdAc0mZotzc+6ipGSFbrqsLQH7+HoPHUVh4H4wF/GdJgOfPA5Cbe1Xj9hYuXIiU\nlBRMmDABQP3Eg3/88YcOIgXCw8MRHh4u99z8+fPllrds2dJku2HDhqGurk4nMegK3dubcKGyYFhY\nWCA2NtYQsRAehYS8jJSUr1FZOQGAFSwtN2DUqGEGjyM4+GXcv78e1dU/ACiFtfV2jB6t3f3ke/bs\nKbcspJtXN0E9DMKFykNS48ePxz/+8Q/cv38fxcXFsgcxLfPnx2Du3BEQCh1hbt4BY8ZYYv36pjMV\n69vmzZ8hOLgW5uYvQSjsiXfeGYeZM2do3F7Pnj1x+vRpAEBNTQ0+//xz3sY7GDMqGIQLlffDcHZ2\nbnIISiAQ4N///rdeA1MH3Q9DdyQSCaRSqc7HLairuroaQqFQ697Ao0eP8N577+Ho0aNgjGHMmDHY\ntGkTOnfuzLkNU78fBgB8/jlw/z6wYYNBdkeMgCb5pbJg6FNKSgoWLlwIqVSKefPmYdmyZXKvJyYm\n4m9/+xvMzMxgZmaG9evXKzzhTgWD6FNrKBjffw+cPAnQzO+th14KRk1NDbZu3YoTJ05AIBBgxIgR\nePvtt2FhYaFVsFKpFG5ubnJz8TS+p3dFRYXsLn/Xrl3Dq6++ijt37jR9E1QwiB61hoKxf399sUhM\nNMjuiBHQ+T29ASA2NhYZGRn4y1/+gtjYWFy6dEknJ8FfnIvHwsJCNhfPixqKBVA/qIzulEaIftA5\nDMKFygPEFy5cwNWr/72scdSoUfDx8dF6x1zm4gGA/fv3Y8WKFbh//z6OHDmi9X4JIU1RwSBcqCwY\nQqEQd+7cgYuLCwAgJydHJ5clch3LMXHiREycOBEnT57EzJkzcevWLa33TVqP6upq7Nu3D7m5uZBI\nJADqc+9vf/sbz5EZFyoYhAuVf/kbTjT37t0bQP3smrq4JzKXuXheFBQUBIlEgidPnii8wsUQk7SR\nlmfChAno0KEDBg0aBEtLS07b6GKStpamYeAeYwCNyyXKcLpKqrq6Grdu3YJAIICbm5tOLrmUSCRw\nc3NDamoqHBwcEBAQ0OSkd05ODvr06QOBQICMjAxMmTIFOTk5Td8EnfQmSnh5eeHPP//Uqo3WcNIb\nAKysgEeP6meuJaZPk/xS2sNITU3FqFGjsG/fPrmGG65SmjRpkhahys/FI5VKMXfuXNlcPED9FAv7\n9u3Djz8fxPfAAAAgAElEQVT+CAsLC9jY2GDXrl1a7ZO0PkOHDsXVq1d1ct7N1DUclqKCQZRR2sP4\n6KOPsHLlSrz55psKzzfo4rCUrlAPgyjj4eGBO3fuoHfv3rKesUAgkLuQQ5XW0sPw8QF++gkQiQy2\nS8IjvYzD+Pe//40+ffqofI5PVDCIMrm5uQD+e5FFQ5403FqVi9ZSMIKDgfj4+n+J6dPLOIzJkyc3\neW7KlClq7YQQvjg7O6OkpARJSUk4cOAASktL1SoWrQldKUVUUXoO48aNG8jKykJJSQl+/fVXMMYg\nEAhQVlaG6upqQ8ZIiMY2btyI7du3Y9KkSWCMYcaMGYiJiUFcXBzfoRkdKhhEFaUF4/bt27JvZAcO\nHJA9b2tri+3btxskOEK09e233+LcuXOyWQOWL1+OwYMHU8FQgAoGUUVpwZgwYQImTJiAM2fOYOjQ\noYaMiRCdMjMzU/gzkUcFg6iicuCer68vtmzZgqysLFRVVclOHn7//fd6D84UMMbw9dfbsWvXQXTq\n9BJWr16B/v378x1WqxEdHY3AwEDZIan9+/djzpw5fIdllDp1AozorgXECKn8ujVz5kwUFRUhJSUF\nwcHByMvLg42NjSFiMwmffPIZ3n9/C06ceBOJib4YPHikUd1LxNQtXrwYCQkJ6NixIzp37owdO3Zg\n0aJFfIdllKiHQVRReVntgAEDcPnyZfj4+ODq1auora3FsGHDFE4UyBdjvqy2a1dnPH58EIAXAEAo\njEN8vD0+/PADfgMzcWVlZXjppZdkd4dsyI+GHnKnTp04t9VaLqtNTQU++QQ4dsxguyQ80ulI7wZt\n2rQBALRv3x7Xrl2Dvb09Hj16pFmEhBjI1KlTcejQIQwcOFDhwFOxWMxDVMaNehhEFZUFIyYmBsXF\nxVi9ejUiIyNRXl6OVasMf6/nluq992Lx6afTUFkZD4FADEvLXZg6NZ3vsEzeoUOHAPx34B5RjQoG\nUYXXW7TqijEfknrxpHfnzu2xevUKeHp68h1WqzFq1CikpqaqfK45reWQ1LNnQPfuQHm5wXZJeKTT\nqUE2vHA3+IaGX+zaL168WMMwdc+YCwbhR1VVFSorKzFy5Ei5qcrLysoQFhaGmzdvcm6rtRQMxoA2\nbeoLhg4mpCZGTqfnMJ49ewaBQIBbt27hwoULiIyMBGMMBw8eREBAgNbBEqJP27Ztw8aNG1FYWIhB\ngwbJnre1tcWCBQt4jMx4CQT1h6WePgXs7fmOhhgjlYekgoKCkJycDFtbWwD1hSQiIgInT540SIBc\nUA+DKLNp0yatR3W3lh4GAHh4APv2AXTU1PTp5Sqphw8fwsLCQrZsYWGBhw8fqh8dITyIi4vDn3/+\niaysLLk50GbNmsVjVMaLTnyT5qgsGLNmzUJAQIDcSNnZs2cbIjZCtBYfH48//vgD169fx9ixY3H4\n8GEMGzaMCoYSVDBIc1SO9P7www+RkJCADh06oFOnTtixYwc++EA3g85SUlLg7u4OV1dXrFu3rsnr\n//znPyESieDj44OXX35ZrZveEAIAe/fuxdGjR9G9e3ckJCTgypUrKCkp0UnbqvIXqO/huLq6QiQS\nITMzU/b8nDlz0K1bN3h7e+skFl2hgkGaxZQoLS1ljDH25MkT9uTJE/b48WP2+PFj2bK2JBIJ69u3\nLxOLxaympoaJRCKWlZUlt86ZM2dYSUkJY4yxw4cPs8DAQIVtNfM2SCvn5+fHGGNs4MCBrKSkhNXV\n1bF+/fqp1Yai/OKSv4cOHWLh4eGMMcbS09Pl8vfEiRMsIyODeXl5qbVffVu4kLENGwy+W8IDTfJL\n6SEpfY+UPX/+PFxcXGQ3s4mKikJiYiI8PDxk6wwZMkT2c2BgIPLz87XaJ2l9/P398fTpU8TExMDP\nzw/W1tY6mX2ZS/4mJSXJDt8GBgaipKQEDx48gL29PYKCgoxyUCH1MEhzlBYMfY+ULSgoQI8ePWTL\nTk5Ozc5P9d133yEiIkIvsRDT9dVXXwEA3n77bYSGhqKsrAwiHdy0mkv+KlqnoKAA9kZ8zWqnTsD1\n63xHQYyV0oKRkZHR7IYDBw7UaseKei3KHD9+HN9//z1Onz6tdJ34+HjZz8HBwQimGxO3apcuXVKa\nYxkZGc3mb1pamtxgP0W45i9rdNmiOnkPGD6vqYdhurjktSpKC8bixYubTe7jx49rtWNHR0fk5eXJ\nlvPy8uDk5NRkvatXryImJgYpKSno2LGj0vZe/GARsmTJEggEAlRVVeHSpUvw8fEBUJ9Pfn5+OHv2\nrNJtG/9hXrlyZZN1uORv43Xy8/Ph6Oio1vswdF43DNwjpodLXquk+1Mp3NTW1rI+ffowsVjMnj9/\nrvCk4d27d1nfvn3Z2bNnm22Lx7dBjNyrr77Krl69Klu+du0amzRpklptKMovLvn74knvs2fPNrlo\nQywWG91J7/PnGfvPdQLExGmSXyrHYQDAtWvXcOPGDZ0OfBIKhdiyZQtCQ0MhlUoxd+5ceHh4YNu2\nbQCA+fPn4+OPP8bTp08RGxsLoH7Q4Pnz57XaL2ldbt68KXfpqpeXF27cuKF1u1zyNyIiAsnJyXBx\ncYG1tTUSEhJk20+dOhV//PEHnjx5gh49euDjjz9GdHS01nFpiw5JkeaonBpE2cCnvXv3GipGlWhq\nEKJMVFQUbGxsMGPGDDDG8PPPP6O8vBw7d+7k3EZrmhrk6VOgTx86LNUa6HS22gZeXl64cuUKBg4c\niCtXrqCoqAjTp0/H0aNHtQpWl6hgEGWqqqqwdetW2dxnw4cPR2xsLCwtLTm30ZoKRl1d/Yy1z58D\n5uYG3TUxML3MJdWuXTuYm5tDKBSitLQUdnZ2cifyCDFm7dq1w+LFi41qOn5jZmYGtG8PlJQAnTvz\nHQ0xNioLhp+fn14GPhGiT1OmTMGePXsUTr0hEAhomplmNJzHoIJBGlN6SOqdd97BtGnTMGzYMNlz\nYrFYZwOfdIkOSZHGCgsL4eDgoHTgacMIbS5a0yEpAAgIADZvBgIDDb5rYkA6PSTVr18/LF26FIWF\nhXjjjTcwdepU+Pr6ah0kIYbg4OAAQL3CQOrRlVJEGZUnvXNzc7Fr1y7s3r0blZWVmDZtGqZOnYp+\n/foZKkaVqIdBGrOxsVE68FQgEKCsrIxzW62thzFtGjB2LDB9usF3TQxIL1dJvSgzMxPR0dG4du0a\npFKp2gHqCxUMok+trWAsWAC4uQHvvmvwXRMD0iS/VN4PQyKRICkpCdOmTUNYWBjc3d3x66+/ahwk\nIXx4+PAh7t27J3sQ5eiQFFFG6TmMI0eOYNeuXTh06BACAgIwdepUfPPNN7CxsTFkfIRoJSkpCUuW\nLEFhYSHs7Oxw9+5deHh44DpNyapUp06AlncvICZKaQ9j7dq1GDJkCG7cuIEDBw5g2rRpVCxIi/PX\nv/4VZ8+eRb9+/SAWi5GamopAuvynWdTDIMoo7WEcO3bMkHEQohcWFhbo0qUL6urqIJVKMXLkSLz3\n3nt8h2XUqGAQZThNPkhIS9WxY0c8e/YMQUFBmD59Ouzs7KinrAIVDKKMypPehLREe/bsQXV1NRIT\nE2FlZYUvv/wSYWFhcHFxwYEDB/gOz6hRwSDKqHVZrbGiy2pJYxMnTsTp06cRFhaGqVOnIjQ0FOYa\nzqbX2i6rffgQ6N8fePTI4LsmBqT3cRjGigoGUaS0tBS//fYbdu3ahcuXL2PixImYOnUqRowYoVY7\nra1g1NYC7doBNTX1kxES00QFgxAlHj9+jH379uEf//gHiouLkZ+fz3nb1lYwAOCll4C8vPqZa4lp\n0svAPX1KSUmBu7s7XF1dsW7duiav37x5E0OGDIGlpSU2bNjAQ4TEFDx9+hS//vordu/ejeLiYkyZ\nMoXvkIwe3dubKMLbVVJSqRQLFizA0aNH4ejoCH9/f0RGRsLDw0O2TufOnbF582bs37+frzBJC/Xs\n2TPZ4aiMjAxERkbif//3fxEcHKx0jinyXw0nvmnuRvIi3grG+fPn4eLiIptNNCoqComJiXIFo2vX\nrujatSsOHTrEU5SkpXJ2dkZYWBjeeecdjBkzBm3atOE7pBaFrpQiivBWMAoKCtCjRw/ZspOTE86d\nO8dXOMTE5Ofno127dnyH0WJRwSCK8FYwdH1YID4+XvZzcHAwgoODddo+aVkmT56MN998E2PHjoWV\nlZXcaxUVFTh48CB++OEHJCcnN9k2LS0NaWlpBorUOFHBIIrwVjAcHR3l7g2el5cHJycnjdt7sWC0\nRs+fP0dycjIqKioQHBys1e/SFCQkJGDLli346KOPYG5uju7du4MxhgcPHkAikeCNN97ADz/8oHDb\nxl84Vq5caaCojQcVDKIIbwXDz88P2dnZyM3NhYODA3bv3o2dO3cqXJcumW1eZWUlhgwZjX//2xyA\nA4DFOHbsEPz9/fkOjTd2dnb4+OOP8fHHH+PBgwe4e/cuAKBXr16wt7fnOTrj16kT8OAB31EQY8Nb\nwRAKhdiyZQtCQ0MhlUoxd+5ceHh4YNu2bQCA+fPn48GDB/D390dZWRnMzMywceNGZGVl0VxAjWzb\ntg23b9ujunofAAGAf2Lu3IW4evU036EZBXt7eyoSaurYEcjK4jsKYmx4nXwwPDwc4eHhcs/Nnz9f\n9rO9vb3cYSuiWF7efVRX+6O+WABAAB48+F8+QzIa+/btw/Lly1FUVCTrqap7i9bWiA5JEUVo4L8J\nGDHiZVhZ7QCQD6AWbdqsw7BhL/MclXH4n//5HyQlJaGsrAzPnj3Ds2fPqFhwQAWDKEIFwwRMmDAB\nK1bMgYVFP5ib22DIkPv4/vvNfIdlFOzt7eXG9hBuqGAQRWguKRMilUpRW1sLS0tLvkMxGu+99x4e\nPHiAiRMnygbvCQQCTJo0iXMbrXEuqYICwN8fKCzkZffEAFrcXFJEt8zNzalYNFJaWop27drhyJEj\nOHjwIA4ePKiz+2GomgsNAOLi4uDq6gqRSITMzEy1tuVTQw+DvocROcwEmMjbIEZKUX5JJBLWt29f\nJhaLWU1NDROJRCwrK0tunUOHDrHw8HDGGGPp6eksMDCQ87bK9mtIlpaMVVTwGgLRI03yi27RSkzS\nunXrsGzZMrz77rtNXhMIBNi0aZNW7XOZCy0pKQmzZ88GAAQGBqKkpAQPHjyAWCxWua0xaOhlNBoo\nT1qxVl8wGGOorKyEtbU136GYjJqaGggEAlhYWPAWg6enJwBg0KBBTV7TxbQ0XOZCU7ROQUEBCgsL\nW8Q8ag0Fo5VPGkBe0KoLRmpqKiZPnoFnz56iW7ceSE7eC5FIxHdYLVZtbS1mz34bv/zyfxAIBJg5\ncw62b9+s8a1RtTF+/HgAwJtvvqmX9rkWHdaCTwJ07w688gpAE/2SBq22YBQVFWHChChUVPwCIBiF\nhT9j9OhIFBRk01TYGvr447VITMyDVPoEQB12745Ev35fYvny93mL6cKFC1izZg1yc3MhkUgA1P+x\nv3r1qlbtcpkLrfE6+fn5cHJyQm1tLed51PicVHP/fqCkxGC7I3p25kwazp5Nky1/8YUGjej8TAoP\nNHkbv//+O2vfPpjVXwdS/7C2dmbZ2dl6iLB1CAwcw4BDL/xO97CRIyfwGpOrqytLTExkOTk5TCwW\nyx7qUJRftbW1rE+fPkwsFrPnz5+rPOl99uxZ2UlvLtsq2y8huqJJfrXaHkb37t1RW3sbQAmADgDu\nQiJ5gq5du/IcWcvVs2d3XLx4DlJpBABAKDyPXr268xpT165dERkZqfN2ucyFFhERgeTkZLi4uMDa\n2hoJCQnNbkuIsWvVA/fi4v4H33//K4AhYOwY1qz5AO+99xfdB9hK3Lt3D/7+w1FZ6QNAChubm8jI\nOIXu3fkrGkeOHMHu3bsxevRoGrhHyAs0ya9WXTAA4OTJk8jJyYGPjw8GDhyo48han+LiYhw5cgQC\ngQBhYWFo3749r/FMnz4dt27dQv/+/WFm9t9xqg3f9rmggkFMERUMQhpxc3PDzZs3tbqUlgoGMUU0\nNQghjQwdOhRZdGMHQnSCehjEpLm7uyMnJwe9e/dG27ZtAah/WS31MIgp0iS/eL1KKiUlBQsXLoRU\nKsW8efOwbNmyJuvExcXh8OHDsLKywo4dO+Dr68tDpKSlSklJ4TsEQkwGb4ekpFIpFixYgJSUFGRl\nZWHnzp24ceOG3DrJycm4c+cOsrOz8c033yA2NpanaJUrKCjAzJlvIShoHFauXIPa2lq127hz5w5e\nf/1NDB8+Hl98sQl1dXVyr0ulUqxbtwFBQeMQFTUHubm5au+jpqYGH364EsOGjcWbb8aiqKhI7TZ0\nobi4GG+9FYdhw8bi/fc/QFVVldptqPqdM8awdes3GDEiEosW/RUVFRVwdnaWexBCNKD5sA/tnDlz\nhoWGhsqWP/30U/bpp5/KrTN//ny2a9cu2bKbmxt78OBBk7b4ehslJSXM3r4PEwqXM2A/s7IazaZO\nnaNWGwUFBaxDh+7MzOwTBvzGrKz82fvvfyC3TmzsImZl9TID9jNz83jWubMTe/jwoVr7mTBhKmvX\nLpwBiUwoXMx69HBj5eXlarWhrerqatavny9r0yaWAYnM0vI1NnLkWFZXV8e5DS6/81Wr1jIrK28G\n7GMCwefMxqYry8nJ0ThuvvKLx48naQU0yS/eMnLPnj1s3rx5suWffvqJLViwQG6dcePGsdOnT8uW\nR40axS5evNikLb4+WHv27GG2tmEvjGx+xszN27KqqirObWzevJlZWs5+oY17zMqqo+z1uro6ZmHR\njgEPZetYWb3Ovv32W877KCkpYUKhFQMqZW3Y2o5gBw8eVOv9auvEiRPM1taXAXX/iaOGWVp2Zffu\n3ePcBpffeZcuvRhwTbaOUPguW736E43jpoJBTJEm+cXbOQxNJ29Tth2fc+4Q05KWloa0tDS+wyDE\n+Oi+bnFz9uxZuUNSa9asYWvXrpVbZ/78+Wznzp2y5dZ6SOqddxoOSf2mg0NS+5mFxWLWs6c7HZLi\niK/84vHjSVoBTfKLt4zUZvK2xvj8YBUUFLAZM2LYsGFjWXz8J6y2tlbtNrKzs9mUKbNZUNA4tmHD\nRiaVSuVel0gkbO3az9mwYWPZG29Eqz15HmOMPX/+nH3wQTx7+eUINnv22woLryE8efKExcS8y15+\nOYK9//4HrLKyUu02VP3O6+rq2FdfbWPDh49nr746g12/fl2rmKlgEFOkSX7xOg7j8OHDsstq586d\nixUrVshN3gZAdiVVw+RtiqbvoOvViT7ROAxiimhqEEL0gAoGMUU0NQghhBC9oYJBCCGEEyoYhBBC\nOKGCQQghhBMqGIQQQjihgkEIIYQTKhiEEEI4oYJBCCGEEyoYhBBCOKGCQQghhBMqGIQQQjihgkEI\nIYQTKhiEEEI4oYJBCCGEEyoYhBBCOOGlYBQXFyMkJAT9+vXDmDFjUFJSonC9OXPmoFu3bvD29jZw\nhIQoxzV/U1JS4O7uDldXV6xbt072/J49e9C/f3+Ym5sjIyPDUGETojVeCsbatWsREhKC27dvY9So\nUVi7dq3C9aKjo5GSkqL3eNLS0qgNaoMzLvkrlUpld4vMysrCzp07cePGDQCAt7c3fvvtNwwfPlzn\nsXGhj9+JvtptSbG2xHbVxUvBSEpKwuzZswEAs2fPxv79+xWuFxQUhI4dO+o9HmP5w0RtGGcbjXHJ\n3/Pnz8PFxQXOzs6wsLBAVFQUEhMTAQDu7u7o16+fzuPiqiX9UWtJsbbEdtXFS8EoKipCt27dAADd\nunVDUVERH2EQohEu+VtQUIAePXrIlp2cnFBQUGCwGAnRB6G+Gg4JCcGDBw+aPP/JJ5/ILQsEAggE\nAn2FQYhGGudvw3k0rvlLOU1MEuOBm5sbu3//PmOMscLCQubm5qZ0XbFYzLy8vJptr2/fvgwAPeih\nl0ffvn3Vzt+zZ8+y0NBQ2fKaNWvY2rVr5dYJDg5mly5dorymBy+PxnnNhd56GM2JjIzEDz/8gGXL\nluGHH37AxIkTtWrvzp07OoqMENW45K+fnx+ys7ORm5sLBwcH7N69Gzt37myyHmNM6X4or4nRUbvE\n6MCTJ0/YqFGjmKurKwsJCWFPnz5ljDFWUFDAIiIiZOtFRUWx7t27szZt2jAnJyf2/fff8xEuIXK4\n5m9ycjLr168f69u3L1uzZo3s+V9//ZU5OTkxS0tL1q1bNxYWFmbw90CIJgSMNfMVhxBCCPkPkxjp\nvXTpUnh4eEAkEmHSpEkoLS3lvK2ywVVc5eXlYeTIkejfvz+8vLywadMmtdtoIJVK4evri/Hjx2u0\nfUlJCSZPngwPDw94enoiPT1d7TY+/fRT9O/fH97e3pg2bRqeP3+uchtFAyy5Dm5rrg11/l+bG+S5\nYcMGmJmZobi4WO0YAGDz5s3w8PCAl5cXli1b1mwbuqJtXiqiy1xtTNvcVUQX+ayIJjnemC5ynmu7\n2vx9a67dBlw/HwD4OSSla0eOHGFSqZQxxtiyZcvYsmXLOG0nkUhY3759mVgsZjU1NUwkErGsrCy1\n9n3//n2WmZnJGGPs2bNnrF+/fmq30WDDhg1s2rRpbPz48RptP2vWLPbdd98xxhirra1lJSUlam0v\nFotZ7969WXV1NWOMsddff53t2LFD5XYnTpxgGRkZchcnLF26lK1bt44xxtjatWtV/p8oakOd/1dF\n2zPG2L1791hoaChzdnZmT548UTuGY8eOsdGjR7OamhrGGGMPHz5stg1d0EVeKqLLXG1M29xVRNt8\nVkTTHG9MFznPtV1N/76papcx9T4fjDFmEj2MkJAQmJnVv5XAwEDk5+dz2q65wVVc2dvbY8CAAQAA\nGxsbeHh4oLCwUL03ACA/Px/JycmYN29esydClSktLcXJkycxZ84cAIBQKET79u3VauOll16ChYUF\nKisrIZFIUFlZCUdHR5XbKRpgyXVwZnNtqPP/qmyQ5+LFi/HZZ5+pfA/K2ti6dStWrFgBCwsLAEDX\nrl05taUNXeSlIrrK1ca0zV1FdJHPimia443pIue5tqvp3zdV7QLqfT4AEzkk9aLvv/8eERERnNbV\n9eCq3NxcZGZmIjAwUO1tFy1ahPXr18sSQ11isRhdu3ZFdHQ0Bg4ciJiYGFRWVqrVRqdOnbBkyRL0\n7NkTDg4O6NChA0aPHq1RPLoenKnO/2uDxMREODk5wcfHR+P9Zmdn48SJExg8eDCCg4Nx8eJFjdvi\nyhCD/rTJ1ca0zV1FdJHPiugyxxszxIBkTT4Hymjy+WgxBSMkJATe3t5NHgcOHJCt88knn6BNmzaY\nNm0apzZ1ObiqvLwckydPxsaNG2FjY6PWtgcPHoSdnR18fX01/oYmkUiQkZGBd955BxkZGbC2tlY6\nR5cyOTk5+Pvf/47c3FwUFhaivLwc//znPzWK50XaDs5U9/8VACorK7FmzRqsXLlS9pwmv1uJRIKn\nT58iPT0d69evx+uvv652G+rS96A/bXK1MV3kriK6yGdF9JXjjeljQLImnwNlNP18tJiC8fvvv+Pa\ntWtNHg0n2Xbs2IHk5GS1/vMdHR2Rl5cnW87Ly4OTk5PasdXW1uK1117DjBkzNBpTcubMGSQlJaF3\n796YOnUqjh07hlmzZqnVhpOTE5ycnODv7w8AmDx5stozoV68eBFDhw5F586dIRQKMWnSJJw5c0at\nNhp069ZNNlL6/v37sLOz06gdTf5fgfo/DLm5uRCJROjduzfy8/MxaNAgPHz4UK12nJycMGnSJACA\nv78/zMzM8OTJE7XaUJeu8lIRbXO1MV3kriK6yGdFdJnjjekq5xXR9HOgjKafjxZTMJqTkpKC9evX\nIzExEZaWlpy3e3FwVU1NDXbv3o3IyEi19s0Yw9y5c+Hp6YmFCxeqGzoAYM2aNcjLy4NYLMauXbvw\nyiuv4Mcff1SrDXt7e/To0QO3b98GABw9ehT9+/dXqw13d3ekp6ejqqoKjDEcPXoUnp6earXRoGFw\nGwCNB2dq+v8K1E/lUVRUBLFYDLFYDCcnJ2RkZKj9IZ44cSKOHTsGALh9+zZqamrQuXNntdpQly7y\nUhFd5GpjushdRXSRz4roMscb00XOK6LN50AZjT8fap9uN0IuLi6sZ8+ebMCAAWzAgAEsNjaW87bK\nBldxdfLkSSYQCJhIJJLt//Dhw2q30yAtLU3jK00uX77M/Pz8mI+PD3v11Vc1uqpk3bp1zNPTk3l5\nebFZs2bJrg5qTsMASwsLC9kAS2WD27i28d1336n1/6pqkGfv3r1VXgWiqI2amho2Y8YM5uXlxQYO\nHMiOHz+u8vehC9rmpSK6ztXGtMldRXSRz4pokuON6SLnubSr7udAVbvafD4Yo4F7hBBCODKJQ1KE\nEEL0jwoGIYQQTqhgEEII4YQKBiGEEE6oYBBCCOGECgYhhBBOqGAYuYapG+7evavwjm3aWLNmjdzy\nyy+/rNP2CWkO5XbLQwXDyDXMRyMWi/Hzzz+rta1EImn29U8//VRu+fTp0+oFR4gWKLdbHioYLcTy\n5ctx8uRJ+Pr6YuPGjairq8PSpUsREBAAkUiEb775BgCQlpaGoKAgTJgwAV5eXgDqp7fw8/ODl5cX\ntm/fLmuvqqoKvr6+mDlzJoD/fuNjjGHp0qXw9vaGj48PfvnlF1nbwcHBmDJlCjw8PDBjxgxD/xqI\nCaLcbkHUHmNODMrGxoYxVj/twrhx42TPb9u2ja1evZoxxlh1dTXz8/NjYrGYHT9+nFlbW7Pc3FzZ\nusXFxYwxxiorK5mXl5dsuaHtxvvau3cvCwkJYXV1dayoqIj17NmT3b9/nx0/fpy1b9+eFRQUsLq6\nOjZkyBB26tQp/b15YtIot1se6mG0EKzRDC5HjhzBjz/+CF9fXwwePBjFxcW4c+cOACAgIAC9evWS\nrbtx40YMGDAAQ4YMQV5eHrKzs5vd16lTpzBt2jQIBALY2dlhxIgRuHDhAgQCAQICAuDg4ACBQIAB\nA1GUNO8AAAFgSURBVAYgNzdX5++VtC6U2y2HkO8AiOa2bNmCkJAQuefS0tJgbW0tt5yamor09HRY\nWlpi5MiRqK6ubrZdgUDQ5EPccLy5bdu2sufMzc1VHksmRBOU28aJehgthK2tLZ49eyZbDg0NxVdf\nfSVL6tu3byu8I1lZWRk6duwIS0tL3Lx5E+np6bLXLCwsFH4ogoKCsHv3btTV1eHRo0c4ceIEAgIC\ndHqDHEIaUG63HNTDMHIN335EIhHMzc0xYMAAREdHIy4uDrm5uRg4cCAYY7Czs8Nvv/3W5E5fYWFh\n+Prrr+Hp6Qk3NzcMGTJE9tpbb70FHx8fDBo0CD/99JNsu1dffRVnz56FSCSCQCDA+vXrYWdnhxs3\nbjS5i5i+7w5HTBfldstD05sTQgjhhA5JEUII4YQKBiGEEE6oYBBCCOGECgYhhBBOqGAQQgjhhAoG\nIYQQTqhgEEII4YQKBiGEEE7+H5u6BE2js1hUAAAAAElFTkSuQmCC\n" - } - ], - "prompt_number": 4 - }, - { - "cell_type": "code", - "collapsed": false, - "input": [ - "print 'Best preprocessing pipeline:'\n", - "for pp in estimator._best_preprocs:\n", - " print pp\n", - "print\n", - "print 'Best classifier:\\n', estimator._best_classif\n", - "test_predictions = estimator.predict(X_test)\n", - "acc_in_percent = 100 * np.mean(test_predictions == y_test)\n", - "print\n", - "print 'Prediction accuracy in generalization is %.1f%%' % acc_in_percent" - ], - "language": "python", - "metadata": {}, - "outputs": [ - { - "output_type": "stream", - "stream": "stdout", - "text": [ - "Best preprocessing pipeline:\n", - "PCA(copy=True, n_components=4, whiten=False)\n", - "\n", - "Best classifier:\n", - "KNeighborsClassifier(algorithm=auto, leaf_size=72, metric=euclidean,\n", - " n_neighbors=29, p=2, weights=uniform)\n", - "\n", - "Prediction accuracy in generalization is 96.7%\n" - ] - } - ], - "prompt_number": 5 + "Best classifier:\n", + "KNeighborsClassifier(algorithm=auto, leaf_size=72, metric=euclidean,\n", + " n_neighbors=29, p=2, weights=uniform)\n", + "\n", + "Prediction accuracy in generalization is 96.7%\n" + ] } ], - "metadata": {} + "source": [ + "print 'Best preprocessing pipeline:'\n", + "for pp in estimator._best_preprocs:\n", + " print pp\n", + "print\n", + "print 'Best classifier:\\n', estimator._best_classif\n", + "test_predictions = estimator.predict(X_test)\n", + "acc_in_percent = 100 * np.mean(test_predictions == y_test)\n", + "print\n", + "print 'Prediction accuracy in generalization is %.1f%%' % acc_in_percent" + ] + } + ], + "metadata": { + "anaconda-cloud": {}, + "kernelspec": { + "display_name": "Python [Root]", + "language": "python", + "name": "Python [Root]" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.5.2" } - ] -} \ No newline at end of file + }, + "nbformat": 4, + "nbformat_minor": 0 +} diff --git a/notebooks/skdata_quick_intro.ipynb b/notebooks/skdata_quick_intro.ipynb index 4a905771..2e770fb8 100644 --- a/notebooks/skdata_quick_intro.ipynb +++ b/notebooks/skdata_quick_intro.ipynb @@ -1,107 +1,122 @@ { - "metadata": { - "name": "skdata_quick_intro" - }, - "nbformat": 3, - "nbformat_minor": 0, - "worksheets": [ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Hyperopt-Sklearn Intro\n", + "\n", + "## Easiest possible thing\n", + "\n", + "As an ML researcher, I want a quick way to do model selection\n", + "implicitly, in order to get a baseline accuracy score for a new data\n", + "set." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "# Skdata-based code\n", + "import skdata.iris.view\n", + "from skdata.base import SklearnClassifier\n", + "from hpsklearn.estimator import HyperoptEstimatorFactory\n", + "\n", + "view = skdata.iris.view.KfoldClassification(5)\n", + "algo = SklearnClassifier(\n", + " HyperoptEstimatorFactory(\n", + " max_iter=25, # -- consider also a time-based budget\n", + " ))\n", + "mean_test_error = view.protocol(algo)\n", + "print 'mean test error:', mean_test_error" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As an ML researcher, I want to evaluate a certain parly-defined model class, in order to do model-family comparisons. For example, PCA followed by SVM." + ] + }, { - "cells": [ - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "# Hyperopt-Sklearn Intro\n", - "\n", - "## Easiest possible thing\n", - "\n", - "As an ML researcher, I want a quick way to do model selection\n", - "implicitly, in order to get a baseline accuracy score for a new data\n", - "set." - ] - }, - { - "cell_type": "code", - "collapsed": false, - "input": [ - "# Skdata-based code\n", - "import skdata.iris.view\n", - "from skdata.base import SklearnClassifier\n", - "from hpsklearn.estimator import HyperoptEstimatorFactory\n", - "\n", - "view = skdata.iris.view.KfoldClassification(5)\n", - "algo = SklearnClassifier(\n", - " HyperoptEstimatorFactory(\n", - " max_iter=25, # -- consider also a time-based budget\n", - " ))\n", - "mean_test_error = view.protocol(algo)\n", - "print 'mean test error:', mean_test_error" - ], - "language": "python", - "metadata": {}, - "outputs": [] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As an ML researcher, I want to evaluate a certain parly-defined model class, in order to do model-family comparisons. For example, PCA followed by SVM." - ] - }, - { - "cell_type": "code", - "collapsed": false, - "input": [ - "from hpsklearn.components import svc, pca\n", - "\n", - "algo_pca_svm = SklearnClassifier(\n", - " HyperoptEstimatorFactory(\n", - " max_iter=25, # -- consider also a time-based budget\n", - " preprocessing=[pca('pca')],\n", - " classifier=svc('svc')))\n", - "mean_test_error = view.protocol(algo_pca_svm)\n", - "print 'mean test error:', mean_test_error" - ], - "language": "python", - "metadata": {}, - "outputs": [] - }, - { - "cell_type": "markdown", - "metadata": {}, - "source": [ - "As a domain expert, I have a particular pre-processing that I believe reveals important patterns in my data. I would like to know how good a classifier can be built on top of my preprocessing algorithm." - ] - }, - { - "cell_type": "code", - "collapsed": false, - "input": [ - "def my_feature_extractor(name, *kwargs):\n", - " # Should return a pyll graph that evaluates to a Sklearn-compatible\n", - " # feature-extraction component (i.e. with a transform() method)\n", - " raise NotImplementedError()\n", - " \n", - "algo_pca_svm = SklearnClassifier(\n", - " HyperoptEstimatorFactory(\n", - " max_iter=25,\n", - " # -- consider an any_preprocessing() constructor that accepts\n", - " # lambdas which provide initial and final steps to all the\n", - " # default pre-processing pipelines.\n", - " preprocessing=hp.choice('pp',[\n", - " [my_feature_extractor('foo-pre-pca'), pca('post-foo-pca')],\n", - " [my_feature_extractor('foo-alone')],\n", - " ]),\n", - " classifier=any_classifier('classif')))\n", - "mean_test_error = view.protocol(algo_pca_svm)\n", - "print 'mean test error:', mean_test_error" - ], - "language": "python", - "metadata": {}, - "outputs": [] - } - ], - "metadata": {} + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "from hpsklearn.components import svc, pca\n", + "\n", + "algo_pca_svm = SklearnClassifier(\n", + " HyperoptEstimatorFactory(\n", + " max_iter=25, # -- consider also a time-based budget\n", + " preprocessing=[pca('pca')],\n", + " classifier=svc('svc')))\n", + "mean_test_error = view.protocol(algo_pca_svm)\n", + "print 'mean test error:', mean_test_error" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "As a domain expert, I have a particular pre-processing that I believe reveals important patterns in my data. I would like to know how good a classifier can be built on top of my preprocessing algorithm." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "def my_feature_extractor(name, *kwargs):\n", + " # Should return a pyll graph that evaluates to a Sklearn-compatible\n", + " # feature-extraction component (i.e. with a transform() method)\n", + " raise NotImplementedError()\n", + " \n", + "algo_pca_svm = SklearnClassifier(\n", + " HyperoptEstimatorFactory(\n", + " max_iter=25,\n", + " # -- consider an any_preprocessing() constructor that accepts\n", + " # lambdas which provide initial and final steps to all the\n", + " # default pre-processing pipelines.\n", + " preprocessing=hp.choice('pp',[\n", + " [my_feature_extractor('foo-pre-pca'), pca('post-foo-pca')],\n", + " [my_feature_extractor('foo-alone')],\n", + " ]),\n", + " classifier=any_classifier('classif')))\n", + "mean_test_error = view.protocol(algo_pca_svm)\n", + "print 'mean test error:', mean_test_error" + ] } - ] -} \ No newline at end of file + ], + "metadata": { + "anaconda-cloud": {}, + "kernelspec": { + "display_name": "Python [Root]", + "language": "python", + "name": "Python [Root]" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.5.2" + } + }, + "nbformat": 4, + "nbformat_minor": 0 +}