{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Meta Models\n", "\n", "Certain models in scikit-lego are \"meta\". Meta models are\n", "models that depend on other estimators that go in and these\n", "models will add features to the input model. One way of thinking\n", "of a meta model is to consider it to be a way to \"decorate\" a\n", "model.\n", "\n", "This part of the documentation will highlight a few of them.\n", "\n", "## Thresholder\n", "\n", "The thresholder can help tweak recall and precision of a model\n", "by moving the threshold value of `predict_proba`. Commonly this\n", "threshold is set at 0.5 for two classes. This meta-model can\n", "decorate an estimator with two classes such that the threshold\n", "moves.\n", "\n", "We demonstrate the working below. First we'll generate a skewed dataset.\n" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pylab as plt\n", "\n", "from sklearn.pipeline import Pipeline\n", "from sklearn.datasets import make_blobs\n", "from sklearn.model_selection import GridSearchCV\n", "from sklearn.linear_model import LogisticRegression\n", "from sklearn.metrics import precision_score, recall_score, accuracy_score, make_scorer\n", "\n", "from sklego.meta import Thresholder" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "X, y = make_blobs(1000, centers=[(0, 0), (1.5, 1.5)], cluster_std=[1, 0.5])\n", "plt.scatter(X[:, 0], X[:, 1], c=y, s=5);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Next we'll make a cross validation pipeline to try out this thresholder." ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 12.4 s, sys: 0 ns, total: 12.4 s\n", "Wall time: 12.3 s\n" ] } ], "source": [ "%%time \n", "\n", "pipe = Pipeline([\n", " (\"model\", Thresholder(LogisticRegression(solver='lbfgs'), threshold=0.1))\n", "])\n", "\n", "mod = GridSearchCV(estimator=pipe,\n", " param_grid = {\"model__threshold\": np.linspace(0.1, 0.9, 500)},\n", " scoring={\"precision\": make_scorer(precision_score),\n", " \"recall\": make_scorer(recall_score),\n", " \"accuracy\": make_scorer(accuracy_score)},\n", " refit=\"precision\",\n", " cv=5)\n", "\n", "_ = mod.fit(X, y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "With this cross validation trained, we'll make a chart to show the\n", "effect of changing the threshold value." ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "(pd.DataFrame(mod.cv_results_)\n", " .set_index(\"param_model__threshold\")\n", " [['mean_test_precision', 'mean_test_recall', 'mean_test_accuracy']]\n", " .plot(figsize=(16, 4)));" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Increasing the threshold will increase the precision but as expected this is at the\n", "cost of recall (and accuracy).\n", "\n", "### Saving Compute \n", "\n", "Technically, you may not need to refit the underlying model that the `Thresholder` model wraps around. In those situations you can set the `refit` paramater to `False`. If you've got a predefined single model and you're only interested in tuning the cutoff this might make everything run a whole lot faster.\n" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 1.21 s, sys: 0 ns, total: 1.21 s\n", "Wall time: 1.21 s\n" ] } ], "source": [ "%%time \n", "\n", "# Train an original model \n", "orig_model = LogisticRegression(solver='lbfgs')\n", "orig_model.fit(X, y)\n", "\n", "# Ensure that refit=False\n", "pipe = Pipeline([\n", " (\"model\", Thresholder(orig_model, threshold=0.1, refit=False))\n", "])\n", "\n", "# This should now be a fair bit quicker.\n", "mod = GridSearchCV(estimator=pipe,\n", " param_grid = {\"model__threshold\": np.linspace(0.1, 0.9, 50)},\n", " scoring={\"precision\": make_scorer(precision_score),\n", " \"recall\": make_scorer(recall_score),\n", " \"accuracy\": make_scorer(accuracy_score)},\n", " refit=\"precision\",\n", " cv=5)\n", "\n", "_ = mod.fit(X, y);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Grouped Prediction\n", "\n", "\"img1\"\n", "\n", "To help explain what it can do we'll consider three methods to predict\n", "the chicken weight. The chicken data has 578 rows and 4 columns\n", "from an experiment on the effect of diet on early growth of chicks.\n", "The body weights of the chicks were measured at birth and every second\n", "day thereafter until day 20. They were also measured on day 21.\n", "There were four groups on chicks on different protein diets.\n", "\n", "### Setup\n", "\n", "Let's first load a bunch of things to do this." ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "import matplotlib.pylab as plt\n", "from sklearn.linear_model import LinearRegression\n", "from sklearn.pipeline import Pipeline, FeatureUnion\n", "from sklearn.preprocessing import OneHotEncoder, StandardScaler\n", "from sklearn.metrics import mean_absolute_error, mean_squared_error\n", "\n", "from sklego.datasets import load_chicken\n", "from sklego.preprocessing import ColumnSelector\n", "\n", "df = load_chicken(as_frame=True)\n", "\n", "def plot_model(model):\n", " df = load_chicken(as_frame=True)\n", " model.fit(df[['diet', 'time']], df['weight'])\n", " metric_df = df[['diet', 'time', 'weight']].assign(pred=lambda d: model.predict(d[['diet', 'time']]))\n", " metric = mean_absolute_error(metric_df['weight'], metric_df['pred'])\n", " plt.figure(figsize=(12, 4))\n", " plt.scatter(df['time'], df['weight'])\n", " for i in [1, 2, 3, 4]:\n", " pltr = metric_df[['time', 'diet', 'pred']].drop_duplicates().loc[lambda d: d['diet'] == i]\n", " plt.plot(pltr['time'], pltr['pred'], color='.rbgy'[i])\n", " plt.title(f\"linear model per group, MAE: {np.round(metric, 2)}\");" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This code will be used to explain the steps below.\n", "\n", "### Model 1: Linear Regression with Dummies\n", "\n", "First we start with a baseline." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "feature_pipeline = Pipeline([\n", " (\"datagrab\", FeatureUnion([\n", " (\"discrete\", Pipeline([\n", " (\"grab\", ColumnSelector(\"diet\")),\n", " (\"encode\", OneHotEncoder(categories=\"auto\", sparse=False))\n", " ])),\n", " (\"continuous\", Pipeline([\n", " (\"grab\", ColumnSelector(\"time\")),\n", " (\"standardize\", StandardScaler())\n", " ]))\n", " ]))\n", "])\n", "\n", "pipe = Pipeline([\n", " (\"transform\", feature_pipeline),\n", " (\"model\", LinearRegression())\n", "])\n", "\n", "plot_model(pipe)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Because the model is linear the dummy variable causes the intercept\n", "to change but leaves the gradient untouched. This might not be what\n", "we want from a model. So let's see how the grouped model can address\n", "this.\n", "\n", "### Model 2: Linear Regression in GroupedEstimation\n", "\n", "The goal of the grouped estimator is to allow us to split up our data.\n", "The image below demonstrates what will happen.\n", "\n", "\"img2\"\n", "\n", "We train 5 models in total because the model will also train a\n", "fallback automatically (you can turn this off via `use_fallback=False`).\n", "The idea behind the fallback is that we can predict something if\n", "the group does not appear in the prediction.\n", "\n", "Each model will accept features that are in `X` that are not\n", "part of the grouping variables. In this case each group will\n", "model based on the `time` since `weight` is what we're trying\n", "to predict.\n", "\n", "Applying this model to the dataframe is easy." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from sklego.meta import GroupedPredictor\n", "mod = GroupedPredictor(LinearRegression(), groups=[\"diet\"])\n", "plot_model(mod)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "And the model looks a bit better.\n", "\n", "### Model 3: Dummy Regression in GroupedEstimation\n", "\n", "We could go a step further and train a [DummyRegressor](https://scikit-learn.org/stable/modules/generated/sklearn.dummy.DummyRegressor.html) per diet\n", "per timestep. The code below works similar as the previous example\n", "but one difference is that the grouped model does not receive a\n", "dataframe but a numpy array.\n", "\n", "\"img3\"\n", "\n", "Note that we're also grouping over more than one column here.\n", "The code that does this is listed below." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from sklearn.dummy import DummyRegressor\n", "\n", "feature_pipeline = Pipeline([\n", " (\"datagrab\", FeatureUnion([\n", " (\"discrete\", Pipeline([\n", " (\"grab\", ColumnSelector(\"diet\")),\n", " ])),\n", " (\"continuous\", Pipeline([\n", " (\"grab\", ColumnSelector(\"time\")),\n", " ]))\n", " ]))\n", "])\n", "\n", "pipe = Pipeline([\n", " (\"transform\", feature_pipeline),\n", " (\"model\", GroupedPredictor(DummyRegressor(strategy=\"mean\"), groups=[0, 1]))\n", "])\n", "\n", "plot_model(pipe)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Note that these predictions seems to yield the lowest error but take it\n", "with a grain of salt since these errors are only based on the train set.\n", "\n", "## Grouped Transformation\n", "\n", "We can apply grouped prediction on estimators that have a `.predict()` implemented but we're also able to do something similar for transfromers, like `StandardScaler`. \n", "\n" ] }, { "cell_type": "code", "execution_count": 49, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
speciesflipper_length_mmbody_mass_gsex
0Adelie181.03750.0male
1Adelie186.03800.0female
2Adelie195.03250.0female
4Adelie193.03450.0female
5Adelie190.03650.0male
\n", "
" ], "text/plain": [ " species flipper_length_mm body_mass_g sex\n", "0 Adelie 181.0 3750.0 male\n", "1 Adelie 186.0 3800.0 female\n", "2 Adelie 195.0 3250.0 female\n", "4 Adelie 193.0 3450.0 female\n", "5 Adelie 190.0 3650.0 male" ] }, "execution_count": 49, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from sklego.datasets import load_penguins\n", "\n", "df_peng = (load_penguins(as_frame=True)\n", " .dropna()\n", " .drop(columns=['island', 'bill_depth_mm', 'bill_length_mm']))\n", "df_peng.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's say that we're interested in scaling the numeric data in this dataframe. If we apply a normal `StandardScaler` then we'll likely get clusters appear for all the `species` and for the `sex`. It may be the case (for fairness reasons) that we don't mind the clusters based on `species` but that we *do* mind the clusters based on `sex`. In these scenarios the `GroupedTransformer` can help out. We can specify a grouping column that the data needs to be split on before the transformation is applied." ] }, { "cell_type": "code", "execution_count": 50, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from sklearn.preprocessing import StandardScaler\n", "from sklego.meta import GroupedTransformer\n", "\n", "plt.figure(figsize=(12, 6))\n", "# we'll drop species because we're not grouping on it\n", "df_peng = df_peng.drop(columns=[\"species\"])\n", "\n", "plt.subplot(121)\n", "X = df_peng.drop(columns=['sex']).values\n", "X_tfm = StandardScaler().fit_transform(X)\n", "plt.scatter(X_tfm[:, 0], X_tfm[:, 1], c=df_peng['sex'] == 'male')\n", "plt.xlabel(\"norm flipper len\")\n", "plt.ylabel(\"norm body mass\")\n", "plt.title(\"scaled data, not normalised by gender\")\n", "\n", "plt.subplot(122)\n", "X_tfm = GroupedTransformer(StandardScaler(), groups=['sex']).fit_transform(df_peng)\n", "plt.scatter(X_tfm[:, 0], X_tfm[:, 1], c=df_peng['sex'] == 'male')\n", "plt.xlabel(\"norm flipper len\")\n", "plt.ylabel(\"norm body mass\")\n", "plt.title(\"scaled data *and* normalised by gender\");" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "You can see that there are certainly still clusters. These are caused by the fact that there's different `species` of penguin in our dataset. However you can see that when we apply our `GroupedTransformer` that we're suddenly able to normalise towards `sex` as well. \n", "\n", "#### Other Scenarios \n", "\n", "This transformer also has use-cases beyond fairness. You could use this transformer to causally compensate for subgroups in your data. For example, for predicting house prices, using the surface of a house relatively to houses in the same neighborhood could be a more relevant feature than the surface relative to all houses. " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Decayed Estimation\n", "\n", "Often you are interested in predicting the future. You use the data from\n", "the past in an attempt to achieve this and it could be said that perhaps\n", "data from the far history is less relevant than data from the recent past.\n", "\n", "This is the idea behind the `DecayEstimator` meta-model. It looks at the\n", "order of data going in and it will assign a higher importance to recent rows\n", "that occurred recently and a lower importance to older rows. Recency is based\n", "on the order so it is imporant that the dataset that you pass in is correctly\n", "ordered beforehand.\n", "\n", "We'll demonstrate how it works by applying it on a simulated timeseries problem.\n" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsUAAADFCAYAAAC8VI8xAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsnWeUFNXWht/q7sk5AgMDQ0YY8pBzFAEDBlQUE4p6zeEqehUzKmY/IxjuxYQBFRUFBMl5yDkPMMQJTE4d6vvRXdUVO0yn6u79rMVa3VWnqg5TXXX22WfvdzMsy4IgCIIgCIIgwhldoDtAEARBEARBEIGGjGKCIAiCIAgi7CGjmCAIgiAIggh7yCgmCIIgCIIgwh4yigmCIAiCIIiwh4xigiAIgiAIIuwho5ggCIIgCIIIe8goJgiCIAiCIMIeMooJgiAIgiCIsMcQiIump6ezOTk5gbg0QRAEQRAEEUZs3bq1mGXZDGftvGIUMwzzCIA7AbAAdgO4nWXZOrX2OTk5yM/P98alCYIgCIIgCEIVhmFOuNLO4/AJhmGaA3gQQB7LsrkA9ABu8PS8BEEQBEEQBOEvvBVTbAAQwzCMAUAsgDNeOi9BEARBEARB+ByPjWKWZU8DeBPASQBnAZSzLLtU2o5hmOkMw+QzDJNfVFTk6WUJgiAIgiAIwmt4I3wiBcCVAFoDyAIQxzDMzdJ2LMvOYVk2j2XZvIwMp7HOBEEQBEEQBOE3vBE+MRrAcZZli1iWNQL4GcBAL5yXIAiCIAiCcJGvNhTgREl1oLsRtHjDKD4JoD/DMLEMwzAARgHY74XzEgRBEARBEC7QYLLg2YV7cc3HGwLdlaDFGzHFmwD8BGAbrHJsOgBzPD0vQRAEQRAE4RoWlgUAFFfVB7gnwYtX1CdYln2OZdlOLMvmsiw7lWVZuiMEQRAEQRB+gjOKAaDwYg3/ef3RYizadTYQXQo6qMwzQRAEQRBEkLP+SAn/+fD5Kv7zlLmbcN+32wLRpaCDjGKCIAiCIIggpsFkwZ3z7JWC9TomgL0JXsgoJgiCIAiCCGJMFovou0EvN4rPV9T5qztBCxnFBEEQBEEQQYzJwoo3sPI2/WYtx+1fbkZtg9k/nQpCyCgmCIIgCIIIYkxmsRXcYLYotltxsAhbCkr90aWghIxigiAIgghhiqvq0e7pP5FPxlDIsPdMOTYfL8VvO8/gfEUdTBIjuMFk/b54zznZsUkxEX7pYzBiCHQHCIIgCILwHVuOl8JkYfHZmuPIy0kNdHcILzDh/bX85zYZcfhqWj/Rfs5TfM/XW2XHykItCB7yFBMEQRBECEMmUGhz+mKtoqd4wdZCxfZGldAKgoxigiAIgghpuJoODKl0hSQsgN92nBFtK6sx4rEfdyq2bzBZYLaw2HGqzA+9Cy7IKCYIgiCIEIYlX3HI89bfh0TfX/xjn2pbo9mCD/45gqs+XIftJy/6umtBBcUUEwRBEEQYQJ7i0IRLqnOVaf/LR5+cFABAUWW9L7oUtJCnmCAIgiBCGD58Agy2nriIIxcqA9shwiOKqzw3ZLcUWD3EEXodNh4rwcXqBo/PGQqQUUwQBEEQ4QADXPPxeox+e3Wge0I0knqTGXkvL/PqOW+YsxG3frnZq+cMVsgoJgiCIIgQhosoJm9g8FNnbLxyRHZqjML5rNXtDpyl1QOAjGKCIAiCCAvOV9QFuguEh3gip3b7wNaybWW1RgCAQU8B5wAZxQRBEAQRFnjiZSS0Qb2bSXVCoiP0sm1lNVajOEJP5iBARjFBEARBhDSsLdPOE4OK0Ab1tnCHxhAdITf5ymqsITUR5CkGQEYxQRAEQYQF9abGG1SENvBkYhNlkHuKL/JGMZmDgJeMYoZhkhmG+YlhmAMMw+xnGGaAN85LEARBEIRnmC3KnuKjRVU4U1YbiC4RjcQTo1ivY7BuxkjRtos1FFMsxFvFO94DsJhl2WsZhokEEOul8xIEQRAE4QEmm1EsLfIw6q1V/OcZl3XCu8sOYd8L46DTkYGkRfafrcCJkupGH9+rZTIMEo9wqU2RRMmLHI54bBQzDJMEYCiA2wCAZdkGAKT7QhAEQRAagPMUC3lHUhb4tb8OAADGvbcaXZsn463J3f3SN8I1GkwWXPbeGtX9kXodumcn8UU5pHwwpScyE6NRXW8SbS+xFQKJUUjCC0e8ET7RGkARgC8ZhtnOMMxnDMPESRsxDDOdYZh8hmHyi4qKvHBZgiAIgiDUOF9RhyV7z/GeYiHvLT+seMyh81VYsK3Q110j3MRpPDgDMJI63nmtUvjPTROjAQBRBrHZV1BSAwC4UFmH/WcrvNDT4MYbRrEBQC8AH7Ms2xNANYAZ0kYsy85hWTaPZdm8jIwML1yWIAiCIAgl5qw+in6zluPur7airoES7IKZnafK0PX5pQ7bxEbqIY16+enegTj8ymX45s5+yMtJBQAY9DqM6pQpO/58Rb1DT3S44A2juBBAIcuym2zff4LVSCYIgiAIIgDM+vMA/7mEKtkFLbUNZlz54Tqn7X64ewB0jDwWPEKvw6B26aJt1/Zu4bX+hRoeG8Usy54DcIphmI62TaMA7PP0vARBEARBeM4nq44GugtEIzFanKtNfHtnP3RokgAFm1iRKAW9YsKKt9QnHgDwjU154hiA2710XoIgCIIgXKC2wYxzFXVonS5L6yGCFLNZHg8uhYsZ5zzFb17XHf1ap6q2j9RTUp0aXjGKWZbdASDPG+ciCIIgCMJ9pn+VjzWHi7Hq38M9Pte58jo0TYr2vFOER7jiKW6eEgPAnmiXFh+J7FR1ZdxIg7qnuMFkcbg/1Anf/zlBEARBhBBrDhcDAIa9sVK0/clxndw+1zO/7vZGlwgPeeaXPar7+rZOxaanR6FtRjwA8Il2XFlvNaQKFEI+WnkERy5Uut/REIGMYoIgCIIIUX7+10DcO7wtmql4fW8flAMAuG9EW9F2o23Z/qsNBciZsUhR65jwPUv3nVfc/tM9AzD3ljw0SbTfVy58wplz2ZEn+N1lh3HVh+vd72iIQEYxQRBEGHLkQhWGzl6BYpt4v5Bp/92CDs/8FYBeEd4mLtIaJXm2vA4AML5rU9H+5y7vgh/uHoBHRncQbbfYvI2v/LkfgFUnd+epMpjMjS8zTHiPvJxUJMVEiLY9PLo9mifHoI+DeGLAbhTHRxkwpV9LDGybJtpfJSnwEU6QUUwQBBGGfL72GE6W1mDJ3nP8tkW7zuJkSQ2WH7ggKwlMaJvyGqPi9thIcVJVO9tSu5C+rVNh0Osw+5pu/DYuFINjz+kKXPnhOrwtqYRHaIduLZKxbsZImbEshQufsLAsZk3qKpNsC2e8pT5BEARBBBFRBquxVG+0Gr8WC4v7vt2G9PjIQHaLaCSny2oVt8dFWYf5lqmxOFlaI0rASo+PErWNMIg1vYQewzO28+85Q1XPgp1IgVEMOI4xDjfIKCYIgghDOK3SOlv52Mo6qwFUXEWFHoIRo0pYQ1yUdfLz3fT+2HriIiIEZc/WPjlC1FZa/KGi1u595mS/nCVxEdqHmxBzscfhrDYhhf4SBEEQYUi0bWC8UFGPl//YhyKF2GKhAbTzVBkW7Trrt/4R7mFSyK4a16UpIvXWYb55cgyu6J7F77u0SxNER4hDKxiJUfzVxhPgfgJmF6TBiOAgygVP8Zfrjvu1T1qBPMUEQRBhCOcp/u/6AgCAkv+v3mThDSeu1OyEbhP80T3CTb5YVyD63i4zHp9M7e3WOaQF0T5eeZQ3mPafDV+ZLq0xsVszj47nJkqX2yZJSp7iF37fh9sHtfboOsEIGcUEQRBhCLeEynHwnNzoqTOaZd5EQpsIvfidmyXi538NVGznqBSwNHwCsE6MAMHkiaInAs5luZ4ZxTodg63PjEaiLSEvhp5xHgqfIAiCCEOk3qG1R4plbaZ+vhmbjpWIlA3GvbsaP+Sf8nn/CNf5dNVR0ffUuEinkxkl49aRwcwfp7imQPgTnQv3yRlp8VGIsHmMU2IpuZaDPMUEQRDhiAsuv92ny3H9nI3o0MQu43XgXCWe+GkXJudl+7J3hItU1hnx6l8HRNsaqybgDWOL8D3S2G9PSSPFGR7yFBMEQYQh7lQoO3S+SvQ9S6U6GuFf6oxm7FOQSGuXKdcidg27sdU3R7kABIVPBB5vT15S46IUt+fMWITCizXevZjGIU8xQRBEiGOxsNhRWIZeLVNQXFWPJ37a5VDg/+qezfHz9tOq+5NouVUTdHp2sWzbU5d1wh2D1ROkuLhhg15uWQkdkBkJyoYSGcX+YebCPahtMCvuU4r99oRkB++CHafK0CIlVnV/qEGeYoIgiBDns7XHcPVH67HhaAlmLz6Afw5cwC8KRm+crfpZ/zZpsn1EcDC+azM+VlSJEZ0ycdvAHLxwRa5sn9DU4vSNpdQ0hG8JYH8yb8MJ/Li1UHGfzsuWm07H4MBL4xT36b1sgGsdMooJgiBCnL22Jfaz5bV8EQYl5t6ah5evyhVVPVOCCjhoFyUPsJAIvQ7PX9FF0RPcxxYykZ0ao5p8tbOwHPUmZQ8m4R3qjPa/r7DqYEK0dXHf2zHFABAdocesSV1l23VhFmhORjFBEDJ2FZbBaLagut6E2gYz6oxmGgiDFJZlcba8DgDw4h/78Nfuc6ptB7ZNx839WyHS4HggFNrEh89XYvWhIq/0lfAcgwduxJS4SBS8NgFrnhjpUL1CqEZCeJ9qQXntCMEk59reLQAAHZok+OS6U/q1xNAOGaJt4eYppphigiBEHC2qwhUfrMMdg1rji3XHkRBtQHW9CSmxkdj67JhAd49wk1+2n8bm46UAgDIXjRlHy++AvRIWAIx5ZzUAoOA1KuqhBSKceIpdJSZS3SiuqDMiM5GSLX1FjSCWWLiyc02vFpg5sbNPPMUc0p+PnjzFjYNhGD3DMNsZhvnDW+ckCML/lNdaDaetJ6yGVGWdCRYWKKlugMUNxQJCG+w8Vea0TZRBh5evsseYOjOKD1+owtnyWo/7RngHoXfP4OTeuUq0A1k3VydXROMQGsU1Aq+x0WzxqUEMKBjB4WUTezV84iEA+714PoIgAgCncbqzsFy2b+L/rfV3dwgPcUUpYvUTI3Bz/1b8d2dGMQCsOVSMFQcueNQ3onGYzBa8vtiuTZwhiDv1lqc4gozigCFMZqwWGMhGs++dElJlC5PtmvUmc1gkWXrFKGYYpgWACQA+88b5CIIIHI7kfvadrcBD87ejqj70X46hwNrDxXh/+WGn7aRGsCuGVVyUAX8ISgsDwK/bT+OlP/a510nCbRbvPYePV9qr2AmT5iK8JE1gcLBs/tKiffh5m7IyAuE5alJsRrPF59eWvv+5a1723hp0nrnE59cPNN7yFL8L4AkAqneMYZjpDMPkMwyTX1RESRkEoSUOn6/EG0sOgGVZp0UdFu44g54vLgUAnK+ow8ZjJf7oItEI5m0o4D/f0Ee9Ap3UCHbFUxwbqZd5jh7+fgc+X3vcrT4S7iM1jjISojBnam8MaZ/uNbUALmFvUs/msn0nSmrw6A87vXIdQk61xCju2TIZTROj0bVFks+v3SpNrDzD/daOFVX7/NpawGOjmGGYiQAusCy71VE7lmXnsCybx7JsXkZGhqOmBEH4ico6IzYfL8XNn2/ChyuOorS6wSVvBLeMN/ad1bhhzkZRtjShDTYdK8HJUns1qiHt7e/dDU+NFLWVe4qdDw0sWNngTfgHRhLo2Tw5BmO7NMVX0/p57RqctJsjCT/CN0gnm1lJMdj49CgkRqsX2fAWj43tiE9u7sV/N0lCNuZtKAjp3BJveIoHAbiCYZgCAPMBjGQY5msvnJcgCB9z79fbMPnTDThfUQ/Aqn/pavnf//yym0/KqzOaUVbTgE9XHcXJkvAqC6o1ymoacLqsFtfP2YgD5yr57Slx9gG1WVKM6BipERypYBTf1K+l6HuDiRUlARH+4WRJDVjYn9HuLZIwpnMTr1+HS7gymS0IM1WugFNvEjsmiirr/XbtSIMO43Kb2e+/xYJjRfYy7zMX7sWqEJZg9FiSjWXZpwA8BQAMwwwH8DjLsjd7el6CIHzPrkKxMoHJYnE5meObTSf5z9d8vB6t0+Ow4mARXv3rAMlzBZDBr69QjPl25GWSZpxHKOgUS4s5GM0WGAUTqFD2HmmF7ScvYtJH69HXVmQDAHq3SvWJbJZBZ/cU6xkGJoWCLRuOliA+yuCXZf1wgWVZNEiMYuEkyF+seGw4hr6xAkYzi5FvrRLtc9VxEoyQTjHRaKrqTch9bgmuz8vG69d2C3R3iEYgHecuVhtx49yNbp+noKQmAK9tQgm1JMikGNeXXpXCJ7pkJYq+G80WmC32wdto8X0SULjDef53CiazFh9VF4yJtJoH0RF6a5yygiHEvStoEuw9bpizEZtsuuIc1fX+D1OKt1XPW3u4WLbPUWGXYMerFe1Yll3JsuxEb56T0C65z1kzUb/PPxXgnhCNRTqgPvvrHtW2kQ4kmgBr8g0HlQHWHlyJWFeQKg9MzmuBS7s0FW0zmi2ieMPrPtngWQcJp3Dlf6MEz6KvvHZD2qXjkdEd8OIVXfDEpR19cg1CjtAg/u6u/gDkMcb+gIspX7xXXgHTVxMxLUBlngm3WX2oCB+uOBLobhAe8MXa41iwtVDm/NlcUKp8AIBBbdMUM9GVqKilWFOtER/lulEsLRDQq2WKTNXgyQW7RTHLuxR0rQnvUme0euMr6uzPl9lHBopOx+Ch0e2REheJO4e0cdp+1aEiXPnhOpj8IBsWLjRJtErtBSKh1ZG0nzS8I5Sg8AnCbW75YnOgu0B4yIs2LdnoCNfnxSxcLwxQbzID8H2mNOE6Br0Oj43pgCEdXFP/+fL2PvhyXQFWHypyOzSGZVmfV94KR6zPlRitrMrMWLALZ8vrcLa8Dtmpsc4PIJzCldJ+cGQ7v1/b4OBd3xDCEx/yFBNEGOPOyquFBfQ278GQ9ukO2z67cI9mBmvCzgOj2qNHdjIA4M3rujtsO6JjJrKSrIMydytv7NsSvVomO70OyXj5Bs5TLEQrSU/ptqp6Z8qo/Le3iInQo+C1CZg6IMfv15aGUN0/wm6Yh7KnmIxiwut0evYvqmqlAWoaTHwMohruvNysRq51AHaWaLFk73kUXqTBUctc27uFTKhfylCbV5lLsnv16q6YP32A03NrxVALNcwKyYyB+FO3y4yXbUuNs6qTnCmn574xWCwsZizYJdrmC1URV2EYBs1sk+J/DW+LJrbPABnFRJhRWWfE87/tRZ3RjHqTGRV1rtW5n/r5Jny18QTqjBaqaqUBOs9cgsGvr2j08dJQCZYFuDG5eXKMwhFiQjkZQ6soLa87YvFDQ7Fz5ljV/eO7NsOu58eie7bdOyz8XdzcvyUeGtVedhx5in2DUgl2f0vhLX1kqGKxHs6zGAilhFDgYk0D5m/RVtL6tMGtAQBjuzRFpOC5f2LBLhy5UInNx9VzUIIViikmZHy44ij+u74ALVJi8Meus9hxqswlyZ01h4uxRkG+hQgcxVVy0XdXq8/pdYxIs5gFyxu6bTPjkRoXidLqBtXja514qQnvc9l7a2TbWqbG4tExHRTbx0TqERPp2Osv1TcWxgrf0KcljgqE/TnMLmpdE+6h5IFvlhyt0NJ3pMVFYs7UPFz+wVp+24XKOr7Ah7PVKUIZacEOLTBtcGsMaZ+Bjk0TcLxY/JyPfns1gNCT4yNPMSGDW6IzW1jsOFXmpLWVxXvksi17z5Rj+rx8lz3NhH+4/9ttLrXjvFKXdrFWy7JYgDrbizsuUo/0+EjVYwGghkoA+51jRdWybX8+NARXuaga4i4pcZGKmsYm0iz2CUoe+IdGKU94fIWOYdC1RRJmTuzMb+v7ynIs238BABnFjcVVZ4U/YRgGHZsmAHCt9HsoEB7/S8ItOE+QO76ee77eKts24f21WLrvPP63rsA7HSO8wraT9olOQpRBVYGCM4o7NrG+FFmwOF9eBwBomhiN63pny47hJIQAoMa2jLrhaAm2OJB6I3xLjA+F9lNiI2QJOQDFFPsKpb+rM/1wb8NJ88VFKf+uaIWocQRCds0dogzq75GtJ0px0cGqYTBBRjEhgxvilEJC8xth3Ej1TYnAIgxLrGowYfqQNorL65x6RJxN39bCAvU2KZ6mSdG4a2gbLLh3oOiYHoLYU05w/sa5G6mwgx9QWq0BfJusExOhHH5BMcW+QQt/V+7nFKeie62kkEE4R4ueYiFqzhOWZXHNxxsw6aN1fu6RbyCjmJCjMIb++8edaPf0n5i9+KD7pyObWFMI4z1Z1lrOs1/rVFm7KJuHkR/8WODd63vgkdEd0Do9DoD83nZsai8FXNNg1vyLPpSYuVC9GqGvYBgG/duk4Z5hbUXbOY/mD/mn8DIp0XgNofpElEGHd6/v4bdrc8YwN8lSS7ZdcfACvt54AjkzFoW0SoG30fq7Um3FiZsEFQgqmgYzZBQTMhhw4RN24+nHrYUwWVhcqKxz+3xW1YLAezjCFYuFxfvLD6PwYg0mvL8GlZKXb3xUBG8Acxx/dTy/LJ7bPAkAcGXPLLROj8NDo9vzITadmyWKjmshGCgv1jSgi60UuJSD5yrx5E+7aJndi1yolCdV+op/HhuG+dOtJWgj9DrMuKyTaP9Lf+zD6kNFeOKnXfiMlGi8woFzFfghv5D/vvLfw30WK66EwaZRzoVV9WyZgpeuypW1O1ZUjZcXWSdCVRo39LSCyWzB9K/kIYhaQk2Gs7w2tHKGyCgmZHCyTkrhE41Zin1jyUE88N120bbCizXoP2s5TpXaZ5cWC0ueBR+w8VgJ3v77EG76bBP2nqmQ7U+Ni0T3Fkl4/Zqu/DaGYZCRYI0PbpESg6OzxmNK35ayY6Mj9Djw0jj++yXNEnlD+dB5uSoBx0Pzt+P7/FM4dL5StQ3hOmqTzrcnOy7Q0VjaZMSjf5s01f1L952nypdeZty7YmURfyc+DWxnvd/C1aGrVYxyTrWGSj67RmWd9icPakbxI9/v8HNPfAtJshEiPltzDF86SIxT0sl0hUW7z+JDwfcf8gtxrqIOC7YV4uHR1njW+77dhr/2nAs5iZdAY7QZTBUqM/qWqbFgGAbX92mJER0z+YnP57f2wapDF/hKVWoIX5Zp8ZFY9OBgXP3xeny3+aTqMfG2kAy1PhGu88mqoxjYVtlAvbpXCz/3hvAXETr/GsUf3dQLp0prRQlXsSpyftwKkBZlxrSIOQg03dViijccK+E/Wyxs0OcQkaeYEPHyov38ZyXv0+EL6t4/d+A8CEJvx18qiUKEd1BLgGmeYg95yEyMRprNCG6aFI3r+8i9w45IjYsEwzA47aSaXWKMVfs2GDwkWsZiYfHaXwdwxQeBT3JZ++QI3Du8rfOGhFeIMPjX+IiNNPDyXByMxEkSKfFekzybawTDCqkrKjYzf/N/XoO3IaOYUMWXEjFGm1GsJOdE+AY1qaR4lSzyxsB5jRsUlk1Pldag8KI1XCYh2uYpJg1rjzBqSA+4RUos2iuU/wWA3YXlyJmxCAXFch1lwjlKRpMWdWO5yS4HeYpdQ/p3enBkO/zvjr4B6o0yauETQr7eqL46GCxo76kiAsY/B86LvrubVJccG+G8kQ0u5kzpxc4GwVJSMOHs7+kLyS69QpjNkNkr+LLTnCFOiTieYdRY5bjxXZspbv9ui3WwvM/FwjGEmAe+k//dtOhQSIwRT7DdLTserlz67mrR98HtMzCsQ0aAeqOMK0ZxKEBGMcFzx3/zRd+3nrjo1vEPj2rvcL9QcobzFJfVGrG7sFzUjhQJvEsgDCdnhjYXm06qJJ6hlMjUKi0WgDVB0t9ER+hxUz95yA337CslehLOWbJX7LC4fVCOLHRBC0hLgpNmsWtIVwIsGnQM6XUMnh7fCX88MFiUlC3lwxVHkDNjEb8qGGyQURzmnCqtUTV+T7ipO+jMvnlo/nbsOV2Oosp6mGyG2vvLD+PyD9aK2mlBoD6UUPLWDG6X7tVrTOzWTJSJ7qr3WWOOzqBDGqZyWW5T/HD3AHx+ax5+umegylH+h2LHG09ZjbxS2HOXdwlAT5QRPuvy8AnyFDcGZ8nNgWL60LbIbZ6Eid2yVNu8t+wwAOCfAxf81S2v4nEwIcMw2QDmAWgCa2XgOSzLvufpeQn/MGS2dTnbG4oPFpbF5d2z8PvOM4r7l+2/gGX7nT8oDWZL2CzV+IN6BW/NJc0SsPZIsdeu8cGUXqLvrqqUmDUUExuMSFcBRl/SBE0So9EkMTpAPVIu1kMqI43jeHE1Rry5MtDdcMib13XDI9/vBGDPFeAQeop7vfQ3miVFY9GDQ/zav2Bj6SND0U4lNl8rcEl3HZsk4KBEVjM2So+GGkvQToS94Sk2AXiMZdnOAPoDuI9hmM5eOC+hAbJTnS/BXtqlCQCrUfx/N/ZUNLA5zVs1hHGvRkrO8CpKyS6+nnQ4sonLa+wGEq0KeIY0fMKgD/ySOqNQErNGkLRrFPTZaLbg+y0nKWRKhRMl2k9MnNSzBV8RM1FmFNvve2l1A/aeqUBNg4nyRhzQoUmC80YBRqdjMH96f3xzVz/ZvrhI628gbI1ilmXPsiy7zfa5EsB+AP4rs0N4hdJq+RIdAOSkxTk87o1ru/FtHI1rSoUfhAiNIzKUvMtbS+WluX1tOrVMjVXdd/XHdvkwM8VPeIRRYhRrUZEAEPdTqILyxdrjeHLBbvy09VQguqV5Sqrs7+UbnbxDAwn3zk6QxBSX1chXCDrPXIL/ri/wR7cIH9K/TZpimEeMTbu6MkiVhbz6BmUYJgdATwCbFPZNZxgmn2GY/KKiIm9elmgEdUazyGPX66W/FdtxSTtC2qTbDWULaxfrFiYH9G6VIjomSkX4m0PozQwGzcZgokQw4eG8+qUKcYre5ENBOEWSJM7waFE1vtp4AkBwiNZrmQaT+O+nRUUCQGwUC72HxVXW0tQXFYwnwv73AeShCVqCW7GIMojf88L+C/lNJcQuHNl/NrSST4/YahmoSYBqHa8ZxQzDxANYAOBhlmVld5ll2Tksy+axLJuXkaEtqZFw5MoP1qH7i0udtstKlodPTOyehTGduZAJoHeMOcwtAAAgAElEQVRLqwGcm5XEt2maZI1pvKpHFlY8PtzpdR4VlIqUer8Iz+DspB0zx+CqHtZFHF8bISlxkfznpg7iWz1ZNj9yoTLsE3ne/vuQ6LsWPMVKoTPC2Oe6Bnq+XaVMEIutVj1OCzTY7q80wfZEaQ1qG8zYfLxUtJ3e8cD8zSfxzt+HcNl79vLdr0zKDWCPGodaUqBSGFUw4JWpJ8MwEbAaxN+wLPuzN85J+BZpcLwa0gpFAHB5t2YoqrR6ACwsi9Gdm2DT06NEyT2cNE/vnFS0To+DM4fg0n12ySEKn/AcYZGElNhIjMttiuTYSIy8JBOT81rggZHtsWjXWb/0xVGBicbe6+Kqeox+ezWuz8vG69d2a2zXgpZfthfieFE1lu0XS3VpIaZYidNl9gqHwepBCgTC/Aol7W+twHmK0wSTYQBYtOss1h4ulq1gGE30jp/x827R905NE3BTv1YB6k3j+efxYagzmtH3leWi7ZuOW8s/my0sahpMstAareIN9QkGwOcA9rMs+7bnXSK0xMC2YukuLomOe8dxOrPSbHdOI1XazhUofMJzhgsy1qsbTHy2cJRBj9nXdvdrX9pmxONYkXLCUL3RggaTBZEG9zyc5TYP2uaCUictQxMu21+KFgwnZz2oaZAn4FCinRyzheV/5wBQpfB30wrc5DYvJxWJ0QZUCJKsymuNSJEUdiJPsZxuLZKcN9IgidERMn1qACi8WIsVBy/g733n8e2mk1g3YySaK6w8aw1vrLUNAjAVwEiGYXbY/o33wnmJAHP81fHonJWIz2/Nk+3T83HEysfeNaQNnp3YGZPzsh22U4JemN6lzmjhkx8CwduTxUb4v4a35T9/se44Hv9R2cBzBLfyoAEbUFMYNWxccgPimTJ7pUwul4DKfdupbTDjVGkNnvl1D37cWshvr6434ZkJl+DTqb0D2DtluHd2dIQewzpmyvZLf5ZKZeAtFhbvLjuECxXuVVINFQwaCH3yNieKq/HtJms1y2C5rx57ilmWXQvfJ7MTAYCrmDTqEmv8cG7zRH4fp0Or5uGJNOgwbXBr/rs7FXoofML7BLJCknTZLC5K/NpZvOccAODQ+UowANq7IEnE/X9c1UMOF7QgZ6hWaa13qxScKa/lE3EAu2xTXQOFVHBM/yofaxRCDlqlxuHOIW0C1CvH2I1iHV+9sEliFM5XWMPsmifHiLzeSo6P/BMX8e6yw9hzuhyf3dpH8Tqz/tyPKIMOj43t6O3/QsCJ0GiSrCfUmYTyi8Exrofe1ITwCav+PRzf3dWf/86Ne64aW5N6NkdspB4L7xuE5Y8Nc9i2lgbIRtNgsuBcuXxGXlot98TNvrYbPrm5l2y7r8mQJGbERunBsizGvrMaY95ZDYuFdapjajeKfdbNoETLqyypcZFIi4vCRyuP8OFUnGyTtzzcdUazqrxksLDmsLWojjCx7sd7BuAOgZNBa3AVSiP0Ot74Fert7pMoLCgZSFzlPkeP/pzVx/B//xzB1hOluPWLzWgwWYIm9GbxnnPImbEIF1V+n1pIkvU2wrG88GKNSHlGq4TeXSA8Qi22s1VanMjjx8UuumoU56THYd+L49A9OxltMxxX6znkYhIgIefJBbvQ/9Xlsu2l1XJppMl52RiX28wf3eL55s5+Mpm/mnozLlTa+9fm6T8x6LV/HJ6Hy90L1gxnT9hdWC7bdvdQqwexRYq6PrS/6NTUbgwlCFYFoiJ0KK6qR73JgsV7rasDXOzphqMlYFkWm46V4L/rjsvOeaGyzqXJ8pS5G1XlJYONKEGBnT45qS6XTg8EPbKTAVgrnXFGsaPfIrei8d3mk5g+Lx9mC4tqW8x0vE16rrS6ATkzFuHX7adlxz/43Q6sOlSEDs/8hSGvO35XaIXP1x4DABwWrJQICcXwCaER/OgPOzH9q60B7I1rhN5dIDxiycNDXWrnLKbYE46ovDQI5/y1R1lR4oGR7f3cE2UGtUuXVdNrMFtk3r0zCt5uIdxkLByjJ+asOSb6fk2vFphxWSes+vdwdNVAss71fbLxxwODcXTWeGyfOYYPuxKGA/yQfwqHz1eiymYUHy+uxserjuL6ORvx/O/7ZOfs+8py3Dh3o9NrbztZ5qX/ReDRsA0s4/0be+K3+wchLsrAG0JZSepSjFxM8VM/78bSfedR3WDifwucKs5D87cDAB+TKkS4kuTsXaEVnPmPIjSqHOMJUqWZ1Ye0X6OCjOIwYd+ZCuTMWIQTJdU4VqRsdLbNiEPr9Dj88cBg/N+NPR2ejxNp98WLm6uCdOh8Jfq8sgwnS2q8f5EQRe3Fm9s88MYSh1IhF6WQj58ESUZSuDCBcIwplsaaFpRUg2EYtHJSfdJfMAyD3OZJ0OsYGPQ6jLblJOgZBm/Y5PNWHizCmHdWo7LeHtazbN95xfNx7DglN3gX7jgdUsUPhCo9XDxuMBAXZUC3FlZv8ZypeZg2uDVy0tV/j/UmC+77Zhv/3WRmUWmLRTZZWCzde44PI1F6XwRz0R+1V5ZBF3rmmFJFQ60TeneBUOSX7VYDY/Gecxj51irFNtx7Jrd5Ei7vnuXwfHcPa4tpg1vjlgE5jeqPsOKZFG75bew7q1FUWY+hb6zAy3/sw5+7/aOrG8wEw1ARbZArYSgZPI//uBPHiqpQXmsUJekA9mTMEBxHnCJdRj+v8axuC3+vGFzbu4VoX5VAuqtGEB6x+Xgp/udCKeCH5u8QFT8Idl5aJPeSBxudsxLx7MTOsup2TRLFuQSLBO9zk8Ui+i0Il9k5rXyhd1jYNlgQvptPlcodPRd9XGU0EARjfL9260YSXoXTqS28WKvaRkkmR424KAOendi50f2Z0K0Z7vtWeV95rREXKsUD/WdrjwNrj/M6yYQclmU1pfG84vHhfOz52idH8KoE0vAJAHhv+WHFc9Q0mDHyBWvlReG9D2dPsVSLWK2UrlbgnJ86hgHDMIgy6Hgptjqj/fd64Jw9l2DypxsAALcOzHGadKmE2cJqOgZXjd9DqPyx9O/foUmCqvfbZGZF1fuERBp0qKo3oaegAmt1ECZjc79jBsCQ2Stk+2/om+3nHvkerb+blAhDP0t4EhNpnf8cvqCexFah8lLyNxV1RhQUU8iEu2w9cTHQXRDROj0OLW1JdS1SYnmd2lRJ1StHqGWWc9nuavJfoYxeEnv4xW3K8lVawSxRCrm6V3N+n8lBtUOOxkg5aVmFwxGXNEt03ihI0LkxKTFbWF59QkqkQYdz5bUOfwerDxVh/ZFit/uoFVqmxqJT09C59xzCiW6wQEZxmBBji8tylMRWEaAlqU+n9kYzQVJGea2R9xRJqawz4kSJcnU0JUa8uRIvKiTuAEC9yYwXft8rW5oPVoLFe+JO9Tq12EHOmHJl3P3PL7v5pJ1QQBpTLK06qTWE4ROAOHbSmcFrtrDo8txi0TZu6dmRB1lp1euTVUcx7t3VrnU6QDhT5gkmpL9TRw7/ZfvP4y+bXrmUhTvO4JqPlccDjlu+2Iwpn21yu49a4Z3rewS6Cx7zwhVdAt0Fr0BGcZjAOdyKq7QX43Npl6bY8NQo/nulA+P8uk82YNgbK10+9/HianyhIPEEAAu2nsaX6wrwzt+HAFg1FR/7YSeKKoNvycdiYfHhiiOB7obLfHtnP5famVSMJs6YktrEDSYLhsz+B8v3W5O2Zi7cg282ncTCHcG7LG0yW7DpWAkAYPn+8ygJsjg9ztvPhX1IJfkcUWs0iwznJXvPYcjsFbjli82iIj9L94oNKqUiJq/9dUDzniuj2YK4AFaf9CYZCeIYYtZBxsMLv+9zaDSHiuMCUE60CxatZUfcOjAn0F3wCmQUhwnzNhQ4bfP1NNcMFW/x1bS+WPXv4W4d486g9vriA4rbNx8vxfHian6JlZP3WnnwAhZsK8Rzv+1xq0+BhmVZrDtajM3HSwPdFZcZ2E7u3eyuICdWb7J7v+uMZl7uiTOWpTHF5yvqcKq0FtP+lw8AmLfhhNf6HChm/XkA18/ZiG83ncS0/+XzklUA0DkIltu72TRsO2dZ+3r7oNai6piOkIZ07Tlt1WhefahIFCJxUPJeCNaqmGYLy+v0cgxokxag3nhGp6aJ+OmeAfhSEN4j9R6HE9wvUml1pG2GNpRjfE0w3H4yil2kut4U1LPVAhdkzQa39+8y7JD2GSIZqQ1PjcRLV7q2BONK8s3HK48qbp/86QaMeHOlrFRwii3W9ViRPTwj7+Vlml96f2/5YUz9fLPq/qn9W/mxN42jb04q+uSkyrYLE7E6PbsYuc8tAcuy+GSV9d5KjWJhHGPhxeCPS2dZll/pePqX3bL937jocQ8kV3TPwponRmBI+wwA1gSssZ2bunTs+yoJmIBYgeAt22oPR3W9CR+uOCKaVHE0JnHPX5gsLAw6HR4b0wGANcb427u0f4/VyMtJ5Su1sSxg8LEWr5bvLYd0JXTeHX2RJqnyGaoEQw4IGcUuMmT2CnR/YanzhhrElaWZNDeSn3xFs6QYNEuKcaltg9mCepPZo+xW7u/Cl6y2fRdOfoqr6jW/9P7uMnXD4fLuWXjpqlw/9sZ13rm+O967oQeemXAJPripJyIUYo1rGsQDiMnCoqS6AbttHkPu3i3ccRql1Q0inddg1MiU4ijm9u9HhvITOa2TnSoOmXB1uXj+llOi78IhVapkIDzn3DXH8caSg/jvugLZObW8VG0yW2DQM/zfi0FwGBKO4LrPsr7X4v16o/ZXhqTvtCC/vW6h5WePI+yM4toGM/7Y5b6R40hvr95k1kxN75wZizBbEjbgrG+ZCVH47YHBvuyWy6TGuzbIN5gseG7hXuS9vAxV9Y1LEOQ8xVyso9H2wHLOBuEydbBidiG7P1BM6tkCV/ZojjuHtEFmQrSiEau0rabe/ntmGGvi1UPzd+Ch+dtFS+pSSSiWZbHmcBFyZizCYltSz7wNBfhMUiFOSzgaRNo3SVDdp3Ua69ETJpNKw6NWHrzAf+Ymy8KVBo7TZbWoFrwz6oxmzeQRmGxScpx3NRR0uIVPodBTPLBtGl+e3FvMXWNdVSmqrMf8zSdRUWdEZZ02JsfcT766Xjweh1Kp+gdHtceIjhmYM7U3nh7fid/OFfAJBkLgkXONhTtOo/PMxZj+VT7u/3a7S/JVf+87j++3nMSsP/c7bHfpO6vR6dnFDtv4k49sYQMVdUa8/Mc+p2Efv90/mJfLCjSuxkg2mCx8DO0fjdT25OyNz9Yex9K95/jkHC4h5L5vt6kdGjSoJappkQqFwWvD0RLZthqj3aCJ0Ov444oq60WeValBWW+y8GEmL/y+FwAwc+FevLzI8fMdSIwantR4gpqqiLPwqc/X2pNm10okuLg4csDuxHhn2SHc/uVm0W9r2Bsrcc3H6/nvN3+2CX1eWeZ6532I2cLCoGP4kr9c4YpgJiE6AgCQnRqDJy61G0qRBh2Gdczw6rU4A/iJn3Zixs+70e35pej6/FL8kH/KyZG+h/vFSz3FwRBn6yqPjumAL2/vi7FdmmL60Lb89raZwRMzHfxPnBvUNJhRYJPzmr/5pKxAhJT5m0/i/eVHMGe1Y0+SK/G6/sAkkSF65+9D+GztcczfLK4d/+zEzvjk5t7894Ro7dRwiY7Q4/ZBOU7bnSitQUKM9WU742drrOXIN1c6ncAIsQgG5ulfbcXCEBLO5wiG5SoOpSpVi/fKZZqEnpbEmAjeG2jQMyLJwd8lK0LCUBtXCzscPFeJOqMZUz/fhEvf8b+clzmIJjXuoCYjHKVQ2KUxCJ0eKw4W4TPJO1yYsJuvIX1vLqaYCyVyR75Qq3RtkYRPbu6NF67IxZR+LTF/en8A1lCKCAdG/9JHhrp9rYs1RhRerJHd0yd+2uX2uXyF1FMcQo5iGS9flYub+rXEo7YY+Z4tkwPcI+cE/xPnItzDxw0yP24txK1fbHF4TK3RrCoorkXqJTJE3BJhjUS/NkLPiLKAYzUmAfTc5V0wRJD0p9S/qz9aj52S0sDHiqsxZ/UxlNcYZeL9G4/JPY6zFx8UfeeqSZVUNWDgq8tV+/fNphNYfajI+X8kgGTZdJ/7tJYnr2kVzsvTqanjsAChl2/RrrO421YSds/pCpF3/9NVYkPornn20rGVdSYMfv0f/vv6o8XYd6YCD8/fjnJbyEZVvQmXvrsaj/6wA2sOF+Pgef/LeYWqpzglNkJxu6+Kbqh5prUGF1PMEaVQEj0YGZfbFDG293jfnFQ8OLIdZk3KdTg57aASHtTLiWE1+PUVirKe9SYzxr272iUlJl8iiykOYav45v6t8Mqkrogy6DG8Y0ZQOGnCzygWvBz3n63A2fJa1JvMWH9UXg2nzmgOmoIIVfUmmZeUWzr/zLbkyNWeZxhG5IHQYiLH57f2wWe35OHxsR1citcWPmzdX5Qvl90wZ6PL1zZZWJwpF68ilNca+X7855c9uOULdbUHf1FRZ8Sot1bKto/r0hTf3tUfa54YgelDvBuz50tS46y/zw9v6uXWca4mW+4/W8F/Lq81ikqeT5m7CePfX4Nfd5xBd1s52Xrb/V6vEMLhL4JhEGkMdwxuzX9+URAycUX3LEwT7PMWSjZxQXE1ftgS+GV1IVxMMVeuPRQ8xVJ0OgaPju2IzMRomURbZoJzFYamgkJP7lBS1YAD5yoxc+HeRh3vMbYf4aeSVYs+OSmB6I3fMeh0japO6W+88sQxDDOOYZiDDMMcYRhmhjfO6W24GC3pIDPg1X/w2ZrjmDJ3E9ZJYtSUkjSEVNebZCELgeLNJQfxzSZ7mMS2kxdlWp0x3NIky8qywbVGpEGH0Z2b4P6R7eGKXSCVXvrPL3Kt4ZFvrkRBsevV8IT0eulvDHejaIg/2HSsFEeL5P+fT6b2Rk56HLJTY90qtRpo3ryuG96/safLVb18vcLB/e4CqWKhFBM+pV9L9AuiFQAlIvQ6TM5rAUAcNxsTocetA3JEbW/sm42OHiYVSlfRAGD4myvxxAL7srpSTLs/KatpwLYTFxGh0/H9jQpBo1iILBnWhWNSYpWTse8Y5HgypZVkeCHf3tUPhhCIG3cFg47RdOI3h8d3g2EYPYAPAVwGoDOAGxmG6ezpeb0N5ylWqujGLWfc9NkmWbEAKVzWtMXCostzS3DnvHxZm0AgXZK5+qP1fClcDqFKQ3aKNbFOKwl2nrJ073mnbY4VV8tm6a5itrA4V1GH+zWUfBcMmpzukBwbiSu6Z8m2t0qLxcRuzWTbYyN9GwsvfX4CgVIRilmTuuL7uwcEoDfeRW+TVjAK/o96nXgV68BL4/DyVV2RFKMcbuEqrigOjQtAzLiQyZ9uQHWDGTqdPc/Dnep/wYizRywhyoBmEs9wVnIMPlJYTcpKduxBFjq5/LkCs+3kRVTXm2QG/18PDdF8iXZvYtAzQZH47Y0pSl8AR1iWPcaybAOA+QCu9MJ5vYqjgP6mAm3cnafK+c9KRrHRzKK4qp4vqbvyoDZiS5Uc1sKliiHt0zG0gzXTlwVg0Ovw0z0D8Mt9A/3UQ9/y8Pc7XGqnJObvDn8IZNp+2V7o0bkI10iJjcSIjpmibU+M64gWKb6b0OXMWCSrkhYIgsGz0li41TvhahvD2JUXAGvirV7HuBQT/ODIdqr7pJrGSkhDpvzNofPWJFGzhcXwDhn4YEpPPDy6Q0D75Guk95W785wDefvMMVjzxAhRG4OOQUvbSuclzRL5/Inc5vKKmEJqBeO5vwpxldcYcfVH6/HQfPn4FOfjSb3WiNDrgiJHwhtGcXMAwsCsQts2EQzDTGcYJp9hmPyiIv8bkhEOKun8n6BqUnW9CSdtahJ1CktuVfUmTP50g6yCUqBRGjyFmfgf3NiLfwi591BeTioyExoXn+VPruwh9x5+OMW9uFMOpWXUxvLI9ztRXW/C0aIq542JRsNCHlv5r+HtcLa8VvkAJwxs61rZ3G83nXTeyMf8tjP4tbLV4BwVUu+RUhEXqcecy2YXwnme/zW8LQ6+PM5b3fQ5tQ1mPoYYsL6jGIbBxG5ZDp05oYA0/JBLb+H+3wa9ThZeIAy5EBY3iYnQo0MT9dArYcnwBi+OA46oszlhlu0/jzNl9knX/SPa+XRSr0WeHNcJX0/TfnVGvz1xLMvOYVk2j2XZvIwM72oTuoKjl8sFgXD77f/dgqFvrABgfVlJ+dc3W0VlgLWC0qrEcUH8bHRk8L5c35ncA09dZte3vHtYm0brW9Z7Oa7s5s83YdRbq7x6TjXKahpECaHaX4hqPKKEG5ZVfH5d8f4p4epSfKD/vmU1DQ7LHAc7dw9tgyHt03GdLbaYQ0mbVzjpH9UpE5PzsmVtOINKr2OCSrXhkpmL0eGZv/jv/jLYtIA0f4BTYnA0XqspVsRE6h0ed/t/7WpTvlI5kSJ0hAsTgh+/tGNQ5Xt4g6ZJ0WiVpn29Ym9YSqcBCN9QLWzbNIW7M26WZflZnpCNx0pVj/lr91mc88MSnMlswd/7zotiSi1Olhcj9TpBuc1AD/fuodMxiI2yLzVFGfSNFrVftv+C80ZusP2kVRbO4ocYtcmfbsCUuZt470qQ3Ua32PjUKP4zC+8lHEUZdEhWkQOTEujnZJ9ALSMUyUyMxlfT+iFZkjil9K4W2jC5zZNEsmUcofI8aDEhzFekxEWi4LUJeP/GngDEExsh8+7oi3RbtVO9jkG8bTxoL/AMRyp4ldXwl1Hsr+sQ3sMbI80WAO0ZhmnNMEwkgBsA/OaF83oVR+ETSjSYLW6/ZO/9ZhsufXc1Tpc1blnXVT5ZdRR3zcvHPwfsBp4zo0yLsmvuECV42UUZdG7fT19z6qLvC7hwMYdV9SbsOV2OHRKd5lBC6EWxsKyiNNW8O/qqHj9dpXysjmGQFONaKXEl9Rl/VsYSSsaFE0qeQKGnmGGACAf1j7mjb+rXUrT9pSu7YPN/RuHa3i3kB6kw8s2VeGXRPpfbu0pFnVE1v8GbIV7BQoTtnnP3TjoxGtohA/3aWMOedAyDnPQ4fDWtL167uhv/bmAYiGZG6fGReHyscky2UgKrLwjHexnseGwUsyxrAnA/gCUA9gP4gWXZAAkBquOup/j3RsbyldcaMei1f/iiHxcq6rBga6FXvU7Hi60GGFfKFHDtIdcFsWEcFSE2irVm5A97Y6XPPDwHzlXgVKnd6L7/2+2Y+H9r8cmqoz65nlb48R6rwgLLKuu1psXLjduPbuqF16/piku7NFE8p17HIDHGtQQXqaILALzqRsVET5EWDrqyR1ZQxOQ1hvdu6IGp/Vup7u+TY5egY8AoeoqlvDKpq+j7jX1bIjMh2uWYcsCqWDN3zXHnDd2k2/NLMei1FciZsUi2L5jf042F8/By73UlpwdXeItbNRrSPgMxkXrMvSUPtw/KQXZKrChxb8NTo1THCfIUE2p4ZU2SZdk/WZbtwLJsW5ZlX/HGOb2Nu0bx4z/u9Oh6023Vs275YjMe+3En7+XzBlyohNCj8vc+55JkD49uj8l5LTC5jzweT+tESjzFQr69qx9mX9NNdsxjYzpg9CXKxpEaN/Zt/N9m3Lt2Sac3lxxEzoxFXpkMjXt3DYbMXsF/X3tEXmiGY/HDQzy+nlaItsWFWljl51dJa3t812a4vk9LdGqaCABoky6OYWMAJEa7Fj5Rq+Ap9udkrLRanCH/3g09Mbh9aEo4XdmjOV66Kle0TRj7/cIVXfgSsQwDl4xiAEiPt8amH501nje84qICm/XPTbbUis7835Se/uyOJpA+3kr3lzMwoyWlwNtlxuO5y7tAp2NEYTYRep2sbW5z63tBScfeF4RTfHioELzZV27ijeV2oYC8s7jEvWes0m5cadhKF4ThzRYWG1yonsVpLHJGsSPD66WrcnFs1nhbnyMx+9ruPtd39QVCT6E0iWZg23RFQ/+BUe1d9gpypMU5r6ikRkGJ3Zv7gU2yb/n+C3h4/vZGn9NdWqdrP5HBVbiBkWVZ6AXG6IJ7rR7kxOgIRb1SwGr4FLw2QVQ5DbAaVFcoqJkosd+PMb1V9SZ8tfEE/yyfKKmWFRMKJxbcOwBLHxnKfzfodRjczjohYGCtjiWFVUiNXPzwECx6cLDIgRBoKayCYvVQq78fGYpeLcOjwpkQ6WRTKTymwWbxxkSoJ1FKwwi7t7DLtBl0DJ4cZ03Y3nGqTDRultcYsXCH91OhlDzFYZZfF3SEj1HshUSdZgJx8KaJjqXM6k0WmC0sH+LEFc74a/dZzFmtvOz9wT9HcOPcjdh0zLFhzBnF3DJbg4MlmpgIfUhkuYqMYlsoxdfT+uGfx4Y5PM7dpcgYF6qkTc5zPSbxznn5+HXHGZ8nbX01rS++ubNfUGXdO4O7dywLcGNkp6YJ6N3KvpTuLOHyml4t8MokuwdSp2OQGB2BHtnJjepTaXUDHp6/XdUDdLKkBg9+t91tPexXFu3Hs7/uwerDVkN42Bsrsfu0XTN94X2DGtXfYKV3q1Q0kbxjuZWB7NRYVQUCKenxUeiSJdav7Z6dhKt7NkeGQknhyjojHvthp2JYg7dw9NuIDbAXO1BwzzpXMEfJU8w9c1LvrxBpwnnvVilIsP1NIw060WSq9VN/4quNJwAA93+3DQ/N34FCL+eGKL0nQjUEKlQIG6PYG9nr3HKejnFeCc5kYTH49X/47zU2ebd7v9mGWX8eUDxmx6mLAIDKOnksoxDOKOZeAHUN6kZxqJQJFZb05f5Pg9uno42TksDu/v/VjKy0uEgse3QYCl6bgA4OSs7+sr1QcUBtbM33+75xrYJek8RoDGoXWkvrvFoKWJGBLMTZZDcmUo+h7e3yfd6I1/x1xxlVZYinf9mN33aewSabSs3KgxdEsf9qcPHDVXUmbD95Uba/NowUCdS4rncLfHNnP0XdcgC4fWBrTOzWTLY6ICUhOgJvX99DsVR21+eXYsE2a1EeoYbui7/bk+2u/Xi9qlReea0RN322UVVDu3p2W9EAAB2KSURBVMFkQYlCVVWOOB+XLtcq3ByHMyKVVgI4r6swv0SK1ChmGAbX21YRIw06RBrEz//XG6xGMTcB9XbFNanDqk16HAaG2Hs61AgNi8kFvOFBy7V5HCIE8mYAcN+Itortzwrk2arrHRu6AFBhM4bvnJfvsOIOl1THZbbuP6e+zGsIAS8xAHRrYffsKSVdqSG97w+Pbu+wvTTMJiHagHUzRmLrs2PQLjPe1kb9+o98rxyL7k7ChfC3smi3awmf0SHkIeZIi7Mm0o2+pAlvzEoHPeEkZoSKdrXwWZU+Do7E/h2h9lgJQ5vqjGbc9uUW3P7lZqfn4+JdTRYLJn20XrY/VOTGPIFhGAxql64a150UG4EPpvSSSbyp4SzP5Oft9uX0L9bZk+3yT1zE2yrFm37bcRrrjpTwFU+l3PftNtw5L1/1msEY2uYNuOebG9OUwh05x4Kj+8adR6g8oredK1KvkxnbJosFHZ75C2U11vG2pLpx2udqSD3FX97ex6vnJ7xP2BjFgHXp1RG9bIkcSpVm/nhgMPrYPAvWR9P6oH06tTdSXYhD3XO6HEcu2MvGKi2nCyvubDiqHk/IGQbcC+SGORtlbSZ2awbAcWhFMCF8ESa4mCgFANESrwIXj6bmQZZ6Hn/51yDZqkBjqkwJjeIdp8pwTKUK3pELVejy3BIs2Fro0kSKw5H3JFhJi49C/jOj8djYjny4kqzQg83zkxQTgS9vV5do45AaVEoeWC45yxFVCvemzmgWhTZxz6crSbbc5LVGUjDo6p7NMfuabujfRu7VJDzjtBPJuyd+2uX2ORl+8ibeXm8y4/nf9iomRF/VIwu9W1njiN2Z8IcS3GPZvkkC2mfG49mJnWVtuHeoo5CpT6b2xrTBrfHSlfaQKT2vaKGThWVcqKwXGa7XfLxBNQGyMUhXCIOheEW4E1ZPIOfpU+O1a7qhb+tULH54KLJTrYbQ9XnZmDmxM3KbJyHZFj7Bsiy/PzE6wiVv7P82nMDot+3qBDtOlYkG1g1HS1BQYq9Al+ig6hZnUKtVZ9v8n1H8i6Oxy/ZaplWaXHVADekg09+mdTnAJsvUq2Uy1s0Yye+XGrxKxrOrme9CuPtwrKgKV324DiNVquAdPGedOC0/cF600uCMUAmTkZIeHwW9jkFSbASOzhqPu4aI9Ye5++UoxjReEKcpbVarEHrULtP5wPXDllOYt6GA/15vMqPTs4uxuaCU7w832DorrAPYjeLZi8WhVc2SozG5T7bmJAi1wo19WzpvpMJWhTAVZwgTuZbvlxu4/IqGoF1JVT06PrMY/11foHjOtyb3wLd39cPm/4xS3B8OcH+3KL0Ofz86DHk58kkg5+CRhkAIaZthNaiFeTTcsxVl0MkMaqVH05sFuBrMFPYUbITmSKrCq1d3Vdw+pnMTDGybhg5NEvDD3QMQH2XA42M7AgDuH9mOj1ETSgTNuKwTPpzSCwPapskG5MyEKKcG+KSP1uPer7fi/eWHsbuwHDfO3SgyYB1JuehtS0ANZotihnxmQjSibB7RUNJJHNTOashmuODJ4xCGT/xn/CXonp2MAy+Nw4iOmQCALllJIk+w9KWp5LlpTDU97j6oGcMcnAHFgHFJsYQjlBLs1NDrGJlxyN0fR7HCybGR+J+t0Ie0nXRi+d/b++DTm/Oc9uXXHWcwc+FeGM0W1BnNOFsmHkj1OntClSuhD9xE62KN+J7HR7m+KhJuFLw2gX+nJzQiQW1Up0y3jxGuLEz7nzwMghsKzAKjeJmC8SyEK0udmeA4eTuUUQuPEsK9Q91dqePGywiFindK46w3x0ySZAs+wiqASWnZveC1CYptr+zRHBO7ZYkMXs57O21wG0QZ9JhgC1GQGsXTh7bBnUPa8AlXDKM8MK47Uow1h4vx8Uq5GoW0mtaxoipkJcegrMaIo7al93qjBT/mFyr2//GxHVBvNKsmpQQjc2/JQ0WtyS2v2SXNrCEzb17Xna9kFR2hh9oppEawkgHcmPAJVxM4eKOYAarrlb0MEXpGtgIQqp5iZ9g9xY7bcaFRV/VsDgB4YlxHPPjdDmQlR2NXoV3lYXhHsaF0/NXxaP3Un6rnfXPpQXy66pjiPm5AlEqFrT9SjCmfbcKqfw/nl1OVEosAcYIpoczC+wahWZL7BuX7N/ZEp2cXK+67plcLPuFOSLVCQRchnIdSGD7haML670s7utDT0Ce3eSLaZcbjqfGdVNt8OKUXPl19DC1SXF8pBOzvBiUVJqXwQm8askpVMQltE1ZGsbtIjV29jsHRWeNlS7DSdlIDJSZCL4sVBORxZ0Je+2s/htsShzYeK8FtX25Bs6Ro0ZJ6rdEs8l4LSYuPwtvX91C/QBASG2lwmIjy632DkF9QipcX2auODe+YicUPDxFpTAPgY/hGXSI2gqQGr5KnuDGa19ay4c4NY66JjmEU41YBq9KEtARwKMjuNQZu0qJ3MlFKiI7A7ufH8hq1A9umI/+Z0ThRUo0PVxzBDyqTS+EE7J5hbWVVBNUMYpOZ5WOKpc/5ykNFAIAFWwvRoWkC7v92O8Z0Vi4y05gJWLjRvZHyeo6kvYZ1zJAZxd1fWIrf7ncsjafnVVLsN91RnPB1bpScDmViIw1Y9qhjec1uLZLx4RRlXXJHcJ5ivU5cLlyN6+dsxMGXx3ll9c1XVU4J30FGsZsoxS5KB2TpEk2UQadoFHPERxtkCT8FJTUyL4Y0xvTLdccDXp1JS/TITkaP7GSRUQyAr24mpEtWEo7OGi+7n1KDV9EolmxLjDbwyiFqmCwWVDv4DXBwS3e/7zqD33aeUWxjXfFwnCQULvCTCBcmBUorRa3S4jD72u4yozhSr5N5kZ4c1xHTh7ZB75f/dhoSYbawdk+xpHE7m4xgYVkt9pyxhj8duaCcjBcq6jHBRs/sZGQmROFCpT3pqrzW6PA9Dtj1tIVhAI5Wcej97Xu44dig0yHTSX0BjkPnqtC1hVjfeuOxEnRtnuTWPasXeJ0dlTEntAO5IbyANPGKG8e4l6EztYSiysZluxrNLC8lo1bZi1BHahDnNk+UhWYoGSWcdzImQo/50/vjGYVMaSlGEyuT2TMrLBVwkyNHRld8FC2pc0RHWu9F9xaN8xZyvHRVLl/tCgDynx2NHTPHAAAGt0vHJc2sv43UuEiX5O+OFFXxeuPS28wZTCYz67RqnatFKojGMcUm3XVz/5a4vLs91CwuyqD47M9drbwyIEV4zx3FuzuqzkZ4B85TnBBtQGJ0hGrIpBCjxKN8obION8zZiEd/2KHYfunec8iZsUj2jhd6im/q3/ikUMJ/0DTVCyREi/+MnGG14alRqDOa8cj3O3Cy1LuVcqSEWuEGf7PvxUth0OlkSTFK8cvcYNk2Mw7926TxlbYccfkHa2XbGkwWWQU9Z54oAGiXmYAtBfbM+cYkGYUKmQnRWHDvAFzSTL4a4A5SL06iYCL79Z3iClQxkXqnxTRmLtyruu///rFq2FbXm3hPUo1KrGpjlE4I15k1qStmTbIm672xxK78ERel5/VthQi1i5XgcgeEnmK1xK2xnZuEbdiTP+FW/4Tx+d9P7483lx4UvUeFSOOKa2z5HfvPVio1x/v/WIu5FBRXo3t2MsprjVi695woppgB3etgIGw9xd70wAxtn4E7B7fmFSe4cIrUuEhkJcfgiXEdkRjtfcNFKDUl1eMl3CM20oBIg44fzG4bmKPqUeCSbTh96ubJMRjftamozZonRji95tTPN8m2OTOK37i2G5673O6Z3jlzLDY8Hb5SToC1JLA/ix544t2rqjfhdJk19GWLTb4NAC5WKyuNkKfYf0zqaY/vjTLoVZMfHcEVVvpj11k8+v0O0TYpampIhHfhVD2E79Z+bdJElS6lNQxe/euASFaP+6T2OHKTIe55ffmPffj3T7vw7eYTfBtSoggOwtaS2vbMGGz5z2ivnMug1+GZiZ2Rm2X1Vknfpb1bpWLX85eis4ferOEdM3jDu3lyDO4YlMPvc5ZoRLjGZbnN8OS4TnhinHpWOJes1Tcnhd82OS9bVDY2NU69qtaErlbVkvwTFzFk9j94e+lBfl+tisdw3h19MfvabrguLxvREXrERxmQ1yoFSbERoskR4XvUJqCuPN9Cj7AwDl2tyE6zJMfl5Anv0S4zHs9MuISvcujKhEQaLy4sDf3z9tNgWRZHVePFw3b49SsZCdZ3cYUktEGYaPn93QNE+3aeKsP2U2X8d7sqkPJvgguFW7jjNBZsLeQNcKGnmFZ9goOwG00/vqkXLtYYkRTrff1Ps0A5QImF9w+ChWXR8RllGSBnPDiqPS5U1OGer7ehzmhGveAFTB4lO69d3RVtnehEq6HXMbh3uHLZbo5+bdLw/fT+6CMQmB/eMRPDO2byMnyOVAMSY+yP3anSWrz/zxE8OrYjymoaUFzVoHhMbvMkDO1g92zseeFSl/4/hPeRhrxwZCREASpVuS0WFjodoyqzpwankkL4hzuHWOU0AXWvoBCjmYXRbIZBz+DA2UqcKKmR7X9LpSQ0GUn+oUtWEvrmpGKGRO6NS5jLTo1RVHFavOcszBYWfVun8l5j6R07UVKNZkkxvFE8d421HPgtA+zhWM2SovHkuE4eh3gR/iHsjOLLbF46X+BsNumpvFKb9DiU2xLrTILsdkfXDEdu8KDKlav0s1XGk6LXMTBbWD6OjfsuRC3xsseLf6teTxq3TgSOBFtBjcu7Z6FdRjwOXajEol1nkeZgdeCVP/fj35d2VCwtHGXQibLUORbcO0C2jfAfyTHq95OjwWxB7nNLVPc7ij0no9g/REfo8cM98meJ07bmikF9dkse7pxnL8gyd81xzF1zHEdnjbev5AhuWXmNEcPeWIlWabF8Ui1HSqz9t5MWH8nroxPah0ZaL8LNNuN8ILj/0z0DkBwbyUuEWVi7UfzYmA5evx7ROP58cAjWHSkGwzA4/MplOFdehyGzV4jaKMWXcx5mKU+O64SrezUnvVoN8cZ13fDxyqN4/oouiNDr8MB32wEA6QnqlRY/X3scX647rqhN3r1FMl8emsOgY9C7lbzULeE/3prcHVsKSvHoDztV2ziLE331z/2q+yh8IrCkxVsNVy7UYbSKVvi9X2/F0n3WBOxjRdW4/tMN+N8dfVFjtBrC0tUBQCzF1irVedl4Qjt4ZBQzDPMGgMsBNAA4CuB2lmXLHB8Vujw9/hK0SY/jSwh7E64WPGccWSwspg9tg12F5bhJkDkvLVJB+JeOTRPQ0Za0EaHXITs1Fm3S43CsuJpvk6hScEWJhGgDmriorUn4hxYpsXhlkj1JaubEzmiVGotB7dIxRyLZ1a1FEl8xT61YT4SBPIZaJDs1FtmpsWibEY8rP1yn2MaZUTx/yynVfRTyFljaZcYjOTbCaVVBziDm2HS8FP8cuICsZPV4f67EOwBMIi9xUOHpVPVvALksy3YDcAjAU553KXiJjzLgziFtnMrszL6mG/q3aZwXiFuWN7MsWqXF4fcHBvNJXYsfHqK4TEQEFukyaaIT3WohjqpuEdogIyEKj1/aUTEBb3jHTKcliLmJ7m0Dc/htP9JzrBkchS4dOFfhx54Q3iQ20oAdM8di1CV2D7Grz92/vtmGq1QmSgBwxqYwc1WPLAzpQHKpwYRHRjHLsktZluWCaTYCoJqVLjC5TzbmT2/coMd7ihW8Tp2aJqqWfSYCh1Sf0p344BYppD4QLCiFuMRG6p2qk94zzJrY+S9BgmfPlpRgpxXiBc/rYIke/G1fbnH7fLcNzMGnU3t73C/C+wiTpz1hyd7zyE6Nwbs39PRKuWjCf3gzqOkOAH+p7WQYZjrDMPkMw+QXFRV58bKhw+xruzltw1XJs6itxRKaQ5oD6UznljOaR3bKRH+VhD5CeyiVBE+JjXCaBNu/TRoKXpvgcglawr8IV3ZSHCRTusolzRJwaZemzhsSQQ0Zw8GJU6OYYZhlDMPsUfh3paDNfwCYAHyjdh6WZeewLJvHsmxeRkaGWrOwYmznJkhxUxrO7ikmozhYkBpFSkoDHAWvTeArbFEJ2OBCyVPcNClGpltOBBdRgslOn5wUWaEHabvpQ9s4PJ+jss9E6EAFtYITp+u4LMs6rHDBMMxtACYCGMVKlcwJh8y5JQ+AuvIAR3q8Pas9wqAePkFoE2mIuVSiTUqGTcXAlfLRhHaIkMSOv3hlFwxtn44IgVU8uF061h4p5vdz1e0I7SKc1E7t3wpX9miOX7YV4vnf94naReit0npmC4vV/x6BoW+skJ4KX03ri4FtKcY0HCBPcXDiqfrEOABPABjGsqxcl4RwiSu6Z+G3nWfstSQBzL0lD3/uPotZk7qKigVIB14ieLipX0vc1K8VOjSJx73D2+LnbYU4X1HP71/y8FAA1uX0L2/rg0HtaPAMJqThE7cMyAEARAk8/v1ap/JGMbef0D4zLuuETk0TwDAMkmIi0ExBeeDy7lmoN5nxr+FtVZUlBrdLJ035MIE8xcGJpzrFHwCIAvC37UHfyLLsPR73Ksy4qV9L/LbzjCh+dEznJhijoJsYpafZZ7BRXW/NRb28exY620qBPzmuE/7YdUbULi7Kfm9HdPK+rB/hWyIF4RPCz8LBMZ6KsAQlXDIkx5D28gnrhK7NMNi23aRStpsM4vAhmjzFQYlHb2iWZdt5qyPhTD9boo0r0SekaRp8lNdaqxBmSoo71DaIq10pJWoRwQMnxWjQMdg+cwy/XTg40j0ODWIjDXj5qlw88+seAMDm/4xCZoI9UdIgiS9vnxmPwxeq/NpHwr9kJEShqNK+8hdFnuKghO6ahuC8CFLjSQhVNgs+uBKgUnWBGolRTDFowU2sLUziuSu6IC7K7m+4e5g98YrucejQtXkSAOCTm3uLDGIlfrxnABY9ONgf3SI85OpGFtvo3CxR9J08xcEJWVgaY+4tefj1vkGq+w1UBSno+ObOfrixbzbio8QLM63SxOU/o8iLGNQY9DoUvDYBUwUVJgFrAY8JXZsBoHscSnTPTsbOmWMxLldZXu2PB+xGcHJsJLpkJfmra4QHvH19j0Yd12CyYNakrrh1QCvnjQnNQm9ojTGmcxOH5SM5b/KlXZTrtBPao1+bNLx6tVyDet4dfUXfaRUgdOFk+MgoDi2SHEhq5jYnIzgc+L8be6JT0wQ8M/ESTOnXEh2bWj3GtUazkyMJLUJZH0HIjpljRMuzRHCSkRCFxGgDKupM+PimXqoZ60Tw89zlnREXpcfQDo412tukx+FYcbWfekUQhDswDCBN/emTk4rFNuUgAEiLtxZ44cLmiOCC3BZBSHJsJHkVQ4QGW5Y6Va4LbbJTY/HeDT0RHaHHjX1bYlA75fu95JGhOPDSOD/3jvAlJM0VvDw4qr3o+67nxoq+63UM0uPFVQ45nfmKOqNvO0f4BHI3EkQAMZqtbgehFjUR2rx6dVfVfRF6HaiQYeiw/LFhojLRRHDAeYT7tU4VbU+Q3MumidEypZGkGGsbTnWICC5oCksQAeS9G3ogt3kixZoSRAjSNiOe9xwSwUNWkjWvp3lyDG4bmKPaLjNRfm9bpsaiX+tUvH6NPI+E0D40EhNEAJnYLQt/PDCERP0JgiA0wv/u6Ivr87KRnRqL56/ootpOqigEWFd7vr97AIXEBSkUPvH/7d1riFzlHcfx749EgzRRVxJTm0QTNW1JKWgqEvCC9bJJRIyWUhKkphew0kgrbZWooKJvGksVfKMoBhW8Y0MD9mIEad8Yb2lsEjXNJl1bQy60KaZF0Ub/vnieMSfrzuzM7smeyZzfB4Y98+zs7DO//Z+ZZ59zMzMzM8tOP3Eyq7498kzvFF+hsud4ptjMzMysDSu+efCS3z7gvff4L2pmZmbWhhsWfvWzqxPO7Gt+TQE7Mnnu38zMzKxNX/vScTz3k3M5bdrkqrtiJfOg2MzMzKwDvmx3b/Kg2MzMzKyJaVMm8cFHvmxzHXhQbGZmZtbESysvrLoLNk48KDYzMzNrYuhV66x3+S9tZmZmZrXnQbGZmZmZ1V4pg2JJP5cUkqaW8XxmZmZmZuNpzINiSbOAfuAfY++OmZmZmdn4K2Om+B7gRiBKeC4zMzMzs3E3prNPSFoC7IyINySN9NhrgGvy3f9J2jqW3z1KU4F/VfB7e5XzLI+zLJfzLI+zLJfzLI+zLE+vZ3lKOw9SROsJXkkvAF8c5lu3ADcD/RHxnqRB4KyI6NpQJb0WEWdV3Y9e4TzL4yzL5TzL4yzL5TzL4yzL4yyTEWeKI+Li4dolfR2YAzRmiWcCGySdHRG7S+2lmZmZmdlhNOrdJyJiE3Bi4/6RMFNsZmZmZjacup2n+IGqO9BjnGd5nGW5nGd5nGW5nGd5nGV5nCVt7FNsZmZmZtbr6jZTbGZmZmb2OR4Um5mZmVnt1WZQLGmRpK2SBiStrLo/3U7SLEkvSnpT0hZJP83tt0vaKWljvl1a+Jmbcr5bJS2srvfdSdKgpE05t9dy2wmS1knalr/25XZJujfn+VdJ86vtffeQ9JVC/W2UtF/S9a7N9klaLWmvpM2Fto5rUdLy/PhtkpZX8Vqq1iTLX0l6O+e1RtLxuX22pA8KNXp/4We+kd8fBnLerU/+34OaZNnxeu3P+6RJnk8VshyUtDG3uzYBIqLnb8AEYDtwKnA08AYwr+p+dfMNOAmYn5enAH8D5gG3A78Y5vHzcq6TSKfq2w5MqPp1dNMNGASmDmm7C1iZl1cCq/LypcDvAQELgJer7n833vK6vZt0YnbXZvu5nQ/MBzYX2jqqReAEYEf+2peX+6p+bV2SZT8wMS+vKmQ5u/i4Ic/zSs5XOe/FVb+2Lsmyo/Xan/et8xzy/V8Dt+Zl12ZEbWaKzwYGImJHRHwEPAksqbhPXS0idkXEhrz8X+AtYEaLH1kCPBkRH0bE34EBUu7W2hLgkbz8CHBFof3RSNYDx0s6qYoOdrmLgO0R8U6Lx7g2h4iIPwP7hjR3WosLgXURsS8i/gOsAxYd/t53l+GyjIjnI+JAvruedB7/pnKex0bE+kijkEc5mH9tNKnLZpqt1/68z1rlmWd7vwM80eo56labdRkUzwD+Wbj/Lq0HeFYgaTZwJvBybroubxZc3djEijNuRwDPS3pd6bLnANMjYlde3g1Mz8vOsz1LOfRN3bU5ep3WonNtzw9Is2sNcyT9RdKfJJ2X22aQ8mtwlofqZL12XbbnPGBPRGwrtNW+NusyKLZRkjQZeBa4PiL2A/cBpwFnALtIm1+sPedGxHxgMbBC0vnFb+b/wn2OxDZJOhq4HHgmN7k2S+JaLIekW4ADwGO5aRdwckScCfwMeFzSsVX17wjh9frwWMahEwquTeozKN4JzCrcn5nbrAVJR5EGxI9FxG8AImJPRHwcEZ8AD3JwM7QzHkFE7Mxf9wJrSNntaewWkb/uzQ93niNbDGyIiD3g2ixBp7XoXFuQ9D3gMuCq/E8GeVP/v/Py66R9X79Myq24i4WzzEaxXrsuRyBpIvAt4KlGm2szqcug+FVgrqQ5eXZpKbC24j51tby/0UPAWxFxd6G9uF/rlUDjqNa1wFJJkyTNAeaSds43QNIXJE1pLJMOxNlMyq1x1P5y4Ld5eS1wdT7yfwHwXmHTtiWHzHS4Nses01r8I9AvqS9v0u7PbbUnaRFwI3B5RLxfaJ8maUJePpVUiztynvslLcjvvVdzMP9aG8V67c/7kV0MvB0Rn+0W4dpMJlbdgfEQEQckXUd6w54ArI6ILRV3q9udA3wX2NQ4ZQtwM7BM0hmkTauDwI8AImKLpKeBN0mbC1dExMfj3uvuNR1Yk89kMxF4PCL+IOlV4GlJPwTeIR34APA70lH/A8D7wPfHv8vdK/9jcQm5/rK7XJvtkfQEcAEwVdK7wG3AL+mgFiNin6Q7SYMQgDsiot2DpHpGkyxvIp0VYV1e59dHxLWkswHcIen/wCfAtYXMfgw8DBxD2ge5uB9yLTTJ8oJO12t/3ifD5RkRD/H5YzHAtQn4Ms9mZmZmZrXZfcLMzMzMrCkPis3MzMys9jwoNjMzM7Pa86DYzMzMzGrPg2IzMzMzqz0Pis3MzMys9jwoNjMzM7Pa+xRQLEcjf7ZsaAAAAABJRU5ErkJggg==", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from sklearn.dummy import DummyRegressor\n", "from sklego.meta import GroupedEstimator, DecayEstimator\n", "from sklego.datasets import make_simpleseries\n", "\n", "yt = make_simpleseries(seed=1)\n", "df = (pd.DataFrame({\"yt\": yt,\n", " \"date\": pd.date_range(\"2000-01-01\", periods=len(yt))})\n", " .assign(m=lambda d: d.date.dt.month)\n", " .reset_index())\n", "\n", "plt.figure(figsize=(12, 3))\n", "plt.plot(make_simpleseries(seed=1));" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We will create two models on this dataset. One model calculates the average\n", "value per month in our timeseries and the other does the same thing but will\n", "decay the importance of making accurate predictions for the far history.\n" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "mod1 = (GroupedEstimator(DummyRegressor(), groups=[\"m\"])\n", " .fit(df[['m']], df['yt']))\n", "\n", "\n", "mod2 = (GroupedEstimator(DecayEstimator(DummyRegressor(), decay=0.9), groups=[\"m\"])\n", " .fit(df[['index', 'm']], df['yt']))\n", "\n", "plt.figure(figsize=(12, 3))\n", "plt.plot(df['yt'], alpha=0.5);\n", "plt.plot(mod1.predict(df[['m']]), label=\"grouped\")\n", "plt.plot(mod2.predict(df[['index', 'm']]), label=\"decayed\")\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The decay parameter has a lot of influence on the effect of the model but one\n", "can clearly see that we shift focus to the more recent data.\n", "\n", "## Confusion Balancer \n", "\n", "**Disclaimer**: This is an experimental feature. \n", "\n", "We added an experimental feature to the meta estimators that can be used to force balance in the confusion matrix of an estimator. The approach works " ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "n1, n2, n3 = 100, 500, 50\n", "np.random.seed(42)\n", "X = np.concatenate([np.random.normal(0, 1, (n1, 2)), \n", " np.random.normal(2, 1, (n2, 2)),\n", " np.random.normal(3, 1, (n3, 2))], \n", " axis=0)\n", "y = np.concatenate([np.zeros((n1, 1)), \n", " np.ones((n2, 1)),\n", " np.zeros((n3, 1))], \n", " axis=0).reshape(-1)\n", "plt.scatter(X[:, 0], X[:, 1], c=y);" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's take this dataset and train a simple classifier against it." ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "from sklearn.metrics import confusion_matrix" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 72, 78],\n", " [ 4, 496]])" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "mod = LogisticRegression(solver='lbfgs', multi_class='multinomial', max_iter=10000)\n", "cfm = confusion_matrix(y, mod.fit(X, y).predict(X))\n", "cfm" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The confusion matrix is not ideal. This is in part because the dataset is slightly imbalanced but in general it is also because of the way the algorithm works. Let's see if we can learn something else from this confusion matrix. I might transform the counts into probabilities." ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0.94736842, 0.05263158],\n", " [0.1358885 , 0.8641115 ]])" ] }, "execution_count": 14, "metadata": {}, "output_type": "execute_result" } ], "source": [ "cfm.T / cfm.T.sum(axis=1).reshape(-1, 1)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Let's consider the number 0.1359 in the lower left corner. This number represents the probability that the actually class 0 while the model predicts class 1. In math we might write this as $P(C_1 | M_1)$ where $C_i$ denotes the actual label while $M_i$ denotes the label given by the algorithm. \n", "\n", "The idea now is that we might rebalance our original predictions $P(M_i)$ by multiplying them;\n", "\n", "$$ P_{\\text{corrected}}(C_1) = P(C_1|M_0) p(M_0) + P(C_1|M_1) p(M_1) $$ \n", "\n", "In general this can be written as; \n", "\n", "$$ P_{\\text{corrected}}(C_i) = \\sum_j P(C_i|M_j) p(M_j) $$\n", "\n", "In laymens terms; we might be able to use the confusion matrix to learn from our mistakes. By how much we correct is something that we can tune with a hyperparameter. \n", "\n", "$$ P_{\\text{corrected}}(C_i) = \\alpha \\sum_j P(C_i|M_j) p(M_j) + (1-\\alpha) p(M_j) $$\n", "\n", "We'll perform an optimistic demonstration below." ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [], "source": [ "def false_positives(mod, x, y):\n", " return (mod.predict(x) != y)[y == 1].sum()\n", "\n", "def false_negatives(mod, x, y):\n", " return (mod.predict(x) != y)[y == 0].sum()" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "from sklego.meta import ConfusionBalancer" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [], "source": [ "cf_mod = ConfusionBalancer(LogisticRegression(solver='lbfgs', max_iter=1000), alpha=1.0)\n", "\n", "grid = GridSearchCV(cf_mod, \n", " param_grid={'alpha': np.linspace(-1.0, 3.0, 31)},\n", " scoring={\n", " \"accuracy\": make_scorer(accuracy_score),\n", " \"positives\": false_positives,\n", " \"negatives\": false_negatives\n", " },\n", " n_jobs=-1,\n", " iid=True,\n", " return_train_score=True,\n", " refit=\"negatives\",\n", " cv=5)" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsMAAADFCAYAAACxSv92AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3XlYlWX6wPHvw74ddgQEFNxQ3BVNJZfKNR2XxszSsjRtz5aZsuZXjU0zY+U0ZVm2qe1plpOpFVouLW6oqLmCK5uKKKvsPL8/3iOgoqICL0fuz3W91znvem5ADzfPud/7UVprhBBCCCGEaIjszA5ACCGEEEIIs0gyLIQQQgghGixJhoUQQgghRIMlybAQQgghhGiwJBkWQgghhBANliTDQgghhBCiwZJkWAghrgFKqcFKqb1KqUSl1LQq9jdVSv2klNqulFqtlAo1I04hhKhvlPQZFkII26aUsgf2AQOAZGATcLvWelelY74ClmqtP1JK3Qjco7W+05SAhRCiHpGRYSGEsH3dgUSt9QGtdRHwJTDinGOigJ+tz1dVsV8IIRokh7p8MX9/fx0eHl6XLymEEDVi8+bNJ7TWAWbHcQEhQFKl9WTgunOO2QbcArwBjAIsSik/rXVG5YOUUlOAKQDu7u5dW7duXWtBCyFEbbmc9+w6TYbDw8OJi4ury5cUQogaoZQ6bHYMV+kvwFtKqbuBtUAKUHruQVrr94D3AKKjo7W8ZwshbNHlvGfXaTIshBCiVqQAYZXWQ63bymmtUzFGhlFKeQB/1lpn1lmEQghRT0nNsBBC2L5NQEulVIRSygkYCyypfIBSyl8pdeY9/xlgbh3HKIQQ9ZIkw0IIYeO01iXAw8CPwG5godZ6p1LqRaXUcOth/YC9Sql9QCDwT1OCFUKIesb0Moni4mKSk5MpKCgwOxRxmVxcXAgNDcXR0dHsUIRo8LTWy4Hl52x7vtLzRcCiuo5LCCHqO9OT4eTkZCwWC+Hh4SilzA5HVJPWmoyMDJKTk4mIiDA7HCEuSWvNr4kn+HJjEq+P7YSjvXwwJkSN0Bry0iErGbJTjMeCbCgrqVh0WaX1Uuu2asxzYGcPAZEQ1B4C24Gbb+1/PaLBMT0ZLigokETYBiml8PPzIz093exQhLikrUdO8coPe1l3IIMQb1eSTp6mWYCH2WEJYb7iAiORzUuH0yehtOjiCWxZiTXxTYGsJGvymwKlhVVf386hYlH2RnJbvl6NP0hL8mHLRxXrXmFGYlx58W4K5+QQOQXFpGYWkJqVT2pmPmmZBeQXn9c85Tzero50aepDpzBv3J1NT5FEHanWT1op5Q18ALQDNDAR2AssAMKBQ8AYrfWpKwlCEmHbJD83Ud8lHMvh1R/3ErvrGH7uTrzwpyjuuK4Jzg72ZocmRO0qyK40UpsE2anWpPdERfKbdwIKsy//2soOLMHgFQrBnaD1UCNJ9QwxtnmFgquPkfhaaa3JyrcmqJn5pGXlczKvuFov51aUgX9eAgF5ewnI24f/kd347P0BO8oAKLR3J90plDTViIMljdhT6ENisR9HdCNSdADFOGCnwM3p0ilPXlEJWoO9naJNsIXopr50bepDdLgPwV6ul/+9Ejahun/2vAH8oLUebb1T2Q14FvhJaz1DKTUNmAY8XUtxCiFEtSWfOs3rKxP4Zksy7k4OPDGgFROvj8BDRnqErSvOPzuZzUuHnDTrSG2lMoVzk1xlB25+4B4A7v7QuEvFc/cAY3HzAwfn8tHbnCLNoZMFJJ4oYH9GAfvS89l/Ip9M7Y5biTOWXEc8ShzwPO2A5ZQjFhcH65JFUckpUjPzScnMJy3LSIBPF116ZPbCfIGe1gVcKCRSJRFld5ioksO0KE2nid1+OpWtx1EVg5NxlkZR5hGMnW9TVEBraNzZWBq1Afvz73fJyi9m65FTbD58irhDp1iwKYn5vx8CIMTblS5NfejZzI9buoTg4ih/VF8rLvmbQSnlBfQB7gawTvVZpJQagXF3MsBHwGpsNBmeNWsW77zzDl26dOGzzz6r8pjVq1czc+ZMli5dWsfRnW/JkiXs2rWLadOm8b///Y9WrVoRFRUFwPPPP0+fPn3o37+/yVEKUfdO5Bby1s+JfL7hCCi4t3czHujbHB93J7NDE+J8xQWQfwryTxqPp0+evZ5/CvIyKiW/6VCUW/W13PzBKwR8m0F4b+O5Vyi5zkEkFnlzMN+dMnWRX/l5UJxdxsETeew5msO+YzmkZVXc2O7h7ECrQG+6tQ3D0d6O3MIScgqKyS4oISWzgJyCHOu2EkrLjFpgfw9nQrxdaNnIgz4tA2js7UJjb1dj8XLB190Juxr6hNHOznqdsjLjj4NThyDzMOrUYewzDxvrf3wDm+cZx9k7Q1A7Y2T7TIIc0BovV0f6RTaiX2QjAIpLy9idlk3cISNB3ngwg++2pfL6yn08cmMLbuvWBCcHuf/A1il9iQJ2pVQnjNmIdgEdgc3AVCBFa+1tPUYBp86sn3N++dSeTZo06Xr48NkTguzevZs2bdpc/VdyFVq3bs3KlSsJDQ294DH1KRmu7O6772bYsGGMHj3alNevDz8/IQAWxiXx9yU7KSwpY0x0KI/e1LJGP9ZUSm3WWkfX2AVtgMxAd5nKyoxENudoFSUJx88vUSg+feFr2TsZpQZu/sborUej80dy3RuV7yvAicTjueWJ7N4qEtrqcHKwo0WAB5FBFloFWmgdZKFVkIXGXi7VKo3TWpNfXIq9nap/5UhlZXDqIKRuhbR4SLUuRTnGfgcXCO4ITXtB0+uhyXXgbDnrElprNh48yX9i97Hx0ElCvF2Z2r8lt3QOwUFuyq1XLuc9uzqfGToAXYBHtNYblFJvYJRElNNaa6VUlVn1uVN7XuyFpn+3k12pV1C/dBFRjT154U9tL7j//vvv58CBAwwZMoSJEycSExPD1KlTKSgowNXVlXnz5hEZGXnWOWvWrGHq1KmAUTe7du1aLBYLr776KgsXLqSwsJBRo0Yxffr0817Pw8ODyZMnExsbS1BQEF9++SUBAQHEx8dz//33c/r0aZo3b87cuXPx8fFh1qxZzJkzBwcHB6Kiovjyyy+ZP38+cXFx3HHHHSxZsoQ1a9bw0ksv8fXXX/OPf/yDYcOG4eHhwYcffshXX30FnJ3Mx8bG8sILL1BYWEjz5s2ZN28eHh4eTJs2jSVLluDg4MDAgQOZOXNmDf4khKg9vyac4JlvdtA93Jd/jmonN8eJmlNcAIU5RtlBQZYxWptz1Bh9zDkKuUet69alrIo6WDuHsxNZvxZGkuvmayyuPuDqi3b1Jt/eixw7C9kljmQXlpaPwOYUWB/zS8g5VVKxXpDG0eyDHM7IwzogW57Q9mjmR2SQhchAC0393HCwu3iyZmcHQZ4uV5XUKaWqVZtrCjs78GtuLO2tA0hlZXDygJEgp26F5I3w+5vw63+NG/7OJMfh10OTHihXH65r5seC+3qwNuEE/4ndy1OLtjNnzX4e79+Koe2DK0aphc2ozr/YZCBZa73Bur4IIxk+ppQK1lqnKaWCgeO1FWRtmjNnDj/88AOrVq3C39+f7OxsfvnlFxwcHFi5ciXPPvssX3/99VnnzJw5k9mzZxMTE0Nubi4uLi7ExsaSkJDAxo0b0VozfPhw1q5dS58+fc46Ny8vj+joaP773//y4osvMn36dN566y3uuusu3nzzTfr27cvzzz/P9OnTef3115kxYwYHDx7E2dmZzMyzZ07t1asXw4cPr3JkuH///kyZMoW8vDzc3d1ZsGABY8eO5cSJE7z00kusXLkSd3d3Xn75ZV577TUeeughFi9ezJ49e1BKnfdaQtRXB0/k8dDnW2gR4MH7E6KlLlicrawUSgorEtmLliVkGklvYY5xA1phttFd4UJcvMESBJYgyprGUOTaiHyXRuQ6+pNj702m8uaU8uJkqSvZhaXkFJSQW2hNbLMrJ7kl5BQUkFuYTJlOvuiXo5RRsuDp4oiHs1Gj2zrIwvCOjctHc8P93GSUsrrs7MC/hbF0uNXYVpQHSRvh8G9w6DfY+B6sewtQRnu35v1QPR6ib6tg+rT0J3bXMV6L3ccjX2xl9qpEnhwYSf82jeQmcxtyyd8aWuujSqkkpVSk1novcBNGycQuYAIww/r47dUGc7ER3LqSlZXFhAkTSEhIQClFcfH5f+XHxMTwxBNPMG7cOG655RZCQ0OJjY0lNjaWzp07A5Cbm0tCQsJ5ybCdnR233XYbAOPHj+eWW24hKyuLzMxM+vbtC8CECRO49VbjP2WHDh0YN24cI0eOZOTIkdX+OhwcHBg8eDDfffcdo0ePZtmyZbzyyiusWbOGXbt2ERMTA0BRURE9e/bEy8sLFxcXJk2axLBhwxg2bNjlf/OEqGPZBcXc+9Em7BR8IImwbdLaSD4KsysS0PLnOUYpQVGe9fE0FOdZHyttLymAkiLjsbTISH5LCo12X2UlF399eydwtY7OuniBRyD4tQQXT3D2BGcLhQ4eHMy1Z89JOJDrRHKpF8nFnmQU2pGbVULOsZIqbg4rA05aF4OjvcLiUulGM2dHwnzdsLgYya3FxcGa4Fqfuzjg6eJQ6RxH3BztZeSxtjm5Q/MbjAWMTwdS4ozE+PBvsH4ObPwAej2M6vUog9oG0b9NIEu3p/L6ygQmfxxHpzBvXr+tE+H+7uZ+LaJaqvub4xHgM2sniQPAPRhTOS9USk0CDgNjaifEuvXcc89xww03sHjxYg4dOkS/fv3OO2batGkMHTqU5cuXExMTw48//ojWmmeeeYb77rvvsl7vUn85Llu2jLVr1/Ldd9/xz3/+kx07dlT72mPHjuWtt97C19eX6OhoLBYLWmsGDBjAF198cd7xGzdu5KeffmLRokW89dZb/Pzzz5f1tQhRl0rLNFO/2MrhjNN8Muk6wnzdzA5JAKz6N+SkXjg5PfO8OB8Ks4yEV5dd+rp2DuDoDk5u4OhmfXQHJw+j9MDeyaj5dHAybo5ysC72zsY2F6/ycgTj0ccoUXB0O69HbUpmPnGHThodBXadYs/RbMo02Clo6ueOl6sjFlcHWvtUjM5eKIGtnNxK9wEb5ehilEmEX2+snzwIP/8D1r4KcfOg3zTsu97NiE4hDG0fzDdbUvj397u54/31LLivp7w32YBqJcNa63igqiLkm2o2HPNlZWUREhICwPz586s8Zv/+/bRv35727duzadMm9uzZw6BBg3juuecYN24cHh4epKSk4OjoSKNGjc46t6ysjEWLFjF27Fg+//xzrr/+ery8vPDx8eGXX36hd+/efPLJJ/Tt25eysjKSkpK44YYbuP766/nyyy/JzT37TmKLxUJOTk6Vcfbt25eJEyfy/vvvM3bsWAB69OjBQw89RGJiIi1atCAvL4+UlBQaN27M6dOnufnmm4mJiaFZs2ZX+Z0Uona98uMeVu1N56WR7ejZ3M/scMQZiSuMNl8O1uS0cmLqbDFqZR2cjCTU2dM6Amup9NzLWD+z3dHNSHgdaq8jSPKp06zem876AxlsPnyq/KYzNyd7Ojfx5uEbWxLd1IfOTbyxuMj08w2ebwSMngs9H4LY52H5X2D9O9D/BRzaDGdMtzDahXhx+/vrGffBBhbe15MgLxezoxYXIZ8pnuOpp55iwoQJvPTSSwwdOrTKY15//XVWrVqFnZ0dbdu2ZciQITg7O7N792569jR6IHp4ePDpp5+elwy7u7uzceNGXnrpJRo1asSCBQsA+Oijj8pvoGvWrBnz5s2jtLSU8ePHk5WVhdaaRx99FG/vsxt2jB07lsmTJzNr1iwWLVp01j57e3uGDRvG/Pnz+egjYwafgIAA5s+fz+23305hoTFj0EsvvYTFYmHEiBEUFBSgtea11167+m+mELXkmy3JvLvmAON7NGF8j6ZmhyMqm1z/P1EqKikj7vBJVu9NZ9We4yQcNwYZgjxdiA73IbqpD9HhvrQOskjtrbiwkK5w91JIiIUVz8PCuyC0Owz8B1FNevDxxO6M+2ADd3ywngVTehJgcTY7YnEBl2ytVpOqatPT0FpzeXh4nDe6a8sa2s9PmG/rkVPc9t56ujTx5pNJ1+FYR8mKtFazbUezCli99zir9h7nt8QMcgtLcLRXXBfhR7/IAPpFNqJ5gLvc9CSuTGkJxH8Gq/5ldBhpPQwG/oNN2d7c9eFGmvi68cWUHvhKz/M6U9Ot1YQQol44mlXAfZ9sJtDTmbfHda2zRFjYjrIyTdKp0+w9avTa3Xssh91p2exPzwOgsZcLwzs1pl+rAHq18JebLkXNsHeArhOMlm3r3obfXofZPejW+wnmjr+Tuz/Zzp0fbuDzyT3wcpVSm/pG3gXq2LU0KixEXSooLmXKJ3HkFZbwyaQYGWFp4E4XlZCaWUBKZj4JlSaZ2Hcsl/ziis4OYb6utGpkYUx0GP0iG9Eq0ENGf0XtcXKHvn+FzuPgx7/B6n/T0+cLvrrpb/x5pebueRv5ZNJ18kdYPSM/DSFEvae15qlF29mRksV7d0YTGWS59EnCpp3KK2J/ei4pmfmkZhaQlpVPqvV5alY+mafPbnvp7+FM6yALt3dvQmSQB60CjZ677pJ0CDN4NoZb5xmjxcv+Qoc1k/m1aX9GHxrOxPl2fHRPd1ydpLtIfSHvEkKIem/OmgMs2ZbKXwdFMiAq0OxwRA3KKywh4Xgu+47mVEwlfCyH9JzCs47zdHGgsbcrjb1d6dLUm2AvV0Ks680D3PHzkJuTRD3UrB888Duse4vAta+y2vVXXksawQMflzJnQk9pt1dPSDIshKjXDp7I47UVe7m5fRAP9mtudjjiMmmtOZFbdPbIbmY+hzJOs/dYNkkn88uPdXG0o1Wghb6tAogMtNAi0INQb1eCvV3lY2VhuxycoPcT0P5W7H+Yxl/3LGD/kbW89cFjPDp5Ck4Ocu+D2eTdRQhRr/1z2W6c7O34+/C2UutpA2avSmR/ei5p1nKGtKwCikrOnlTDxdGOMB83OoZ6M6ZrGK2CLEQGWgjzdcNeZlcT1yrvMBj7GSSsoNHix/nLsaf58a2t3PTw2zg4yAixmSQZBmbNmsU777xDly5d+Oyzz6o8ZvXq1cycOZOlS5fWcXSX71//+hfPPvts+XqvXr34/fffTYxIiCvza8IJVu4+xlODI2lkkab1tmDp9jSyThcR7O1Kh1BvBrd1obG3K8FeLuVlDj5ujvKHjWi4Wg7A8ngcu+c/xKCUL1k+x41BD74mfwiaSJJh4O2332blypWEhoaaHUqNODcZlkRY2KKS0jL+sXQXYb6uTIyJMDscUU3LHrkeO/mlLsTFObrQZtL77Hk3n5uPzeXb9y38afJ0+b9jkvqVDH8/DY7uqNlrBrWHITMuuPv+++/nwIEDDBkyhIkTJxITE8PUqVMpKCjA1dWVefPmERkZedY5a9asYerUqQAopVi7di0Wi4VXX32VhQsXUlhYyKhRo5g+ffp5r+fh4cHUqVNZunQprq6ufPvttwQGBpKens7999/PkSNHAGOWu5iYGNLT07njjjtITU2lZ8+erFixgs2bN+Pv78/IkSNJSkqioKCAqVOnMmXKFKZNm0Z+fj6dOnWibdu2fPbZZ+UTfYwdO5Y777yzfGa9u+++m2HDhjFq1CimTZvG6tWrKSws5KGHHuK+++4jLS2N2267jezsbEpKSnjnnXfo3bt3Tf1khLioLzYlsfdYDnPGd5GbTGyI/DIXoprs7Gg95WMSZ9/CiLQ3+Ga+B6Pu+at8amKCBl+1PWfOHBo3bsyqVat4/PHHad26Nb/88gtbt27lxRdfPGuE9YyZM2cye/Zs4uPj+eWXX3B1dSU2NpaEhAQ2btxIfHw8mzdvZu3ateedm5eXR48ePdi2bRt9+vTh/fffB2Dq1Kk8/vjjbNq0ia+//pp7770XgOnTp3PjjTeyc+dORo8eXZ4sA8ydO5fNmzcTFxfHrFmzyMjIYMaMGbi6uhIfH39eycdtt93GwoULASgqKuKnn35i6NChfPjhh3h5ebFp0yY2bdrE+++/z8GDB/n8888ZNGgQ8fHxbNu2jU6dOtXY912Ii8nKL+a12L1cF+HLoLZBZocjhBC1w96B5g8s5KBnN0Yc/hfffPYOdTkzsDDUr5Hhi4zg1pWsrCwmTJhAQkICSimKi4vPOyYmJoYnnniCcePGccsttxAaGkpsbCyxsbF07twZMCbXSEhIoE+fPmed6+TkxLBhwwDo2rUrK1asAGDlypXs2rWr/Ljs7Gxyc3P59ddfWbx4MQCDBw/Gx8en/JhZs2aV70tKSiIhIQE/P78Lfm1Dhgxh6tSpFBYW8sMPP9CnT5/yRH779u0sWrSo/HuQkJBAt27dmDhxIsXFxYwcOVKSYVFnZv2UQGZ+Mc//KUpGSYQQ1zTl6EL4Q/8jadYghiU8x9eLLIy+9U6zw2pQ6lcyXA8899xz3HDDDSxevJhDhw7Rr1+/846ZNm0aQ4cOZfny5cTExPDjjz+iteaZZ57hvvvuu+j1HR0rbhyxt7enpKQEgLKyMtavX4+LS/VuElq9ejUrV65k3bp1uLm50a9fPwoKCi56jouLC/369ePHH39kwYIFjB07FjBaH7355psMGjTovHPWrl3LsmXLuPvuu3niiSe46667qhWfEFdqf3ouH/1+iLHdwmjb2MvscIQQotYpZw9CH1zK0Tdv4uY/nuQbJzduGfFns8NqMBp8mcS5srKyCAkJAWD+/PlVHrN//37at2/P008/Tbdu3dizZw+DBg1i7ty55dMtp6SkcPz48Wq/7sCBA3nzzTfL1+Pj4wFjFPpMaUNsbCynTp0qj9PHxwc3Nzf27NnD+vXry891dHSsckQbjFKJefPm8csvvzB48GAABg0axDvvvFN+zr59+8jLy+Pw4cMEBgYyefJk7r33XrZs2VLtr0eIK/XPZbtxcbTniQGRlz5YCCGuEXbuPgQ++D15Tv703/Iwi7//weyQGoxqJcNKqUNKqR1KqXilVJx1m69SaoVSKsH66HOp69iCp556imeeeYbOnTuXj9qe6/XXX6ddu3Z06NABR0dHhgwZwsCBA7njjjvo2bMn7du3Z/To0eTk5FT7dWfNmkVcXBwdOnQgKiqKOXPmAPDCCy8QGxtLu3bt+OqrrwgKCsJisTB48GBKSkpo06YN06ZNo0ePHuXXmjJlCh06dGDcuHHnvc7AgQNZs2YN/fv3x8nJCYB7772XqKgounTpQrt27bjvvvsoKSlh9erVdOzYkc6dO7NgwYLymwaFqC1r9qXz857jPHJjCwIsMqPY5VBKDVZK7VVKJSqlplWxv4lSapVSaqtSartS6mYz4hRCXJi9ZyA+9y+nxNGD69dPZsnPa8wOqUFQ1SnUVkodAqK11icqbXsFOKm1nmF94/XRWj99setER0fruLi4s7bt3r2bNm3aXEnsDUJhYSH29vY4ODiwbt06HnjggfJR4/pAfn6ippSUljHkjV8oKi0j9vE+ONezJvRKqc1a62iz46iKUsoe2AcMAJKBTcDtWutdlY55D9iqtX5HKRUFLNdah1/sulW9Zwshal/Rsb0UvDuQnFJ7dgz8ksEx3c0OyeZcznv21ZRJjAA+sj7/CBh5FdcSF3DkyBG6detGx44defTRR8u7TwhxrflswxESjufy7M1t6l0ibAO6A4la6wNa6yLgS4z36Mo04Gl97gWk1mF8QojL4BQYifM9/8PbroDmKyaSl1v9T5rF5atuMqyBWKXUZqXUFOu2QK11mvX5USCwqhOVUlOUUnFKqbj09PSrDLfhadmyJVu3bmXbtm1s2rSJbt26mR2SEDUu83QR/125j17N/RgYVeVbibi4ECCp0nqydVtlfwfGK6WSgeXAI1VdSN6zhagfnMM6kzrgbVqSxJEvnzA7nGtadZPh67XWXYAhwENKqbP6hWmj1qLKegut9Xta62itdXRAQECVF5eeerZJfm6ipry+MoHs/GKeGyat1GrR7cB8rXUocDPwiVLqvN8B1XnPFkLUjZa9RrLE7RbaJC+kdNdSs8O5ZlUrGdZap1gfjwOLMT6SO6aUCgawPla/dUIlLi4uZGRkSGJlY7TWZGRkVLsVnBAXkng8h0/WH2Zs9ya0Cfa89AmiKilAWKX1UOu2yiYBCwG01usAF8C/TqITQlwx18HT2VEWTuniByFbqptqwyX7DCul3AE7rXWO9flA4EVgCTABmGF9/PZKAggNDSU5ORn5OM72uLi4EBoaanYYwsa9tGw3bk72PDmgldmh2LJNQEulVARGEjwWuOOcY44ANwHzlVJtMJJheeMVop67qV0Yd//4FO/mPYH+ZjLqriVgJ/dV1KTqTLoRCCy2fnTpAHyutf5BKbUJWKiUmgQcBsZcSQCOjo5ERERcyalCCBsXn5TJ6r3pPDOkNX4e0krtSmmtS5RSDwM/AvbAXK31TqXUi0Cc1noJ8CTwvlLqcYyytru1fCQnRL1nZ6cY0rc3z397F68eeg9+/S/0+YvZYV1TLpkMa60PAB2r2J6BMcoghBBX5ON1h3B3sueO65qYHYrN01ovx7gxrvK25ys93wXE1HVcQoirN6pzCP/5cSDrnffRY9W/IKIvhMkN9TVFZqATQpgiI7eQpdvSuKVLKBYXR7PDEUKIesvF0Z67Y8KZcnIcxe7B8PUkKMgyO6xrhiTDQghTLIhLoqi0jLt6NjU7FCGEqPfG92hKiZMn7/g/A1nJsOxJkEqnGiHJsBCizpWWaT5bf4SezfxoGWgxOxwhhKj3vN2cGBMdxqx9vuT0eBJ2fAXbvjQ7rGuCJMNCiDr30+5jpGTmM6GXjAoLIUR1Tbo+Ag28VTICmsbA8r9Axn6zw7J5kgwLIercx+sOE+zlQv82MtucEEJUV5ivGze3D+azjSnkDH0b7ByM+uGSIrNDs2mSDAsh6lTi8Vx+TTzBuOua4GAvb0FCCHE57uvTjNzCEj7fXQrD34TUrfDzP8wOy6bJbyIhRJ36dP1hnOztGNtd2qkJIcTlahfiRa/mfsz77RBFrYZBlwnw+5tw8oDZodksSYaFEHUmt7CERZuTubl9EP4yyYYQQlyRKX2acTS7gO+2pUK/Z4yQsQQhAAAgAElEQVRyifXvmB2WzZJkWAhRZxZvTSG3sIS7eoWbHYoQQtisvq0CiAy08P4vB9CWIGh/K2z9FPJPmR2aTZJkWAhRJ7TWfPz7IdqFeNI5zNvscIQQwmYppZjcpxl7juawZl869HwQik9D3DyzQ7NJkgwLIerEugMZJBzP5a6e4SilzA5HCCFs2vCOjQnydOG9tQcgqD006wcb35POEldAkmEhRJ34ZN1hvN0cGd6xsdmhCCGEzXNysOOemHB+35/BHylZ0PMRyEmDnd+YHZrNkWRYCFHr0rLyid11jNuiw3BxtDc7HCGEuCbcfl0TPJwdeHftAWhxEwS0ht/fkmmaL5Mkw0KIWvf5hiOUac34HjLjnBBC1BRPF0fG9WjC0u2p/L4/A3o+BMd2wME1ZodmUyQZFkLUqsKSUr7YeIQbIxsR5utmdjhCCHFNefTGljTzd+fRL+M5HjEc3ANg3Wyzw7Ip1U6GlVL2SqmtSqml1vUIpdQGpVSiUmqBUsqp9sIUQtiq73cc5URukbRTE0KIWuDu7MDb47qSW1jMo1/tpiz6XkiIhfS9ZodmMy5nZHgqsLvS+svAf7XWLYBTwKSaDEwIcW34eN0hIvzd6d3C3+xQhBDimhQZZOEfI9qx/sBJ5pzuBw4usO4ts8OyGdVKhpVSocBQ4APrugJuBBZZD/kIGFkbAQohbNcfKVlsOZLJ+B5NsbOTdmpCCFFbbo0O49auobz6awap4aNg2wLITTc7LJtQ3ZHh14GngDLruh+QqbUusa4nAyFVnaiUmqKUilNKxaWnyw9FiIbk43WHcHW0Z3TXULNDEUKIa96LI9rRqpGFhw70gNJC2PSB2SHZhEsmw0qpYcBxrfXmK3kBrfV7WutorXV0QEDAlVxCCGGDTuUV8W18KiM7h+Dl6mh2OEIIcc1zdbJn9rgu7C0JIs65O3rTB1Ccb3ZY9V51RoZjgOFKqUPAlxjlEW8A3kopB+sxoUBKrUQohLBJS7enUlhSxvgeTcwORQghGowWjTz416j2zMwZiDp9ArYvMDukeu+SybDW+hmtdajWOhwYC/ystR4HrAJGWw+bAHxba1EKIWzO0u1ptGjkQVSwp9mhCCFEgzKycwgRXQeyoyycvDVvQFnZpU9qwK6mz/DTwBNKqUSMGuIPayYkIYStO5ZdwMZDJxnaPhjjflshhBB16YXh7fje48+4Zx/gxLZlZodTr11WMqy1Xq21HmZ9fkBr3V1r3UJrfavWurB2QhRC2Jrvd6ShNQzrEGx2KEII0SC5ONpz64RHOar9SF3+KkUlMjp8ITIDnRCixi3dnkZkoIWWgRazQxFCiAYrItCbk+3voUPxNj76ZonZ4dRbkgwLIWpUWlY+cYdPyahwHVNKDVZK7bXOCjqtiv3/VUrFW5d9SqlMM+IUQtStqKGPUGjniu+O9/kjJcvscOolSYaFEDVq2fY0AIZKMlxnlFL2wGxgCBAF3K6Uiqp8jNb6ca11J611J+BN4Ju6j1QIUedcvaHznQy3W8fauHizo6mXJBkWQtSoZTvSiAr2pFmAh9mhNCTdgUTrvRxFGG0wR1zk+NuBL+okMiGE6Zx7P4JSELzzfbNDqZckGRZC1JjkU6fZeiRTRoXrXgiQVGn9YrOCNgUigJ8vsF9mDRXiWuPdhAONhzOk8AeOHD5gdjT1jiTDQogas3yHUSIh9cL12lhgkda6tKqdMmuoENcmy4CncKCUzJX/MTuUekeSYSFEjVm2PY32IV409XM3O5SGJgUIq7R+sVlBxyIlEkI0OMERUaxyvoHIpK8gVz71qUySYSFEjTiScZptyVkyKmyOTUBLpVSEUsoJI+E9r4+SUqo14AOsq+P4hBD1QHK7B3DQRZxe87rZodQrkgwLIWrEMmuJxM3tJRmua1rrEuBh4EdgN7BQa71TKfWiUmp4pUPHAl9qrbUZcQohzNWta3eWlPXCccuHkJdhdjj1hiTDQogasXR7Kp3CvAnzdTM7lAZJa71ca91Ka91ca/1P67bntdZLKh3zd631eT2IhRANQ9vGnnzlehv2pQWwfrbZ4dQbkgwLIa7awRN57EzNlhIJIYSox5RStGoXzfdlPdAb3oXTJ80OqV6QZFgIcdWWS4mEEELYhAFRgbxRPBJVlAsb5pgdTr0gybAQ4qp9ty2Vrk19aOztanYoQgghLqJ7hC9HnSPYZukL6+dAvszMLsmwEOKqJB7PZc/RHIbKqLAQQtR7jvZ23Ni6Ef/KHQqFWbDxPbNDMp0kw0KIq7JsexpKSYmEEELYigFRQWzID+Vk2ABYNxsKss0OyVSSDAshrsqyHal0a+pLkJeL2aEIIYSohr6RATjZ2/G1xx1QkNngR4cvmQwrpVyUUhuVUtuUUjuVUtOt2yOUUhuUUolKqQXWRu9CiAZk37Ec9h3LZVhHGRUWQghb4eHsQK8Wfnxy2AfdchCsewsKc8wOyzTVGRkuBG7UWncEOgGDlVI9gJeB/2qtWwCngEm1F6YQoj5auj0NOwWD2wWZHYoQQojLMCAqkCMnT3Ok3UOQfwo2fWh2SKa5ZDKsDbnWVUfrooEbgUXW7R8BI2slQiFEvaS1Zun2VK6L8KORRUokhBDClgxoEwjAkhONoflN8PubUJRnclTmqFbNsFLKXikVDxwHVgD7gUzrFKAAyUDIBc6dopSKU0rFpaen10TMQoh6YM/RHA6k5zFUJtoQQgib08jThU5h3qzYfQz6Pg2nT0DcPLPDMkW1kmGtdanWuhMQCnQHWlf3BbTW72mto7XW0QEBAVcYphCivllmLZEYIiUSQghhkwZEBbI9OYs0rw4Q0Rd+ewOK880Oq85dVjcJrXUmsAroCXgrpRysu0KBlBqOTQhRT50pkejV3B8/D2ezwxFCCHEFBrU1SiVW7rKODucdhy2fmBxV3atON4kApZS39bkrMADYjZEUj7YeNgH4traCFELULztTszmUcVpKJIQQwoY1D/Agwt+d2F3HIDwGQqJh47tQVmZ2aHWqOiPDwcAqpdR2YBOwQmu9FHgaeEIplQj4AQ33NkQhGhCtNW+vTsTBTjG4rZRICCGErVJKMSAqkPUHMsguKIbr7oeMRNj/s9mh1anqdJPYrrXurLXuoLVup7V+0br9gNa6u9a6hdb6Vq11Ye2HK4Qw2we/HGT5jqM8OTASH3dpLy6EELZsYFQgxaWa1XvTIWoEeAQao8MNiMxAJ4Sott/3n2DGD3sY0i6I+/s2MzscIYQQV6lzEx/83J1YsesYODhB9ERIiIWM/WaHVmckGRZCVEtqZj6PfL6VcD83Xr21I0ops0MSQghxleztFDe1acTqPccpKimDrveAnWODmqJZkmEhxCUVFJfywKebKSwp4907o/Fwdrj0SUIIIWzCwKggcgpLWH8gAyyB0O4W2PpZg5miWZJhIcQlTf9uJ9uSs5h5a0daNPIwOxwhhBA16PqW/rg62hulEgDd74OiHIj/wtzA6ogkw0KIi/py4xG+2JjEg/2aM1gm2BBCiGuOi6M9vVv6s2LXMbTWENq1QbVZk2RYCHFB8UmZPP/tTnq39OfJgZFmhyOEEKKWDGwbxNHsAnakZBkbGlCbNUmGhRBVysgt5MFPNxNgcWbW2M7Y28kNc0IIca26qXUj7O0Uy7anGRvOtFnbMMfcwOqAJMNCiPOUlJbxyBdbycgr4t07u0o/YSGEuMb5uDtxY+tGfL0lmeLSsoo2a4krrvk2a5IMCyHO8+qPe/l9fwb/HNWediFeZocjhBCiDtwWHcaJ3CJW7TlubGggbdYkGRZCnGX5jjTeXXuAO3s0ZXTXULPDEUIIUUf6RQYQYHFmYVySsaFym7WCbHODq0WSDAshyhUUl/LCkp10DPXiuWFRZocjhBCiDjnY2/HnLqGs2pvO8ewCY+OZNmvbrt02a5IMCyHKfb7hCOk5hTxzcxucHOTtwZYopQYrpfYqpRKVUtMucMwYpdQupdROpdTndR2jEKL+GxMdSmmZZtGWZGNDeZu1967ZNmv1+7ed1pB3wngUQtSqguJS5qzZT49mvvRo5md2OOIyKKXsgdnAECAKuF0pFXXOMS2BZ4AYrXVb4LE6D1QIUe81C/Cge7gvX8UlGz2H4Zpvs1a/51Q9fRJebQ72TmAJAktj8AwGzxCwBBvPK2+zdzQ7YiFs1hcbj3A8p5A3xnY2OxRx+boDiVrrAwBKqS+BEcCuSsdMBmZrrU8BaK2P13mUQgibcGt0KH9dtJ1Nh07RPcLXaLMW+zejzVrL/maHV+PqdzJs7wiDX4acVMhOg5w0SNsGe3+Akvyzj7VzBN9mENAK/FuBfyT4tzSeO8v0sUJczJlR4e4RvvRsLqPCNigESKq0ngxcd84xrQCUUr8B9sDftdY/nHshpdQUYApAkyZNaiVYIUT9NrRDMNO/28WCTUlGMnymzdrqf8OJRPBvYXaINeqSybBSKgz4GAgENPCe1voNpZQvsAAIBw4BY86MONQYF0/ocf/527WGgkxrgpwK2alw8gCcSIDje2DPctClFcd7hhhJsW+z80eTLcHG6wjRgC3YlMSx7EL+e1sns0MRtccBaAn0A0KBtUqp9lrrzMoHaa3fA94DiI6Olho1IRogNycH/tQxmP9tTeXvw6OwuDgabdbWzoRN78OQl80OsUZVZ2S4BHhSa71FKWUBNiulVgB3Az9prWdYb9aYBjxde6FWohS4+hhLYBV3vJcUwamDkL4XTuyrWHZ+A/lV5OtOHhVlF15h4NfCSJ4DIsEnAuzr9wC6EFejoLiUt1cn0j3cl55SK2yrUoCwSuuh1m2VJQMbtNbFwEGl1D6M5HhT3YQohLAlY6LD+GJjEku3p3F79yZnt1m74W/X1EDiJbM8rXUakGZ9nqOU2o3xkdwIjBEGgI+A1dRVMnwpDk5GIhsQef6+4nxjJDknrdLIcqXHxJ8g/rOK46sqvwiIhMB2kiSLa8LCOOuo8JhOKCVTLtuoTUBLpVQERhI8FrjjnGP+B9wOzFNK+WOUTRyo0yiFEDajU5g3rQI9WLApyUiGwWiztn0BxH9e9Sf3NuqysjmlVDjQGdgABFoTZYCjGGUUVZ1Tv+rPHF3Br7mxXEhBtlFyccI6spy+7/zyC2dPaBoDzfoZS0CkMWIthA0pLCnl7VX76RbuI7XCNkxrXaKUehj4EaMeeK7WeqdS6kUgTmu9xLpvoFJqF1AK/FVrnWFe1EKI+kwpxZjoMF5atpt9x3JoFWgx2qyFdoP1s6HbpGumcUG1k2GllAfwNfCY1jq78giS1lorpaqsLbPJ+jMXT+sPvOvZ28+UXxz7Aw7+AgdWw77vjX0eQdCsr5EYR/QFr5A6DlqIy7dwUxJHswuYeWtHGRW2cVrr5cDyc7Y9X+m5Bp6wLkIIcUmjOofw8g97WLApqWIipt5PwhdjYcci6HS7uQHWkGolw0opR4xE+DOt9TfWzceUUsFa6zSlVDBw7bfpqVx+0e7PxrZTh+HgGiMxTvzJ+PgAwK8lNOkBodEQ0hUC2khZhahXCktKeXv1fqKb+hDTQkaFhRBCnM3Pw5n+bQJZvDWFpwe3NiZjajXYKBX99TXoMAbs7M0O86pVp5uEAj4EdmutX6u0awkwAZhhffy2ViKs73yags9d0OUuY2aW47uMxPjgGtizFLZ+Yhzn4AqNOxmJ8ZnFu4mUVgjTLIxLJi2rgFdGd5BRYSGEEFUa0y2M7/84yk+7jzGkfbCRt/R+EhbdA7uXQNtRZod41aozVBkD3AnsUErFW7c9i5EEL1RKTQIOA2NqJ0QbYmcHQe2MpdfDRgu4UwchZQskx0HKZtj4PpS+ZRzv5m90rnD2AGeL0dXC2fOcdQu4ehvdLizBRgcNSVzEVSosKeWdVYl0berD9S38zQ5HCCFEPdWnZQBBni4siEsykmEwJuHwawlr/wNRI20+L6lON4lfgQt9lTfVbDjXGKWMThS+zaD9aGNbaTEc22kkxilbICvJmGkv8wgU5kBhLhTlXPiaDq7GbHyeIdaeycHg2Rg8AsHFy0ieyxNpD3CySHmGOM+izcmkZhUw488yKiyEEOLC7O0Uo7uG8vbqRNKy8gn2cjVKI3o/Af97APb9CJGDzQ7zqkiWVNfsHY1yicadjDsxq1JWBkW5xlKYY/RGLm8Hl1rxPGmj8VhadPHXdHCtSJI9GlUk0Gd6K5+ZfMQSbNRFi2taUUkZb6/aT+cm3vRuKaPCQgghLm5MdBhvrUrk683JPHxjS2Nj+1uNGenWvgqtBtn06LAkw/WRnZ3R0aI6Da21NkaWc49WjCwXZlck0mdGmgtzjJZxucchLR72fn/+lNZglG74hBs3Cfq3rOir7N1URpivEYs2J5OSmc+/bmkvo8JCCCEuqYmfGz2b+bEwLpkH+7XAzk4Zg3sxU2HZk8Z9Us36mR3mFZPsxtYpBe5+xnI5tDZGnM+dfCQ7xZja+tzJR+ydwLd5xeQj3k2s5RjWEedz65yvkd6D15qikjJmr0qkU5g3fWRUWAghRDXd1i2MxxbEs/5gBr2aW39/dBoPa141pmlu1s/M8K6KJMMNlVLg5mssgW2rPiY/0zr5yD7rBCQJcPQP2P0d6LKLX9/BBdz8ILgThHQxumc07mzcDChM8/UWY1T4n6PayaiwEEKIahvcLgjLtw4s3JRUkQw7ukCvRyD2b3BkAzS5ztwgr5Akw+LCXL0hrJuxVFZSCLnHrCUZORVlGOXr1lKNnKOQuhX2Lqs4169lRe/lkC5Gr0IH57r9uhqo4zkFzPopgY5h3vRtFWB2OEIIIWyIi6M9Izo15qu4ZKbnF+Plav0EOPoe+OU/8MtMGPeVuUFeIUmGxeVzcDbKJKor/5SRFJ/poJH4E2z7wtin7IwSjEvxiahIoEO6GqPZUopRbZmni7jzg41k5RczZ3xXGRUWQghx2W6LbsKn64+wZFsqd/Zoamx0coeeD8LPL0HaNgjuaG6QV0CSYVH7XH2g+Y3GAka9claykRwf++PS3TDKSo0SjX0/QPynxjYHFwjqYCTGodFGkuwTYdN3s9aW3MISJszdyMGMPObd3Y2OYVKqIoQQ4vK1C/GkXYgnc389yNhuYTja2xk7uk+B3940aodv+8TcIK+AJMOi7ikF3mHG0nZk9c/T2ujHnLK5Ytk8Hza8Y+x3shgdMAIijZv8/FsZz30iGmwnjILiUibN38TO1GzmjO9KjEywIYQQ4goppZh6UysmfxzH15uTGdvd+imxixd0n2yUSxzfA41amxvoZWqYGYKwTUpZp79uCu1uMbaVlkD6bmOGv+O7IH2vMR32mTIMADtHY+KTM50w/CON534tjS4Y16iikjIe+HQzGw+d5I2xnekfFWh2SEIIIWxc/zaN6NzEmzd+SmBk5xBcHO2NHT0ehPVvw6+vwS3vmRvkZZJkWNg2ewcIam8slRVkn98J4/ge2LMcdGnFcZ6hlUaTz/RVbg0etn2DWUlpGY8viGfV3nRm3NKe4R0bmx2SEEKIa4BSir8OiuSO9zfw6frD3Nu7mbHD3Q+iJ8L6d6DfM+AbYW6gl0GSYXFtcvGE0K7GUllJEZw6aIwgn9hnLOl7YcsnUJxXcZxnaMXNeiFdjRkDnS11+zVcobIyzbRvdrBsRxr/N7RNxcdYQgghRA3o1dyf3i39mb0qkdu6hWFxsd7Q3vNh2Pge/PY6/OkNc4O8DJIMi4bFwckYBQ6IPHu71saEI+l74fhuSN1i1CTvXmI9QBkjxmc6WoRGQ6OoetfRQmvNi0t3sWhzMo/1b1nxF7sQNaS4uJjk5GQKCgrMDkVcJhcXF0JDQ3F0rF/vW8I2/WVgJCNm/8aHvx7ksf6tjI2ewdD5TtjyMfR5CrxCzA2ymiQZFgKMemSvUGNpcVPF9ryMisQ4ZTPsXX52R4vgjhWjxyFdjamsTexo8dqKfcz//RD3Xh/B1JtamhaHuHYlJydjsVgIDw+XFn02RGtNRkYGycnJRETYzsfXov7qGObN4LZBfPDLQe7qGY6vu7VNasxU4+b232fBkJdNjbG6JBkW4mLc/aDlAGMBa0eLw8YNeynWJDlurnHTAICr79nJcVB7sATVSYI8Z81+3vw5kdu7h/G3oW0kURG1oqCgQBJhG6SUws/Pj/T0dLNDEdeQJwe2InbXUd5ZncjfhkYZG32aQqfbYeP7EHkzNOtrbpDVIMmwEJdDKWP01ycc2o82tpUWG6UV5S3ftsD+nyqmrHb2rLg5r/xmvUjjGjXU8m3ebweZ8f0ehndszEsj20uiImqV/PuyTfJzEzWtZaCFUZ1D+WjdYSZeH0Gwl6uxY9C/jUGjrybA5J+Njk712CV/Eyul5gLDgONa63bWbb7AAiAcOASM0Vqfqr0whajH7B0huIOxRN9jbCvMhbR4I0k+c7PegVWw7fOK8+wcwa+5MZufswWcPIzHM8ul1q31yvN/O8j073YxuG0Q/xnTEXs7+YUnhBCibjzWvyVLtqUw66dE/n2LtbOTiyfc/gW8fyN8PhbuXWlsq6eqMyw1H3gL+LjStmnAT1rrGUqpadb1p2s+PCFslLMHhF9vLJUVZMGJRKPdW7q15Vt2MmTsh8IcKMqF4tPVew17Z/Lt3Lih0JEbPT0JLQnE7ptA8GwMlmDjRgZL44p1h2pMey1EPZeZmcnnn3/Ogw8+eEXnv/7660yZMgU3N7cajkyIhinM141x1zXlk/WHmdKnGRH+7sYO32Yw5mP4ZBR8PQlu/xLs7M0N9gIumQxrrdcqpcLP2TwC6Gd9/hGwGkmGhbg0F6+qW75VVlpiJMVnkuPCHCjMNkaby9dz+eNgMvGJSbT0hm7BjtgVZhuj0Xu/h5L886/r5m8kyI7VSALsHK0j0B6VRqU9z153dLPd6a/De4Oji9lRiCuQmZnJ22+/fVXJ8Pjx401NhktKSnBwcLjguhC25qEbWrBgUxKvrdjHm7d3rtgR0QeGvALLnoCVL8DAl8wL8iKu9H9foNY6zfr8KHDBqa2UUlOAKQBNmki/UyEuyd4BXL2N5QI+WXeI5/bsZEBUIGPu6IKdg13FTq2hIBOy0yAn1fqYBtmpxmNJNVpilRQZI9aFlZLy6pxnK57YDY4yEcnVmv7dTnalZtfoNaMae/LCn9pecP+0adPYv38/nTp1YsCAAbz66qu8+uqrLFy4kMLCQkaNGsX06dPJy8tjzJgxJCcnU1paynPPPcexY8dITU3lhhtuwN/fn1WrVp117RdffJHvvvuO/Px8evXqxbvvvotSisTERO6//37S09Oxt7fnq6++onnz5rz88st8+umn2NnZMWTIEGbMmEG/fv2YOXMm0dHRnDhxgujoaA4dOsT8+fP55ptvyM3NpbS0lOnTp/Pcc8/h4+PDnj172LdvHyNHjiQpKYmCggKmTp3KlClTAPjhhx949tlnKS0txd/fnxUrVhAZGcnvv/9OQEAAZWVltGrVinXr1hEQYNsTBgnbFGBxZuL14cxetZ8H+jYnqnGlkohuk4wZYn9/02hJ2ukO8wK9gKv+U1RrrZVS+iL73wPeA4iOjr7gcUKI6vlk/WGe+3Yn/dsEMvuOLjhVToTBGK119TGWwKiae+HSYuuo9JlyjipGn22Fm7/ZEYgrNGPGDP744w/i4+MBiI2NJSEhgY0bN6K1Zvjw4axdu5b09HQaN27MsmXLAMjKysLLy4vXXnuNVatW4e9//r+Bhx9+mOeffx6AO++8k6VLl/KnP/2JcePGMW3aNEaNGkVBQQFlZWV8//33fPvtt2zYsAE3NzdOnjx5ydi3bNnC9u3b8fX1ZfXq1WzZsoU//vijvNXZ3Llz8fX1JT8/n27duvHnP/+ZsrIyJk+ezNq1a4mIiODkyZPY2dkxfvx4PvvsMx577DFWrlxJx44dJREWpprSuzmfrDvMzNi9zL2729k7B88w7p35bir4Nocm15kT5AVcaTJ8TCkVrLVOU0oFA8drMighRNU+23CY5/73B/3bNOLtcVUkwrXJ3hHcfI1FCLjoCG5diY2NJTY2ls6djY9mc3NzSUhIoHfv3jz55JM8/fTTDBs2jN69e1/yWqtWreKVV17h9OnTnDx5krZt29KvXz9SUlIYNWoUYExcAbBy5Uruueee8nILX99L/78YMGDAWcd17979rJ6/s2bNYvHixQAkJSWRkJBAeno6ffr0KT/uzPkTJ05kxIgRPPbYY8ydO5d77rnnkq8vRG3ycnPk/n7NeeWHvcQdOkl0eKX/E/aOcOtHxg11C8bB5FXgHWZesOe40t+kS4AJ1ucTgG9rJhwhxIV8vuEIf1v8Bze1bsTsuk6ERb2nlBqslNqrlEq03th87v67lVLpSql463KvGXHWNK01zzzzDPHx8cTHx5OYmMikSZNo1aoVW7ZsoX379vzf//0fL7744kWvU1BQwIMPPsiiRYvYsWMHkydPvqJZ9hwcHCgrKyu/ZmXu7u4XXF+9ejUrV65k3bp1bNu2jc6dO1/09cPCwggMDOTnn39m48aNDBky5LJjFaKm3d0rnACLM6/8sBetzykGcPOFOxZASSF8eTsU5ZkTZBUu+dtUKfUFsA6IVEolK6UmATOAAUqpBKC/dV0IUUs+33CEZxfv4MbWjXh7fBecHernHbnCHEope2A2MASIAm5XSlVVI7NAa93JunxQp0HWEIvFQk5OTvn6oEGDmDt3Lrm5uQCkpKRw/PhxUlNTcXNzY/z48fz1r39ly5b/b+/eo6qq8z6Ov78ckCOC3C8GpngNRcVEvK4yJ4p0wgtq98uMqS2qp7Fla5ge81Izzziubo8zZlnZWFnOpKXpomlpadZTVshYaDqKtxHHC4oSyE3g9/xxjggCcfHckO9rrbPY55zf2fuzN+f8+J7Nb++d3eDrL7pYeIaFhVFcXMyaNWtq2sfExLBu3ToAysvLKSkpITk5mTfffHIZOXkAABSGSURBVJOSEtvZXy4Ok+jevTs7duwAqJlHcxQWFhIcHIyfnx979+5l+/btAAwfPpxt27Zx6NChOssBeOihh7j33nuZOnUqFov2Ccr9/Dp489jYXnx7uIDP9zVwgZfwvjBlBZzcDR8+DPYvju7WZDFsjLnLGNPFGONjjIkxxrxhjDljjPmFMaa3MeZmY0zTg6WUUi2WX1TO/PW7eOrDHG7qG84yLYRVw5KAXGPMQWNMBbAa21l/rjqhoaGMGjWK+Ph4nnzySW655RbuvvtuRowYwYABA5gyZQpFRUXk5OSQlJREQkICCxcuZO7cuQDMnDmTlJQUbrrppjrzDQoKYsaMGcTHx3PrrbcydOilMY9vv/02S5YsYeDAgYwcOZITJ06QkpJCamoqiYmJJCQk8NxzzwEwZ84cli1bxuDBgzl9+nSz1yslJYXKykri4uLIyMhg+PDhAISHh7N8+XImT57MoEGDuOOOO2pek5qaSnFxsQ6RUB7lzqHX0jWkI4s+3kt5ZVX9Br2TIflZ2PMRfPas7UBtN5N6u7GdKDEx0WRlZblseUq1VYWlF3ht20FW/N8hyiuruSupK3PH98Pqo4Wwu4jIDmNMortzNEREpgApxpiH7PfvA4YZYx6t1eZB4I9APrAPmG2MOdrAvGqfAWjIkSNH6jy/Z88e4uLinLQmqiWysrKYPXs2X3zxRbNfo78/5QqbfjzJjLeyeGBENxZOiK/fwBhY/yjsfAcQ2zmJowbYbwNtPwOiruj0nS3ps/XEhkp5kNKKKlZ+fZhlWw9QWHqB2wddwxPJfS6dxFyp1tsAvGeMKReRWdjOET/28kZ6BqC2YdGiRSxbtoxVq1a5O4pS9ST3i2T66Fje+PIQSbGhjB/YpW4DEUhdAv1S4fj3cOIH23nyf1x3qY1f2KUCedCdEOm8A3a1GFbKA1yoqubvWUdZ8ul+Tv5Uzpi+4cy5pS/x0YHujqbahmNA7UOzY+yP1TDGnKl193VgsQtyKSfJyMggI6PecZJKeYzfplzHjiNn+e3aH+h/TWe6X75Tx8sCfW613S4qK7SNJz6RYyuQT+TAN69At5FaDCt1taquNmz44T+8uGkfh8+UMKRbMEvuHMywHqHujqbalu+A3iISi60IvhOoc2b7i6fDtN9NBfa4NqJSqj3p4O3F0nuuZ/ySL0hflc0H6SObHupnDbQVvt1GXnqs6oJtWIUTaTGslBscLShhbXYeH2Qf498FJVwXFcAbDyQy9roIpK1e4li5jTGmUkQeBT4BLMAKY8xuEXkGyDLGfAT8l4ikApVAAfCg2wIrpdqF6KCOvDBtEL/+axbPbPyR/5k0oOUzsfg4PthltBhWykWKyi6QmXOctTuO8e3hAkRgRI9Q5tzal18O6IKXlxbBqvWMMZlA5mWPzas1/Tvgd67OpZRq38ZeF8msG3vw6ucHGRYbwoSEaHdHqkeLYaWcqKra8GXuaT7IzuOT3Scou1BNj7BOPHlrXyYOjiY6qKO7IyqllFJONeeWvuw4fJanPsghPjqQnuH+7o5UhxbDSjmQMYYjZ0r4Pu8c//z3OT7edZyTP5UT2NGHKUNiSLs+hoSuQToUQqlWOnfuHO+++y7p6ektfu24ceN49913CQoKckIypVRjfCxe/PnuwYxf8iWPrMrmw/RRdOzgOacK1WJYqStwpric7/POsfNoITuPnuOHvHOcK7kAQEcfCyN7hjL/9hh+ERehF8tQygHOnTvHyy+/3GAxXFlZibd343/WMjMzG33OnYwxGGPw8rp0Hayqqiq9qpy6qnQJtI0f/tVfv2PBR7v505SB7o5UQ4thpX5G2YUqTv5UxonCMk78VGafLuc/50rZ9Z9C8s6WAuAl0CcygJT+UQzqGkRC1yB6R/jjbWnyIo9KtV0fZ9hOfeRIUQPgtkWNPp2RkcGBAwdISEggOTmZ8ePH8/TTTxMcHMzevXvZt28fEydO5OjRo5SVlfH4448zc+ZMwHap5KysLIqLi7ntttsYPXo0X331FdHR0axfv56OHesOW9qwYQO///3vqaioIDQ0lFWrVhEZGUlxcTGPPfYYWVlZiAjz588nLS2Nf/zjHzz11FNUVVURFhbGp59+yoIFC/D392fOnDkAxMfHs3HjRsB2Kelhw4axY8cOMjMz6d+/P7NmzWLz5s0sXbqUzz77jA0bNlBaWsrIkSN59dVXERFyc3N5+OGHyc/Px2Kx8P7777Nw4UImT57MxIkTAbjnnnuYNm0aEyZclRciVG3UmL4RPDKmF3/ZkktSbAhpQ2LcHQnQYli1c8YYzpyvIPdUMbmnijmQX8yh0+drit+Le3lr69TBQmSglUFdg3hgRHcGdQ0iProzfh3046SUsy1atIhdu3axc+dOALZu3Up2dja7du0iNjYWgBUrVhASEkJpaSlDhw4lLS2N0NC6pyvcv38/7733Hq+99hrTpk1j7dq13HvvvXXajB49mu3btyMivP766yxevJjnn3+eZ599lsDAQHJybF8Ezp49S35+PjNmzGDbtm3ExsZSUFDQ5Lrs37+flStX1lx6+fz58wwbNoznn38egH79+jFvnu0YyPvuu4+NGzdy++23c88995CRkcGkSZMoKyujurqa6dOn8+KLLzJx4kQKCwv56quvWLly5RVsaaWc4zc39+a7wwXMXbeLgTGB9I4McHckLYZV+1BVbcg7W1JT8Np+nif3VDGFpZcK3o4+FmLDOhET7Edi92CiOluJ7GylS2BHogJ9iexsJcDq/NO8KNUm/MweXFdKSkqqKYQBlixZwocffgjA0aNH2b9/f71iODY2loSEBACGDBnC4cOH6803Ly+PO+64g+PHj1NRUVGzjM2bN7N69eqadsHBwWzYsIEbbrihpk1ISEiTubt161ZTCANYLBbS0tJq7m/ZsoXFixdTUlJCQUEB/fv3Z8yYMRw7doxJkyYBYLVaAbjxxhtJT08nPz+ftWvXkpaW9rNDRpRyF2+LF3++azDjlnzBw+/s4NX7htArwr0FsX5S1FWl7EIVB/PPk5tfzIFTxTU/D54+T0VldU27MH9feoZ34pcDu9Az3J9eEf70jPCnS2ernuJMqTamU6dLV7baunUrmzdv5uuvv8bPz48xY8ZQVlZW7zW+vr410xaLhdLS0nptHnvsMZ544glSU1PZunUrCxYsaHE2b29vqqsv9T21s9TODbbC9uI44bKyMtLT08nKyqJr164sWLCgwfWo7f777+edd95h9erVvPnmmy3OqpSrRHS28ue7rmfm21nc+tIX3De8G7Nv7kOgn3t2NmkxrNqMqmrDmeJyTtjH8J78qcw+Xc7Jn8o4UnCevLOlNReq8RLoGuJHz3B/bugTTq9wf3pGdKJnuD9Bfh3cuzJKqVYJCAigqKio0ecLCwsJDg7Gz8+PvXv3sn379lYvq7CwkOho2zlRaw85SE5OZunSpbz00kuAbZjE8OHDSU9P59ChQzXDJEJCQujevXvNGOHs7GwOHTrUrGVfLHzDwsIoLi5mzZo1TJkyhYCAAGJiYli3bh0TJ06kvLycqqoq/Pz8ePDBB0lKSiIqKop+/fq1er2VcoURPUPZOmcML2zax1tfH2b9zmM8kdyHu5KudfnxNloMK5eorjaUXKjifHklRWWVnC+vpPjirazWdLn9ubJKisovtcsvKudUUTlV1XUvyWjxEiICbMMXBsUEkXZ9jG0vb7g/sWGdmr70o1KqTQkNDWXUqFHEx8dz2223MX78+DrPp6Sk8MorrxAXF0ffvn3rDENoqQULFjB16lSCg4MZO3ZsTSE7d+5cHnnkEeLj47FYLMyfP5/JkyezfPlyJk+eTHV1NREREWzatIm0tDTeeust+vfvz7Bhw+jTp0+zlh0UFMSMGTOIj48nKiqKoUOH1jz39ttvM2vWLObNm4ePjw/vv/8+PXr0IDIykri4uJqD6JTydKH+vvxh0gDuHd6NZzb8yNPrd/PO9n8z7/Z+jOoV5rIcYpx8vefaEhMTTVZWlsuWpxpXVW04X2ErOs+X1yo8yyqprG76PXGhqrpOMVtnHuWVFJdX1cyvuLyS8xWVzbq0uI9F8Pf1ppOvN/72Wydfb8L8fYkK9K0ZwxsVaCWqs5VQf18sOqxBuYCI7DDGJLo7hys11Gfv2bOHuLg4NyVSP6ekpIQBAwaQnZ1NYGBgg23096c8lTGGT3af4A+ZezhaUMot/SL57/FxdAvt1PSLG9CSPvuK9gyLSArwv4AFeN0Y49CjKcouVJGZc9yRs2yTKqtMnb2mRQ0UoKUVVTRVaxpjKKmoori8kpKKKofl8/aSOsWrv9WbwI4+RAdZ7Y/54G/1xt/Xgr+vD518LQRYvenUwVboBlgvFb2+3l56QQqllGqhzZs3M336dGbPnt1oIayUJxMRUuK7MKZvBG98eYilW3JJfmEbvx4dy6Nje+Hv67zBDK2es4hYgKVAMpAHfCciHxljfnRUuKKySp74+/eOmt1Vwdfbq07x6O/rTVRnKx07WPBqRhHp18FS89o687k43cGbDt5Nz8fby6vmNVrAKqWUe918880cOXLE3TGUumJWHwuP3NSLqUNiWPzJv3jl8wP0CO/EtMSuTlvmlZTZSUCuMeYggIisBiYADiuGg/18+PzJMY6aXZtl8bo0dMBHL+KglHIzY4x+AW6DXDksUqkrFdHZynNTB/GrUd25LqqzU5d1JcVwNHC01v08YNjljURkJjAT4Nprr21ZOItXq8eKKKWUcjyr1cqZM2cIDQ3VgrgNMcZw5syZmvMSK9VW9L/G+cN+nH42CWPMcmA52A7GcPbylFJKOU9MTAx5eXnk5+e7O4pqIavVSkyMZ1z+VilPciXF8DGg9gCOGPtjSimlrlI+Pj51rvamlFJt3ZUMQP0O6C0isSLSAbgT+MgxsZRSSimllHK+Vu8ZNsZUisijwCfYTq22whiz22HJlFJKKaWUcrIrGjNsjMkEMh2URSmllFJKKZdy6RXoRCQfaM2JEMOA0w6O0xqekgM8J4un5ADN0hBPyQGek6W1OboZY8IdHcaTaZ/tUJqlPk/JAZ6TxVNyQNvP0uw+26XFcGuJSJYnXAbVU3KA52TxlBygWTw5B3hOFk/JcTXzlG3sKTlAs3hyDvCcLJ6SA9pXFr2Cg1JKKaWUare0GFZKKaWUUu1WWymGl7s7gJ2n5ADPyeIpOUCzNMRTcoDnZPGUHFczT9nGnpIDNEtDPCUHeE4WT8kB7ShLmxgzrJRSSimllDO0lT3DSimllFJKOZwWw0oppZRSqt3yyGJYRKaKyG4RqRaRRk+lISIpIvIvEckVkQwn5AgRkU0ist/+M7iRdlUistN+c+glqZtaRxHxFZG/2Z//RkS6O3L5LcjxoIjk19oODzkpxwoROSUiuxp5XkRkiT3nDyJyvTNyNDPLGBEprLVN5jkpR1cR2SIiP9o/N4830MYl26WZWZy+XUTEKiLfisj39hwLG2jjks9Oe6B9dp15a59ddznaZ9dfjvbZ9Zfh3j7bGONxNyAO6AtsBRIbaWMBDgA9gA7A90A/B+dYDGTYpzOAPzXSrthJ26HJdQTSgVfs03cCf3NTjgeBv7jgvXEDcD2wq5HnxwEfAwIMB75xY5YxwEYXbJMuwPX26QBgXwO/H5dsl2Zmcfp2sa+nv33aB/gGGH5ZG6d/dtrLTfvs5q+j9tn1ntc+W/tst/fZHrln2BizxxjzryaaJQG5xpiDxpgKYDUwwcFRJgAr7dMrgYkOnn9TmrOOtTOuAX4hIuKGHC5hjNkGFPxMkwnAW8ZmOxAkIl3clMUljDHHjTHZ9ukiYA8QfVkzl2yXZmZxOvt6Ftvv+thvlx8t7IrPTrugfXYN7bMvo312gzm0z66fw619tkcWw80UDRytdT8Px/8CI40xx+3TJ4DIRtpZRSRLRLaLiCM73+asY00bY0wlUAiEOjBDc3MApNn/nbNGRLo6OENzueJ90RIj7P/2+VhE+jt7YfZ/Gw3G9q26Npdvl5/JAi7YLiJiEZGdwClgkzGm0W3ixM+OukT77MvaaJ8NaJ/dHe2zLy7fbX22tyNm0hoishmIauCp/zbGrPeEHLXvGGOMiDR2HrpuxphjItID+ExEcowxBxyd1cNtAN4zxpSLyCxs397GujmTu2Vje28Ui8g4YB3Q21kLExF/YC3wG2PMT85ajgOyuGS7GGOqgAQRCQI+FJF4Y0yDYwVV07TPvupon12f9tnttM92WzFsjLn5CmdxDKj9TTbG/pjDcojISRHpYow5bv/3xKlG5nHM/vOgiGzF9s3KER1rc9bxYps8EfEGAoEzDlh2i3IYY2ov83VsY/fcwSHvC0eo3aEYYzJF5GURCTPGnHb0skTEB1tHtsoY80EDTVy2XZrK4srtYl/GORHZAqQAtTtWV3x2rhraZzeL9tktp3229tmXL8/lfXZbHibxHdBbRGJFpAO2wdQOPSrYPr8H7NMPAPX2fohIsIj42qfDgFHAjw5afnPWsXbGKcBnxhhHX0mlyRyXjWVKxTbuyB0+Au4Xm+FAYa1/m7qUiERdHM8kIknYPm8OL7bsy3gD2GOMeaGRZi7ZLs3J4ortIiLh9r0LiEhHIBnYe1kzV3x21CXaZ9fPqH229tnaZ+MBfbZx8lGTrbkBk7CNjykHTgKf2B+/Bsis1W4ctiMfD2D7V52jc4QCnwL7gc1AiP3xROB1+/RIIAfb0bo5wHQHZ6i3jsAzQKp92gq8D+QC3wI9nPQ7aSrHH4Hd9u2wBbjOSTneA44DF+zvkenAw8DD9ucFWGrPmUMjR7a7KMujtbbJdmCkk3KMxnagwQ/ATvttnDu2SzOzOH27AAOBf9pz7ALmNfCedclnpz3c0D67dgbts+vm0D67fg7ts+vncGufrZdjVkoppZRS7VZbHiahlFJKKaXUFdFiWCmllFJKtVtaDCullFJKqXZLi2GllFJKKdVuaTGslFJKKaXaLS2GlVJKKaVUu6XFsFJKKaWUarf+H/TfvNKgpFlOAAAAAElFTkSuQmCC", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "df = pd.DataFrame(grid.fit(X, y).cv_results_)\n", "plt.figure(figsize=(12, 3))\n", "plt.subplot(121)\n", "plt.plot(df['param_alpha'], df['mean_test_positives'], label=\"false positives\")\n", "plt.plot(df['param_alpha'], df['mean_test_negatives'], label=\"false negatives\")\n", "plt.legend()\n", "plt.subplot(122)\n", "plt.plot(df['param_alpha'], df['mean_test_accuracy'], label=\"test accuracy\")\n", "plt.plot(df['param_alpha'], df['mean_train_accuracy'], label=\"train accuracy\")\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "It seems that we can pick a value for $\\alpha$ such that the confusion matrix is balanced. There's also a modest increase in accuracy for this balancing moment. \n", "\n", "It should be emphesized though that this feature is **experimental**. There have been dataset/model combinations where this effect seems to work very well while there have also been situations where this trick does not work at all. It also deserves mentioning that there might be alternative to your problem. If your dataset is suffering from a huge class imbalance then you might be better off by having a look at the [imbalanced-learn](https://imbalanced-learn.readthedocs.io/en/stable/) project." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Zero-Inflated Regressor\n", "\n", "There are regression datasets that contain an unusually high amount of zeroes as the targets. This can be the case if you want to predict a count of rare events, such as defects in manufacturing, the amount of some natural disasters or the amount of crimes in some neighborhood. Usually nothing happens, meaning the target count is zero, but sometimes we actually have to do some modelling work.\n", "\n", "The classical machine learning algorithms can have a hard time dealing with such datasets. Take linear regression, for example: the chance of outputting an actual zero is diminishing. Sure, you can get regions where you are close to zero, but modelling an output of **exacly zero** is infeasible in general. The same goes for neural networks.\n", "\n", "What we can do circumvent these problems is the following: \n", "1. Train a classifier to tell us whether the target is zero, or not.\n", "2. Train a regressor on all samples with a non-zero target.\n", "\n", "By putting these two together in an obvious way, we get the ZeroInflatedRegressor. You can use it like this:" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "ZIR (RFC+RFR) r²: 0.8992399247461955\n", "RFR r²: 0.8516520051101274\n" ] } ], "source": [ "import numpy as np\n", "from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor\n", "from sklearn.model_selection import cross_val_score\n", "from sklego.meta import ZeroInflatedRegressor\n", "\n", "np.random.seed(0)\n", "X = np.random.randn(10000, 4)\n", "y = ((X[:, 0]>0) & (X[:, 1]>0)) * np.abs(X[:, 2] * X[:, 3]**2) # many zeroes here, in about 75% of the cases.\n", "\n", "zir = ZeroInflatedRegressor(\n", " classifier=RandomForestClassifier(random_state=0),\n", " regressor=RandomForestRegressor(random_state=0)\n", ")\n", "\n", "print('ZIR (RFC+RFR) r²:', cross_val_score(zir, X, y).mean())\n", "print('RFR r²:', cross_val_score(RandomForestRegressor(random_state=0), X, y).mean())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## OutlierClassifier" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Outlier models are unsupervised so they don't have `predict_proba` or `score` methods. In case you have some labelled samples of which you know they should be outliers, it could be useful to calculate metrics as if the outlier model was a classifier. Moreover, if outlier models had a `predict_proba` method, you could use a classification model combined with an outlier detection model in a `StackingClassifier` and take advantage of probability outputs of both categories of models. To this end, the `OutlierClassifier` turns an outlier model into a classification model. A field of application is fraud. When building a model to detect fraud, you often have *some* labelled positives but you know there should be more, even in your train data. If you only use anomaly detection, you don't make use of the information in your labelled data set. If you only use a classifier, you might have insufficient data to do proper train-test splitting and perform hyperparameter optimization. Therefore one could combine the two approaches as shown below to get the best of both worlds." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In this example, we change the outlier model `IsolationForest` into a classifier using the `OutlierClassifier`. We create a random dataset with 1% outliers." ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from sklego.meta.outlier_classifier import OutlierClassifier\n", "from sklearn.ensemble import IsolationForest" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "OutlierClassifier(model=IsolationForest(contamination=0.01, n_estimators=1000,\n", " random_state=0))" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "n_normal = 10_000\n", "n_outlier = 100\n", "np.random.seed(0)\n", "X = np.hstack((np.random.normal(size=n_normal), np.random.normal(10, size=n_outlier))).reshape(-1,1)\n", "y = np.hstack((np.asarray([0]*n_normal), np.asarray([1]*n_outlier)))\n", "\n", "clf = OutlierClassifier(IsolationForest(n_estimators=1000, contamination=n_outlier/n_normal, random_state=0))\n", "clf.fit(X, y)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Anomaly detection algorithms in Scikit-Learn return values `-1` for inliers and `1` for outliers. As you can see, the `OutlierClassifier` predicts inliers as `0` and outliers as `1`:" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "inlier: [0.]\n", "outlier: [1.]\n" ] } ], "source": [ "print('inlier: ', clf.predict([[0]]))\n", "print('outlier: ', clf.predict([[10]]))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The `predict_proba` method returns probabilities for both classes (inlier, outlier):" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0.0376881, 0.9623119]])" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "clf.predict_proba([[10]])" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The `OutlierClassifier` can be combined with any classification model in the `StackingClassifier` as follows:" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "StackingClassifier(estimators=[('anomaly',\n", " OutlierClassifier(model=IsolationForest())),\n", " ('classifier', RandomForestClassifier())],\n", " passthrough=True, stack_method='predict_proba')" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from sklearn.ensemble import StackingClassifier, RandomForestClassifier\n", "\n", "estimators = [\n", " ('anomaly', OutlierClassifier(IsolationForest())),\n", " ('classifier', RandomForestClassifier())\n", " ]\n", "\n", "stacker = StackingClassifier(estimators, stack_method='predict_proba', passthrough=True)\n", "\n", "stacker.fit(X,y)" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "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": "3.7.9" } }, "nbformat": 4, "nbformat_minor": 4 }