{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "Printed copies of *Elements of Data Science* are available now, with a **full color interior**.\n", "\n", "From July 17 to July 31, [get 20% off at Lulu.com](https://www.lulu.com/shop/allen-downey/elements-of-data-science/paperback/product-9dyrwn.html)." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Regression" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "tags": [ "remove-print" ] }, "outputs": [], "source": [ "from os.path import basename, exists\n", "\n", "def download(url):\n", " filename = basename(url)\n", " if not exists(filename):\n", " from urllib.request import urlretrieve\n", "\n", " local, _ = urlretrieve(url, filename)\n", " print(\"Downloaded \" + str(local))\n", " return filename\n", "\n", "download('https://raw.githubusercontent.com/AllenDowney/ElementsOfDataScience/v1/utils.py')\n", "\n", "import utils" ] }, { "cell_type": "markdown", "metadata": { "tags": [ "remove-print" ] }, "source": [ "[Click here to run this notebook on Colab](https://colab.research.google.com/github/AllenDowney/ElementsOfDataScience/blob/v1/10_regression.ipynb)." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the previous chapter we used simple linear regression to quantify the relationship between two variables.\n", "In this chapter we'll get farther into regression, including multiple regression and one of my all-time favorite tools, logistic regression.\n", "These tools will allow us to explore relationships among sets of variables.\n", "As an example, we will use data from the General Social Survey (GSS) to explore the relationship between education, sex, age, and income.\n", "\n", "The GSS dataset contains hundreds of columns.\n", "We'll work with an extract that contains just the columns we need, as we did in Chapter 8.\n", "Instructions for downloading the extract are in the notebook for this chapter." ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [], "source": [ "download('https://github.com/AllenDowney/ElementsOfDataScience/' +\n", " 'raw/v1/data/gss_extract_2022.hdf');" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can read the `DataFrame` like this and display the first few rows." ] }, { "cell_type": "code", "execution_count": 3, "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", " \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", "
yearidageeducdegreesexgunlawgrassrealinc
01972123.016.03.02.01.0NaN18951.0
11972270.010.00.01.01.0NaN24366.0
21972348.012.01.02.01.0NaN24366.0
31972427.017.03.02.01.0NaN30458.0
41972561.012.01.02.01.0NaN50763.0
\n", "
" ], "text/plain": [ " year id age educ degree sex gunlaw grass realinc\n", "0 1972 1 23.0 16.0 3.0 2.0 1.0 NaN 18951.0\n", "1 1972 2 70.0 10.0 0.0 1.0 1.0 NaN 24366.0\n", "2 1972 3 48.0 12.0 1.0 2.0 1.0 NaN 24366.0\n", "3 1972 4 27.0 17.0 3.0 2.0 1.0 NaN 30458.0\n", "4 1972 5 61.0 12.0 1.0 2.0 1.0 NaN 50763.0" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import pandas as pd\n", "\n", "gss = pd.read_hdf('gss_extract_2022.hdf', 'gss')\n", "gss.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We'll start with a simple regression, estimating the parameters of real income as a function of years of education.\n", "First we'll select the subset of the data where both variables are valid." ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "data = gss.dropna(subset=['realinc', 'educ'])\n", "xs = data['educ']\n", "ys = data['realinc']" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now we can use `linregress` to fit a line to the data." ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'slope': 3631.0761003894995,\n", " 'intercept': -15007.453640508655,\n", " 'rvalue': 0.37169252259280877,\n", " 'pvalue': 0.0,\n", " 'stderr': 35.625290800764,\n", " 'intercept_stderr': 480.07467595184363}" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from scipy.stats import linregress\n", "res = linregress(xs, ys)\n", "res._asdict()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The estimated slope is about 3450, which means that each additional year of education is associated with an additional $3450 of income." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Regression with StatsModels\n", "\n", "SciPy doesn't do multiple regression, so we'll to switch to a new library, StatsModels. Here's the import statement." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "import statsmodels.formula.api as smf" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To fit a regression model, we'll use `ols`, which stands for \"ordinary least squares\", another name for regression." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [], "source": [ "results = smf.ols('realinc ~ educ', data=data).fit()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The first argument is a **formula string** that specifies that we want to regress income as a function of education.\n", "The second argument is the `DataFrame` containing the subset of valid data.\n", "The names in the formula string correspond to columns in the `DataFrame`.\n", "\n", "The result from `ols` is an object that represents the model -- it provides a function called `fit` that does the actual computation." ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "statsmodels.regression.linear_model.RegressionResultsWrapper" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "type(results)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The result is a `RegressionResultsWrapper`, which contains a `Series` called `params`, which contains the estimated intercept and the slope associated with `educ`." ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept -15007.453641\n", "educ 3631.076100\n", "dtype: float64" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The results from Statsmodels are the same as the results we got from SciPy, so that's good!" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** Let's run another regression using SciPy and StatsModels, and confirm we get the same results.\n", "Compute the regression of `realinc` as a function of `age` using SciPy's `linregress` and then using StatsModels' `ols`.\n", "Confirm that the intercept and slope are the same.\n", "Remember to use `dropna` to select the rows with valid data in both columns." ] }, { "cell_type": "code", "execution_count": 10, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "{'slope': -35.777496403061036,\n", " 'intercept': 34148.79417109003,\n", " 'rvalue': -0.02000684524772686,\n", " 'pvalue': 3.728019377398043e-07,\n", " 'stderr': 7.038831749655493,\n", " 'intercept_stderr': 346.12532772697745}" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "data = gss.dropna(subset=['realinc', 'age'])\n", "xs = data['age']\n", "ys = data['realinc']\n", "\n", "res = linregress(xs, ys)\n", "res._asdict()" ] }, { "cell_type": "code", "execution_count": 11, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "Intercept 34148.794171\n", "age -35.777496\n", "dtype: float64" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "results = smf.ols('realinc ~ age', data=data).fit()\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Multiple Regression\n", "\n", "In the previous section, we saw that income depends on education, and in the exercise we saw that it also depends on `age`.\n", "Now let's put them together in a single model." ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept -17999.726908\n", "educ 3665.108238\n", "age 55.071802\n", "dtype: float64" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "results = smf.ols('realinc ~ educ + age', data=gss).fit()\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In this model, `realinc` is the variable we are trying to explain or predict, which is called the **dependent variable** because it depends on the the other variables -- or at least we expect it to.\n", "The other variables, `educ` and `age`, are called **independent variables** or sometimes \"predictors\".\n", "The `+` sign indicates that we expect the contributions of the independent variables to be additive.\n", "\n", "The result contains an intercept and two slopes, which estimate the average contribution of each predictor with the other predictor held constant.\n", "\n", "* The estimated slope for `educ` is about 3665 -- so if we compare two people with the same age, and one has an additional year of education, we expect their income to be higher by $3514.\n", "\n", "* The estimated slope for `age` is about 55 -- so if we compare two people with the same education, and one is a year older, we expect their income to be higher by $55.\n", "\n", "In this model, the contribution of age is quite small, but as we'll see in the next section that might be misleading." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Grouping by Age\n", "\n", "Let's look more closely at the relationship between income and age.\n", "We'll use a Pandas method we have not seen before, called `groupby`, to divide the `DataFrame` into age groups." ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "pandas.core.groupby.generic.DataFrameGroupBy" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "grouped = gss.groupby('age')\n", "type(grouped)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The result is a `GroupBy` object that contains one group for each value of `age`.\n", "The `GroupBy` object behaves like a `DataFrame` in many ways.\n", "You can use brackets to select a column, like `realinc` in this example, and then invoke a method like `mean`." ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "mean_income_by_age = grouped['realinc'].mean()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The result is a Pandas `Series` that contains the mean income for each age group, which we can plot like this." ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import matplotlib.pyplot as plt\n", "\n", "plt.plot(mean_income_by_age, 'o', alpha=0.5)\n", "plt.xlabel('Age (years)')\n", "plt.ylabel('Income (1986 $)')\n", "plt.title('Average income, grouped by age');" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Average income increases from age 20 to age 50, then starts to fall.\n", "And that explains why the estimated slope is so small, because the relationship is non-linear.\n", "To describe a non-linear relationship, we'll create a new variable called `age2` that equals `age` squared -- so it is called a **quadratic term**. " ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [], "source": [ "gss['age2'] = gss['age']**2" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now we can run a regression with both `age` and `age2` on the right side." ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept -52599.674844\n", "educ 3464.870685\n", "age 1779.196367\n", "age2 -17.445272\n", "dtype: float64" ] }, "execution_count": 17, "metadata": {}, "output_type": "execute_result" } ], "source": [ "model = smf.ols('realinc ~ educ + age + age2', data=gss)\n", "results = model.fit()\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In this model, the slope associated with `age` is substantial, about $1779 per year.\n", "\n", "The slope associated with `age2` is about -$17.\n", "It might be unexpected that it is negative -- we'll see why in the next section.\n", "But first, here are two exercises where you can practice using `groupby` and `ols`." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** Let's explore the relationship between income and education.\n", "First, group `gss` by `educ`.\n", "From the resulting `GroupBy` object, extract `realinc` and compute the mean.\n", "Then plot mean income in each education group.\n", "What can you say about the relationship between these variables?\n", "Does it look like a linear relationship?" ] }, { "cell_type": "code", "execution_count": 18, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Solution\n", "\n", "grouped = gss.groupby('educ')\n", "mean_income_by_educ = grouped['realinc'].mean()\n", "\n", "plt.plot(mean_income_by_educ, 'o', alpha=0.5)\n", "\n", "plt.xlabel('Education (years)')\n", "plt.ylabel('Income (1986 $)')\n", "plt.title('Mean income versus education');" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** The graph in the previous exercise suggests that the relationship between income and education is non-linear. So let's try fitting a non-linear model.\n", "\n", "- Add a column named `educ2` to the `gss` DataFrame -- it should contain the values from `educ` squared. \n", "\n", "- Run a regression that uses `educ`, `educ2`, `age`, and `age2` to predict `realinc`.\n" ] }, { "cell_type": "code", "execution_count": 19, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "Intercept -26336.766346\n", "educ -706.074107\n", "educ2 165.962552\n", "age 1728.454811\n", "age2 -17.207513\n", "dtype: float64" ] }, "execution_count": 19, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "gss['educ2'] = gss['educ']**2\n", "\n", "results = smf.ols('realinc ~ educ + educ2 + age + age2', \n", " data=gss).fit()\n", "\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Visualizing regression results\n", "\n", "In the previous section we ran a multiple regression model to characterize the relationships between income, age, and education.\n", "Because the model includes quadratic terms, the parameters are hard to interpret.\n", "For example, you might notice that the parameter for `educ` is negative, and that might be a surprise, because it suggests that higher education is associated with lower income.\n", "But the parameter for `educ2` is positive, and that makes a big difference.\n", "In this section we'll see a way to interpret the model visually and validate it against data.\n", "\n", "Here's the model from the previous exercise." ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept -26336.766346\n", "educ -706.074107\n", "educ2 165.962552\n", "age 1728.454811\n", "age2 -17.207513\n", "dtype: float64" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "gss['educ2'] = gss['educ']**2\n", "\n", "model = smf.ols('realinc ~ educ + educ2 + age + age2', data=gss)\n", "results = model.fit()\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The `results` object provides a method called `predict` that uses the estimated parameters to generate predictions.\n", "It takes a `DataFrame` as a parameter and returns a `Series` with a prediction for each row in the `DataFrame`.\n", "To use it, we'll create a new `DataFrame` with `age` running from 18 to 89, and `age2` set to `age` squared." ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "\n", "df = pd.DataFrame()\n", "df['age'] = np.linspace(18, 89)\n", "df['age2'] = df['age']**2" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Next, we'll pick a level for `educ`, like 12 years, which is the most common value.\n", "When you assign a single value to a column in a `DataFrame`, Pandas makes a copy for each row." ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [], "source": [ "df['educ'] = 12\n", "df['educ2'] = df['educ']**2" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Then we can use `results` to predict the average income for each age group, holding education constant." ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "pred12 = results.predict(df)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The result from `predict` is a `Series` with one prediction for each row.\n", "So we can plot it with age on the x-axis and the predicted income for each age group on the y-axis.\n", "And we'll plot the data for comparison." ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(mean_income_by_age, 'o', alpha=0.5)\n", "plt.plot(df['age'], pred12, label='High school', color='C4')\n", "\n", "plt.xlabel('Age (years)')\n", "plt.ylabel('Income (1986 $)')\n", "plt.title('Income versus age, grouped by education level')\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The dots show the average income in each age group.\n", "The line shows the predictions generated by the model, holding education constant.\n", "This plot shows the shape of the model, a downward-facing parabola.\n", "\n", "We can do the same thing with other levels of education, like 14 years, which is the nominal time to earn an Associate's degree, and 16 years, which is the nominal time to earn a Bachelor's degree." ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [], "source": [ "df['educ'] = 16\n", "df['educ2'] = df['educ']**2\n", "pred16 = results.predict(df)\n", "\n", "df['educ'] = 14\n", "df['educ2'] = df['educ']**2\n", "pred14 = results.predict(df)" ] }, { "cell_type": "code", "execution_count": 26, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(mean_income_by_age, 'o', alpha=0.5)\n", "plt.plot(df['age'], pred16, ':', label='Bachelor')\n", "plt.plot(df['age'], pred14, '--', label='Associate')\n", "plt.plot(df['age'], pred12, label='High school', color='C4')\n", "\n", "plt.xlabel('Age (years)')\n", "plt.ylabel('Income (1986 $)')\n", "plt.title('Income versus age, grouped by education level')\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The lines show expected income as a function of age for three levels of education.\n", "This visualization helps validate the model, since we can compare the predictions with the data.\n", "And it helps us interpret the model since we can see the separate contributions of age and education.\n", "\n", "Sometimes we can understand a model by looking at its parameters, but often it is better to look at its predictions.\n", "In the exercises, you'll have a chance to run a multiple regression, generate predictions, and visualize the results." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** At this point, we have a model that predicts income using age and education, and we've plotted predictions for different age groups, holding education constant.\n", "Now let's see what it predicts for different levels of education, holding age constant.\n", "\n", "- Create an empty `DataFrame` named `df`.\n", "\n", "- Using `np.linspace()`, add a column named `educ` to `df` with a range of values from `0` to `20`.\n", "\n", "- Add a column named `educ2` with the values from `educ` squared.\n", "\n", "- Add a column named `age` with the constant value `30`.\n", "\n", "- Add a column named `age2` with the values from `age` squared.\n", "\n", "- Use the `results` object and `df` to generate expected income as a function of education." ] }, { "cell_type": "code", "execution_count": 27, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "0 10030.116309\n", "1 9769.571701\n", "2 9564.324903\n", "3 9414.375915\n", "4 9319.724737\n", "dtype: float64" ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "df = pd.DataFrame()\n", "df['educ'] = np.linspace(0, 20)\n", "df['age'] = 30\n", "df['educ2'] = df['educ']**2\n", "df['age2'] = df['age']**2\n", "\n", "pred = results.predict(df)\n", "pred.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** Now let's visualize the results from the previous exercise.\n", "\n", "- Group the GSS data by `educ` and compute the mean income in each education group.\n", "\n", "- Plot mean income for each education group as a scatter plot.\n", "\n", "- Plot the predictions from the previous exercise.\n", "\n", "How do the predictions compare with the data?" ] }, { "cell_type": "code", "execution_count": 28, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Solution\n", "\n", "grouped = gss.groupby('educ')\n", "mean_income_by_educ = grouped['realinc'].mean()\n", "plt.plot(mean_income_by_educ, 'o', alpha=0.5)\n", "\n", "pred = results.predict(df)\n", "plt.plot(df['educ'], pred, label='Age 30')\n", "\n", "plt.xlabel('Education (years)')\n", "plt.ylabel('Income (1986 $)')\n", "plt.title('Income versus education level, grouped by age')\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Categorical Variables\n", "\n", "Most of the variables we have used so far -- like income, age, and education -- are numerical.\n", "But variables like sex and race are **categorical** -- that is, each respondent belongs to one of a specified set of categories.\n", "If there are only two categories, the variable is **binary**.\n", "\n", "With StatsModels, it is easy to include a categorical variable as part of a regression model.\n", "Here's an example:" ] }, { "cell_type": "code", "execution_count": 29, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept -24635.767539\n", "C(sex)[T.2.0] -4891.439306\n", "educ -496.623120\n", "educ2 156.898221\n", "age 1720.274097\n", "age2 -17.097853\n", "dtype: float64" ] }, "execution_count": 29, "metadata": {}, "output_type": "execute_result" } ], "source": [ "formula = 'realinc ~ educ + educ2 + age + age2 + C(sex)'\n", "results = smf.ols(formula, data=gss).fit()\n", "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the formula string, the letter `C` indicates that `sex` is a categorical variable.\n", "The regression treats the value `sex=1`, which is male, as the reference group, and reports the difference associated with the value `sex=2`, which is female.\n", "So the results indicate that income for women is about $4156 less than for men, after controlling for age and education.\n", "However, note that `realinc` represents household income.\n", "If the respondent is married, it includes both their own income and their spouse's.\n", "So we cannot interpret this result as an estimate of a gender gap in income." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Logistic Regression\n", "\n", "In the previous section, we added a categorical variable on the right side of a regression formula -- that is, we used it as a predictive variable.\n", "\n", "But what if the categorical variable is on the left side of the regression formula -- that is, it's the value we are trying to predict?\n", "In that case, we can use **logistic regression**.\n", "\n", "As an example, one of the GSS questions asks \"Would you favor or oppose a law which would require a person to obtain a police permit before he or she could buy a gun?\"\n", "The responses are in a column called `gunlaw` -- here are the values. " ] }, { "cell_type": "code", "execution_count": 30, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "gunlaw\n", "1.0 36367\n", "2.0 11940\n", "Name: count, dtype: int64" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "gss['gunlaw'].value_counts()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "`1` means yes and `2` means no, so most respondents are in favor.\n", "\n", "Before we can use this variable in a logistic regression, we have to recode it so `1` means \"yes\" and `0` means \"no\".\n", "We can do that by replacing `2` with `0`." ] }, { "cell_type": "code", "execution_count": 31, "metadata": {}, "outputs": [], "source": [ "gss['gunlaw'] = gss['gunlaw'].replace([2], [0])" ] }, { "cell_type": "markdown", "metadata": { "tags": [ "remove-cell" ] }, "source": [ "We can use `value_counts` to check the results." ] }, { "cell_type": "code", "execution_count": 32, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "gunlaw\n", "1.0 36367\n", "0.0 11940\n", "Name: count, dtype: int64" ] }, "execution_count": 32, "metadata": {}, "output_type": "execute_result" } ], "source": [ "gss['gunlaw'].value_counts()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To run logistic regression, we'll use `logit`, which is named for the logit function, which is related to logistic regression." ] }, { "cell_type": "code", "execution_count": 33, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Optimization terminated successfully.\n", " Current function value: 0.544026\n", " Iterations 5\n" ] } ], "source": [ "formula = 'gunlaw ~ age + age2 + educ + educ2 + C(sex)'\n", "results = smf.logit(formula, data=gss).fit()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Estimating the parameters for the logistic model is an iterative process, so the output contains information about the number of iterations.\n", "Other than that, everything is the same as what we have seen before.\n", "Here are the estimated parameters." ] }, { "cell_type": "code", "execution_count": 34, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Intercept 1.483746\n", "C(sex)[T.2.0] 0.740717\n", "age -0.021274\n", "age2 0.000216\n", "educ -0.098093\n", "educ2 0.005557\n", "dtype: float64" ] }, "execution_count": 34, "metadata": {}, "output_type": "execute_result" } ], "source": [ "results.params" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The parameters are in the form of **log odds** -- I won't explain them in detail here, except to say that positive values make the outcome more likely and negative values make the outcome less likely.\n", "For example, the parameter associated with `sex=2` is `0.74`, which indicates that women are more likely to support this form of gun control.\n", "\n", "To see how much more likely, we can generate predictions, as we did with linear regression.\n", "As an example, we'll generate predictions for different ages and sexes, with education held constant.\n", "First we need a `DataFrame` with a range of values for `age` and a fixed value of `educ`." ] }, { "cell_type": "code", "execution_count": 35, "metadata": {}, "outputs": [], "source": [ "df = pd.DataFrame()\n", "df['age'] = np.linspace(18, 89)\n", "df['educ'] = 12" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Then we can compute `age2` and `educ2`." ] }, { "cell_type": "code", "execution_count": 36, "metadata": {}, "outputs": [], "source": [ "df['age2'] = df['age']**2\n", "df['educ2'] = df['educ']**2" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can generate predictions for men like this." ] }, { "cell_type": "code", "execution_count": 37, "metadata": {}, "outputs": [], "source": [ "df['sex'] = 1\n", "pred_male = results.predict(df)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "And for women like this." ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [], "source": [ "df['sex'] = 2\n", "pred_female = results.predict(df)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Now, to visualize the results, we'll start by plotting the data.\n", "As we've done before, we'll divide the respondents into age groups and compute the mean in each group.\n", "The mean of a binary variable is the fraction of people in favor.\n", "Then we can plot the predictions." ] }, { "cell_type": "code", "execution_count": 39, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "grouped = gss.groupby('age')\n", "favor_by_age = grouped['gunlaw'].mean()\n", "\n", "plt.plot(favor_by_age, 'o', alpha=0.5)\n", "plt.plot(df['age'], pred_female, label='Female')\n", "plt.plot(df['age'], pred_male, '--', label='Male')\n", "\n", "plt.xlabel('Age')\n", "plt.ylabel('Probability of favoring gun law')\n", "plt.title('Support for gun law versus age, grouped by sex')\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "According to the model, people near age 50 are least likely to support gun control (at least as this question was posed).\n", "And women are more likely to support it than men, by about 15 percentage points.\n", "\n", "Logistic regression is a powerful tool for exploring relationships between a binary variable and the factors that predict it.\n", "In the exercises, you'll explore the factors that predict support for legalizing marijuana." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**Exercise:** In the GSS dataset, the variable `grass` records responses to the question \"Do you think the use of marijuana should be made legal or not?\"\n", "Let's use logistic regression to explore relationships between this variable and age, sex, and education level.\n", "\n", "\n", "1. First, use `replace` to recode the `grass` column so that `1` means yes and `0` means no. Use `value_counts` to check.\n", "\n", "2. Next, use the StatsModels function `logit` to predict `grass` using the variables `age`, `age2`, `educ`, and `educ2`, along with `sex` as a categorical variable. Display the parameters. Are men or women more likely to support legalization?\n", "\n", "3. To generate predictions, start with an empty DataFrame. Add a column called `age` that contains a sequence of values from 18 to 89. Add a column called `educ` and set it to 12 years. Then compute a column, `age2`, which is the square of `age`, and a column, `educ2`, which is the square of `educ`. \n", "\n", "4. Use `predict` to generate predictions for men (`sex=1`) and women (`sex=2`).\n", "\n", "5. Generate a plot that shows (a) the average level of support for legalizing marijuana in each age group, (b) the level of support the model predicts for men as a function of age, and (c) the level of support predicted for women as a function of age. " ] }, { "cell_type": "code", "execution_count": 40, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "text/plain": [ "grass\n", "0.0 25997\n", "1.0 12672\n", "Name: count, dtype: int64" ] }, "execution_count": 40, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "gss['grass'] = gss['grass'].replace(2, 0)\n", "gss['grass'].value_counts()" ] }, { "cell_type": "code", "execution_count": 41, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Optimization terminated successfully.\n", " Current function value: 0.605895\n", " Iterations 6\n" ] }, { "data": { "text/plain": [ "Intercept -1.971358\n", "C(sex)[T.2.0] -0.373592\n", "age -0.015406\n", "age2 0.000004\n", "educ 0.221171\n", "educ2 -0.004209\n", "dtype: float64" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Solution\n", "\n", "results = smf.logit('grass ~ age + age2 + educ + educ2 + C(sex)', \n", " data=gss).fit()\n", "results.params" ] }, { "cell_type": "code", "execution_count": 42, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [], "source": [ "# Solution\n", "\n", "df = pd.DataFrame()\n", "df['age'] = np.linspace(18, 89)\n", "df['age2'] = df['age']**2\n", "\n", "df['educ'] = 12\n", "df['educ2'] = df['educ']**2" ] }, { "cell_type": "code", "execution_count": 43, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [], "source": [ "# Solution\n", "\n", "df['sex'] = 1\n", "pred_male = results.predict(df)\n", "\n", "df['sex'] = 2\n", "pred_female = results.predict(df)" ] }, { "cell_type": "code", "execution_count": 44, "metadata": { "tags": [ "remove-cell" ] }, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZgAAAEICAYAAABiXeIWAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8fJSN1AAAACXBIWXMAAAuJAAALiQE3ycutAABnVElEQVR4nO3deViU5frA8e8wDPsqIMiuiCgooqLmCu77glul5ZLaqUOn9dc57XXK6tTptJyixaysXE6pqWiWqYj7mvsCrgiIAm4g2zDMPL8/JiaRbYQZGPD5XNdcycu873vPQHPzbPejEEIIJEmSJMnErBo7AEmSJKl5kglGkiRJMguZYCRJkiSzkAlGkiRJMguZYCRJkiSzkAlGkiRJMguZYCRJkiSzkAlGMpg5cyZPPvkkAOnp6Tg5OZGXl2fy+7z11lvcf//9Jr+udHdatWoVwcHB1X4/ODiYVatWNVg80p9kgrFAsbGxfPjhh40aQ2BgIAUFBbi6utbrOgsXLiQqKqrCsRdeeIGlS5fW67qSJFk+mWAkqRoajaaxQ7A4Qgi0Wm1jhyE1ETLBNDFnz55lzJgxeHl5ERQUxLx589DpdIbvf/zxxwQEBODh4cFLL71EVFQUCxcuBPTdXkOGDMHLywt3d3dGjRpFWlpalfdJS0tDoVBw48YNcnJycHJyqvBQKBQkJycD8MADD+Dr64uLiwvdunVj8+bNABw8eJBHHnmEo0ePGs5LT0/ntddeY/z48YZ7nTlzhmHDhtGiRQtCQkIqtN7KW0BvvPEGLVu2xNvbu9rWXU5ODjY2Nly4cMFwTK1W4+7uzq5du2p9/8rv9eqrr+Lj48N9993HtWvXiIuLw93dHTc3N7p162a4/u1dL7d31bz//vsEBgbi7OxMcHAwCxYsqDLu2n4uarWaRx55hBYtWtC6dWu++uorFAqF4TlCCP773//Svn173NzciI2N5eTJk1XeqyqZmZkMGTLE8PN76623KryO4OBg3n77be655x4cHBw4ceJEjT+z23++AG5ubobfl9dee43Ro0cze/ZsXFxcCA0NZeXKlYbn1vZ6MjMzGTp0qCHeEydO1Poajx8/TteuXXFxcWHYsGFkZWUB8NRTTzFz5swKz/3Xv/7FiBEjqrzOhg0biIyMxNnZGW9vbx599FHD96r73SorK6NXr1689NJLhufGx8czaNCgCv/vNktCsjgxMTHigw8+qHS8sLBQBAUFiQ8++ECo1Wpx4cIFERERIRYsWCCEEGLjxo3Czc1N7NmzR6jVavHyyy8La2tr8c033wghhDh//rxYt26dKC4uFnl5eWLSpEli8ODBhuvPmDFDPPHEE4bnAuL69euV4njjjTdERESEyMvLE0II8fXXX4sbN26I0tJS8e6774oWLVqI/Px8IYQQ33zzjejcuXOF81999VUxbtw4IYQQGo1GhIWFiWeffVYUFxeLw4cPi1atWonFixcbzre2thbvvfeeKC0tFZs3bxbW1tbizJkzVb53o0aNEvPmzTN8vWzZMhEaGmrU+/fNN98IpVIpXn/9daFWq0VhYaF4/vnnxejRo0VhYaEoKysTBw8eFFevXhVCCBEUFCRWrlxpuNfKlStFUFCQEEKI1NRUYW9vL06ePCmEEOLy5cvi8OHDVcZc28/lpZdeEt26dRNZWVnixo0bYtSoUQIQ58+fF0IIkZCQICIjI8WpU6eERqMRH330kQgJCRFqtbrK+92uX79+Yvbs2aKoqEikpqaKNm3aGF5H+ets166dSElJEWVlZUKtVtf4M7v151vO1dVVbN682fB9pVIpPv/8c6HRaERiYqKwtbU1/Exrez39+vUT06dPF4WFheLkyZMiODi4Qry3CwoKEsHBweLkyZOisLBQTJ8+XQwYMEAIIcTRo0eFk5OTuHnzpuH5YWFh4scff6zyWq1atRLfffedEEKIgoICsWPHDiFE7b9b586dE+7u7mLz5s1i1apVwsvLS2RlZRnx02naZIKxQNUlmB9//FFERUVVODZ//nwxcOBAIYQQDz30kIiPjzd8r7S0VLi6uhoSzO0OHjwobG1thVarFUIYl2B++OEH4ePjY/hwq4qbm5vYvn27EKL2BLN9+3bh4uJS4cPwzTffFEOGDDGc7+PjU+H8tm3biuXLl1d57x9++EGEhYUZvh47dqx4/fXXhRC1v3/ffPONaNGiheH9EEKIV155RfTq1UscOnSo0r1qSjBnzpwRdnZ2Yvny5aKoqKjKWKtz+8+lTZs2YtmyZYbv7927t0KCCQ8PF6tWrapwDV9fX7F169Za75Weni4AkZubazj27rvvVkowt/4+1vYzMybBdOjQocL3hw8fLt54441aX095vNnZ2Ybv/etf/6o1wbzzzjuGry9fviwAkZGRIYQQokePHob/R3bu3ClatGghSkpKqrxWYGCgeOWVV0ROTk6F47X9bgkhxJIlS4Sfn5/w8PAQa9asqTbe5kR2kTUhaWlpHDt2DDc3N8PjmWee4fLlywBkZWUREBBgeL5KpaJVq1aGr3Nzc5k6dSoBAQG4uLjQv39/1Go1N2/eNOr+u3fv5pFHHmHlypWGLhSdTseLL75IaGgoLi4uuLm5kZeXx5UrV4y6ZmZmJr6+vtjY2BiOtWnThszMTMPX3t7eFc5xdHSsNuaxY8dy+fJl9u7dy5UrV/j111958MEHgdrfPwA/Pz+srP783+LZZ5+lX79+TJkyBR8fH5544gmKi4trfV0hISF8++23fPLJJ3h7ezN06FAOHTpU5XNr+7nc/nMNDAyscH5aWhoPPPBAhdd1/fr1Cu9hdbKysrCzs8PT07Pa699+zJifWW2CgoIqfX3x4sVaX095vC1btqz2WrXdz9vbG1tbW8P9HnroIUM38sKFC5k2bRq2trZVXmflypUcO3aMsLAwunTpwo8//miIubbfrbi4OLRaLb6+vowePdqId6npkwmmCQkICKBbt27cuHHD8MjPz+f48eMA+Pr6kpGRYXh+WVkZly5dMnz9/PPPU1RUxIEDB8jPz2fr1q2Avs+7NmlpaYwfP57PP/+ce+65x3B8yZIlLFmyhJ9//pm8vDxu3LiBq6ur4Zq3flhXxd/fn6ysrAoD6mlpafj7+xvxjlRmZ2fH5MmT+f777/nf//5Hz549DcmwtvevqnidnJx45513SE1NZdeuXWzatIlPP/3U8L2ioiLDc299rwGmTJnC5s2byc7OpnPnzoZEd7vafi63/1zT09MrnB8QEMCyZcsqvK6ioiKjpoL7+vpSUlJS4Q+C269/+/tS28/s9velsLCQ/Pz8Cte7dZys/J5+fn61vp7yeHNycmqM93a33i8nJwe1Wm243/3338/+/fs5ceIEP/zwA7Nmzar2Ol27dmXFihVcuXKFl19+malTp5KdnW3U79b//d//ER4ejlqt5uOPP6415uZAJhgLVVZWRklJieGhVqsZPXo02dnZfPrpp5SUlKDVaklNTTUMnt5///0sWbKE/fv3o9FomDdvHoWFhYZr5ufn4+DggJubG1evXuWf//ynUbHk5+czevRo/va3vzFlypRK37OxscHT05PS0lJef/31Cq0Lb29vLl26VO1f/T169MDb25tXXnkFtVrNsWPH+Pjjj5kxY8YdvmN/mj59Ov/73//45ptvmD59uuF4be9fVdauXcupU6fQ6XS4uLigUqmwtrYG9B82S5cupaSkhHPnzpGQkGA4LzU1lQ0bNlBcXIyNjQ1OTk6G825X28/l/vvv59133+Xy5cvk5eXxxhtvVPh+fHw8r7zyCqmpqYbrrV692vBzeO2114iNja3y3gEBAfTp04cXXniB4uJiTp8+zfz586t9P6D2n1nXrl3ZtWsXKSkplJSU8MILL6BQKCpc49SpU3z55ZeUlZXx888/k5SUxL333lvr6ymP97nnnqO4uJjU1FS++OKLGuMF+OKLL0hNTaW4uJh//OMf9O/f35AQXVxcmDhxIlOnTqV169Z06dKlymuUlpby/fffc/36daysrHBzcwPA2tq61t+txMREfvzxRxYtWsTSpUt56aWXOHLkSK1xN3mN3EUnVSEmJkYAFR639u1PmDBBeHt7C1dXV9G1a1exdOlSw7kffvih8PPzEy1atBAvvviiCA8PF0uWLBFCCHHixAnRvXt34ejoKMLCwsQXX3xRYZylujGYzZs3C0A4OjpWeGzdulUUFBSIuLg44eTkJHx9fQ399+VjE6WlpWLs2LHC3d1duLq6igsXLlTqo09NTRVDhgwRbm5uonXr1uK9994TOp1OCFH1GE7nzp2rHVcSQgidTidat24t7OzsxI0bNyp8r6b3r6p7ffDBB6J169bCwcFBtGzZUjz66KOGsYcLFy6Ivn37CicnJ9GzZ0/xzjvvGH5OR44cET179hTOzs7C1dVV9O/fv8pxHGN+LsXFxWLOnDnC3d1dBAcHi88++0wAhkFinU4nEhISRHh4uHB2dha+vr5iypQphokWs2bNEi+88EK179eFCxfEoEGDhLOzs+jatat47bXXRLt27Qzfv32sSYiaf2ZCCPHMM88Id3d34efnJ7788stKYzCjRo0SDz30kHB2dhYhISEVxphqez0XLlwQgwcPFk5OTqJr165i3rx5tY7BzJs3T3Tp0kU4OzuLIUOGGMZfym3ZskUA4qOPPqr2Omq1WgwfPly0aNFCODk5ifDwcPHDDz8Yvl/d79bFixeFp6dnhXGX999/X4SHh9/x+FxToxBC7mjZXJWWluLh4cGvv/5Knz59GjscyUR27dpFbGwsJSUllVoGVenUqRPJycl4eHgYdf23336bpKQkNmzYUN9Qq/Taa69x6NAhi1pdn56eTmhoKFlZWUa/T1LtZBdZM/PTTz9RXFxMYWEh//jHP/Dw8KB79+6NHZZUDzk5OWzevBmtVktWVhbPP/88EydONCq5ABw9erTGD80DBw6QkpKCEILff/+djz/+mMmTJ5sqfIun1Wp55513mDJlikwuJiYTTDPz/fff06pVK3x9fTlw4ACJiYkVZvtITY9Wq+Wpp57C1dWVqKgo/Pz8TDpInJuby4gRI3B0dGTixInMnTuX2bNnm+z6luz8+fO4uLiwZcsW3nzzzcYOp9mRXWSSJEmSWcgWjCRJkmQWVc+bbMJcXFzqvIZCkiRJujOZmZmV1jmVa3YJxt/f36jid5IkSVL9hYeHV/s92UUmSZIkmYVMMJIkSZJZNLsuMkmSpLqQE2prZuy6q1vJBCNJ0l1No9GQkZGBWq1u7FAsmq2tLQEBAahUKqPPMWuCSU5OJj4+HrVaTWxsLF988QVKpbLK544aNYrU1FTOnDkD6MtmP/PMM4Yy5X369KlQTLCp02h1FKm1ONgqUSllT6UkNZaMjAzDrqN1+Sv9biCE4OrVq2RkZNCmTRujzzNbgtHpdMyZM4fExETCw8OZMmUKixYtqrJK7uLFi2nRokWl43FxcdVuM9tU6XSCradz2Xgyh+LSMuxtrBncoSX9Q72wspK/3JLUkIQQqNVqgoODa91a4m6mUCjw8PDgypUrCCGMTsRme0f37duHr6+vYQrb7NmzWbFiRaXnXblyhYSEBF588UVzhWK8dc/C5regINdst9h6OpeVBy/iYKPE390BBxslKw9eZOtp891TkqSayZZL7eryHpktwWRmZlbahe/WTZPKPfnkk8ybNw87O7tK31uzZg2dO3dm6NCh7Nu3r8r7JCQkEB4ebnhcv369bgFryyD7BGx5Bz6IgNWPQU5K3a5VDY1Wx8aTOXi72GGv0ncV2quUeLvYselkDhqtzqT3kyRJakxmSzDGzMj45ZdfUCqVDBw4sNL3Ro8eTVpaGocPH+bFF18kLi6OgoKCSs+Lj4/nxIkThoe7u3vdAlZaw6yfYc4maD8SDi2GT3vCoklwLhlMMMOkSK3Vd4upKo5D2auUFJWWUaTW1vsekiQ1TQqFgnHjxlU41q1bN8OOrDWJjY1l+/btZoqs7syWYAICAipt83p7CZetW7eyadMmgoOD6du3LxcuXCAyMhIAT09P7O3tAYiJicHf39+ww51Z+UfD5IXw+CG456+Qvgu+Gwef94NDS6GstNZLaLQ68oo0lVokDrZK7G2sKdZUTCTFGi0ONtY42FY9AUKSpOZPqVSSkZFh6IVJSUlBp2vavRpmSzDR0dFkZmYayrZ89dVXTJgwocJz3n77bTIzM0lLS2P79u0EBQUZthHNysoyPO/48eOkpaXd0eyFenMPguFvw1PHYcjrUHwNVj0CH3aCbf+BomuVTtHpBMmpOfxzzQleX3ucf645QXJqDjqdvvWjUloxuENLsvNLDEmmWKMlO7+EQR1aytlkknSXmzRpEsuXLwf0k5+mTp1q+F5xcTFDhgyhW7duRERE8O9//7vKa1y4cIHRo0cTHR1NdHQ0W7ZsaZDYq2L0LDKtVkt2djZlZWWGY4GBgdU+X6lUsmDBAiZNmoRarSYmJoYHH3yQxMREEhMTa50d9sknn7B69WpUKhUqlYqFCxfWvfurPuzdoM8T+tbMidWw82PY9DpsfQ+ipuqPe4QAfw7ge7vY4eFoQ7FGy8qDFwGIDWsJQP9QLwA2nczhaoEaBxtr4rr4GY7XRE5tliQzWx1v8rFXAFq2h3G1L7O4//77mT17NnPnziUxMZFly5YZlmfY2NiwbNky3NzcKC0tpU+fPowZM4b27dtXuMZDDz3Ef//7XyIiIkhPTyc2NpazZ882ykQGoxLMp59+yosvvkiLFi0MU/kUCgWnTp2q8byBAwdWKjw5duxYxo4dW+m5wcHBhjUwAG+99RZvvfWWMeE1DKUKOk1C0yEO9dntOPz+GVb7voJ9X0HYSMp6PsrGE25VDuBvOJ5NRz9XXO1VqJRWxIa1pE9bT6OThZzaLEl3h9atW6PRaFi2bBkdO3bEwcHB8D0hBK+//jqbNm1CCEFmZibHjh2rkGAKCgrYvn0706ZNMxwrLS0lJycHb2/vBn0tYGSC+c9//sOJEydo1aqVueOxWBU/5F2xd3iJMaOeIvrS/7A6shTr1J952C6MY4EPctpjEDora4QQZOeVcPRiHjcTj+PiYGNIDCqlFa4OxrVCjGkZSZJkAka0Mszt3nvv5ZFHHuG7776rcHzx4sWcPXuWvXv3Ymtry8SJEykpKanwHJ1Oh4ODA4cOHWrAiKtn1Cecr6/vXZ1coPL6FXuVFR8c0DH32v280uYH1vv8BRdNLiNPvcRDv4+jW+b35ORkczwrHztrK4I8Heu05kVObZaku8v999/P3//+d4YOHVrheF5eHp6entja2nL+/Hk2bNhQ6VwXFxciIiL4+uuvDccOHDhg9pirY1QLJioqirFjxxIXF4etra3h+K0DUM1ZVR/yOflqrhSUYlOsICLch9/dZvEdo+ldsoWJpYn0v/BfujOfX1WDOBowFSuFT4XE0Ketp1HjKOVTmz0cbSoct1cpuVqgpkitNbolJEmS5fPw8OAf//hHpeMPPvggy5cvJyIiguDgYGJiYqo8f/HixTz22GN89NFHaDQa7rnnngoJpyEZlWAKCgrw8PBg69athmMKheKuSTC3f8hrdYJzVwpxsbemuFSLpkyHvUpJeIAnv+cMJ8t1HN7X9tH90lLiND8z/tw6zt6I4YDvVC66RN1RYrh1avOt62fk1GZJal5unUBVzt/fn7S0NADc3d0rfAbfKjk52fDvoKAg1qxZY44Q75hRCeabb74xdxwW7fYPeY1WR5lWh0qpH6BXWesThYONNe6OKp4aEgaE8f6GHhwWF+mV+yPhOWtpey2ZSw7t2e45BQfrDkbdu3xqc/kYjL1KaZjaHNfFT84mkyTJYhn16VRYWMjzzz/PPffcwz333MOLL75IYWGhuWOzGLevX1EprVAoIL9YQ2svR6z+mP5X3qpwdVDh6WzL0AhvTmq8+TnwWRZEr2Wz/6PYl15hcvrrqD6Ogm3vV7me5nb9Q72I6+JHcamWzOtFFJdqjZ7aLEmS1FiMasH85S9/oUWLFnzxxRcoFAq++eYb/vKXv7Bo0SJzx2cxKqxfKVXj42JPkaaMls76MamqWhUVz7Ehy/MBrHo/Rr/S7Vjt+RQ2/RO2/hs6369fT+PZtsI9b133cqdTmyVJkhqbQhhRNKxz584cPny41mOWIDw8vNLaG1Mq/9C3U1mx69xVNp3Moai0DAcbawZVszalygWSQsCFnbD7U0j5GRAQOgx6xaML6sfWM1fkuhdJMjMhBCkpKbRv315WVK5Fde9VTZ+5RrVgFAoFFy9exM/PD9CXcblbfxi3rl8xtlVR5ZoXhQKC++gfV8/Cni/g4CI4vZ4it/Zk2Y3D2X8kHo4Oct2LJElNklEJ5tVXX6V79+7069cPIQQ7d+7kk08+MXdsTcKdLJislkcIjHwXBryA9vfvKNvyCVNvvEPh1S854jOJIz4TwMXtjqY3S5IkNTajEkxcXBy9evVi7969KBQKPv7440YpO9Ds2btR0PVR3rzYmxjdXrpmLaVXxny6Z35DitdwfnWKo0gdJte9SFIzpFAo6Ny5s+HrDz74gAEDBpjtfsHBwWzfvr1SlXtTMrrYpY+PT5U1xCTTcrBVYmtrxxGbAZz2HIz3zeN0zVpCeM46OuasQfdjf+j1V/14jdziVZKaDaVSaTElXkylxk+oHj16AKBSqbCxsTE8yr9ubqrbx6Uh3T4lOts5gp/avM5rIUu50OEvWF0+Akvvg0+6wZ75oK68CZskSc1DdaX3Fy5cyJgxYxgxYgRt2rTh6aefJjExkd69exMSEmJYkJmRkUFMTAxdu3alU6dO1c78PXLkCAMHDqRbt2707duXo0ePmiT+Glswe/fuBUCj0ZjkZpbK0qoVV1XSf1CPLgSEDoWyV+Hw/2DP5/DLs5A0D7o+CD0e1u9hI0lSnb2y4xXO3jhr8uuGuIXwep/Xa3yOVqslKioKAGtra/bv319t6X2AQ4cOceTIEezt7QkJ0W8ZsnPnTtatW8err77K5s2b8fT05JdffsHBwYH8/Hy6devG6NGjcXNzM9xXo9Hw8MMPs2LFCvz8/Ni3bx9z5sxhz5499X7dRnWRzZw5k4ULF9Z6rKmytGrFVlaK6meo2ThC99nQbRacTYLdCbDrE/105/aj9OtpAnvpZ6n9Qe4jI0mW7/YusppK74N+m+TyPbLCwsIYNmwYoK8def78eUBffuaJJ55g3759WFlZcenSJc6cOUN0dLThmqmpqRw/fpxRo0YZjl27VvsCcGMYlWDKd5m81f79+00SQGOrrVpxY87aqnGGmpUVhA7WP3JS9C2aw/+Dk2vAJxLu+Su68Di2ns+3mJaZJFm62loZDam20vu3Fh62srIyfG1lZWWoa/b+++9jb2/PoUOHUCqVdOvWrVKJfyEEISEhZhn/qfGT8+OPPyY0NJSTJ0/Srl07w6NVq1YVMmBTVl7I8tZCkqBPMkWlZRSptY0U2R1o2R7GfAhPn4DBr0HRVVj1CGXvR1C84U28FDfwd3eo03YBkiQ1DlOU3s/Ly8PHxwelUsmePXuqXBzfvn17bt68yaZNmwB9wjl48GD9gv9DjQlm+vTpbNiwgREjRrBhwwbD4/jx482me+zWQpa3KtZosbNWotHpms6eKw4toO9T8MRhyiZ8xWUrb0ZcWUj8ofEMO/UaQerTch8ZSWpCFi9ezMqVK+ncuTPh4eF3vP7wscceY+XKlURGRvLRRx/RvXv3Ss9RqVSsWrWKefPm0blzZyIiIlixYoVJ4jeqVExdJScnEx8fj1qtJjY2li+++AKlsury8qNGjSI1NdWwbXJZWRlz585l27Zt2Nra8vnnn9OvX79a71mXUjHJqTkVqhUXlZZxNDMPB1sl7g42DdK1ZOpxkrwiDa+vPU436/N0yVpKu6sbUQotF12i2OgSx/j7/oKrk70JIpekpkuWijFeXUrFGPVJdujQIXr37o2Li0uF6co10el0zJkzh2XLlnHmzBny8/OrnSK3ePFiWrRoUeHYwoULUavVnDlzhqVLlzJ79mzMlQtvr1Z8NqcQFBDi5WT2riWdTpCcmsM/15zg9bXH+eeaEySn5qDT1e+1lrfM0uza82vYPL7qtobd/g/hXpTGjMxXcfkyGrZ/aFQ1Z0mSpLowKsE8+uijJCQkEBISwrVr13jrrbf45z//WeM5+/btw9fXl/DwcABmz55dZbPrypUrJCQk8OKLL1Y4vmLFCubMmQNAZGQkHh4eJusXvF35rK1XxoTz/IgO+LjaEenvhoONfg6EObcovn0r5voms/K1PECF9TSFtl4k+T7My21+IKXn2yjs3GHjq/B+OKx5AnJOmvJlSZIkGZdgSktL6dKlC2VlZTg5OfF///d/tfbRZWZmEhAQYPg6MDCQjIyMSs978sknmTdvHnZ2dnU6PyEhgfDwcMPj+vXrxrykKqmUVqiUVqjLtA0y6F/bDLY7SWZVtYR0QjC+c8V9ZMZ0a0O7YY/CI9tg5jr9LLQD38Gn98C3YyH1F9A1gYkNkiRZPKOmKZd3hwUFBfHjjz/i5+dHXl5ejecY0531yy+/oFQqGThwoGFb0Ds5HyA+Pp74+HjD1+UtprpqyC2Kb9+KuZy9SnlH2ypD1Wt5Vh/KIq6LH6+MCa96fKe8mvONdNj7JRz4Vl8lwD1Yv3AzahrYu5ns9UqSpRJCyDGYWtRliMKoT69XX32VvLw83nvvPT7//HOeeOIJ/vvf/9Z4TkBAQIUWR3p6eqWialu3bmXTpk0EBwfTt29fLly4QGRkpNHnm8PtpVrgz83EBnVoadI1MTXNYLuTZFZbSwjA1UFVfexugTD0DXj6JIz+AKztYf0L+u6zn5+B3NS6v0hJsmAKhQJbW1uuXr2KTqdDCCEfVTx0Oh1Xr17F1tb2jhKxUbPIbt68ibOzc4VjhYWFODo6VnuOVqslNDSUtWvXEh4ezpQpUxgxYgSzZs2q8vlpaWkMHjzYMIvsyy+/ZPPmzSxZsoQjR44wYcIETp8+XeuLM8WGY+WlY4zZTKy+bp/BduvOmMZWESifMebv7lDpe5nXi3hldASuDirjgxICzm/R1zpLXQcICBkIPf4CoUNlkU2pWdFoNGRkZKBWqxs7FItma2tLQEAAKlXFz5J6bzjm5uZGfHw8H374IVZ/fLj069evxkU/SqWSBQsWMGnSJNRqNTExMTz44IMkJiaSmJjIggULarznzJkz2b59O23btsXGxoavvvqqwZqwNZZqMbGq6o7FdfEzHDfGnXbr1TolWqGANrH6x7XzsG8BHPgezt4L7q2hx1zZfSY1GyqVijZt2phtlmpzUZfPX6NaMJGRkQwdOpTDhw+zfPlyXF1d6dKli9lmddWHubdMNpf6roMxpiVUr6Ke6gI48oN+580rqaByhM736cdqWravy0uWJKkZqPc6GGtra9577z2mTZtG3759SUlJkQNiJqavO1bDOEktbl/LU1yqrdQSqteUaFsnfZHN+D0wfTW0iYH9X8OnPeG7cZCyTs4+kySpAqO6yMobOTNnzqRdu3aMGzfOZNU2JdOorVvPZEU9b+0+u572R/fZd/C/+/WTBbrPgS4P6svWSJJ0VzPqz+XXX/+zwmjv3r3ZuHEjf/vb38wWlFR31bWEzFLU0z0Yhs7Tzz4b8xHYOMOGV/SzzxL/BpdNs2mRJElNk1lrkTWGpjoGU1fGjt1otDr+ueYEDjbKShMBiku1vDImvP4TGYSACzth7xdwci0ILQT21k8K6DAGlHcwk02SpCahzrPIevTowd69e1GpVBXGXMoXJZWWlpo2UslodzpgX76+p7qJACaZJadQ/Ll4My9TP0bz+7ewfBY4t9JvktZtJjh71/9ekiRZPNmCaaLqsn6muvU9vdp4UKLRmWc6dpkajq+EvfPh4u9gpYKI8dB9LgT0qLDzpiRJTU9Nn7kywTRB9e3uKu9Ws1NZsevc1Ybb8TLzd9j3JRxbAdpS/c6bPeZCx0lgU3mRqCRJlq/O05RVKhU2NjaoVCrDo/zr2sr1S+ZT3wH78okAu85dNWkl51r5d4O4z/WTAga9ot8qIPFv8H4HWP8iXDtnnvtKktQoakwwGo2G0tJSNBqN4VH+tRx/aTymqGFmykrOd8zRE/o9A08chnsXg28U7PoE/tsVFk+GU7+BrnF23Czf7kDu+ClJ9WfUOphy165do6SkxPC1r6+vyQOSameKAXtTVnKuM6U1dBitf+Se0q+pObQETv+mnwIdPRu6PNAga2rqVeVAkqQqGfUJsmHDBtq2bYuvry+RkZEEBATQp08fc8cm1cCYlfs1MVUlZ5Pxagcj34VnTsKo/+grOm94Wd99tioessxblsjUG79JkmRkC+Yf//gHW7duZdSoURw8eJDly5ezbds2c8cm1aC+BTkbZNpyXdg666sBRM/Wr6nZ9yUc+R8cWgR+3fSzzyLiQGVX+7WMZLIqB5IkVWDU/zUKhQJfX1/KysoAmDRpkkwwFqI+Nczq2woyq/I1NZMXwpPHIPYFyM+CVY/oWzW/vayv9GwCZqlyIEmScS0Ye3t7tFot4eHhvPvuu/j5+VUYi2kuLhVcwsvBC2urOxqaarIacluCenFpBbH/gH5P6/en2bcAdv4Xdn4MbQfrWzyhQ8Cqbt16DbmLqSTdTYz6NPnvf/9LcXExH374ISkpKfz0009899135o6tQQkhiE+KZ+RPI/nyyJdcLb7a2CE1mPpWcm4wShWEj4MZayB+L/T8C2TsgaX3wkdRsO0/UHDnYyYNuYupJN1N5ELLP2h1WhLPJrI0ZSknr53E2sqaoUFDub/9/XT26iy3JzCR+u57U0lpIRxdpm/VXD6qrxQQPlbfqgnsZXSlgIbcxVSSmpN6r+QfMmRIpQ9YV1dXevTowWOPPYa9vb1pIjWB+q7kF0Jw5MoRfkj5gV/TfkWj09C+RXvuDbuXka1H4qC6O1acG5MI7uQ5Zq8aIARk7of9X8Gxn0CrBq8O+j1sIu8FOxejLmPyBChJzVy9E8yTTz5JVlYWDzzwAAqFgsWLF+Ph4YFWq6WwsJDvv/++yvOSk5OJj49HrVYTGxvLF198gVL5Z392YWEhAwYMMCze7NOnD59++inW1tYsXLiQZ555hoCAAAD69OlDQkJCvV7snbpWco2fTv/EstRlZBVm4axyZlzbcUwOm0wb1zYmuYelMWY9SF2ec61QQ7GmjE5+rjj8Md5RW+20Oiu6BgcX6YttXj+v330zcjJEPwStOpv2XvUgk5nUHNQ7wXTv3p19+/YZvhZC0KNHD/bt21ftxXU6He3atSMxMZHw8HCmTJnCqFGjmDFjRoXnFBUV4eTkhE6nY9KkSYwdO5aZM2eycOFCtm/fzoIFC0z2YutKq9Oy7eI2/pf6P3Zc3AFAT5+e3Nv+XmIDYlFZNZ8y9MYU0bzT59gorUhKyaa0TEeEnyvBHo6AibcKqIpOB+c26xNN6joQOvCL1ieajhNA1Tgtb7moU2pO6r1lcn5+Prm5fw6eXr16lfz8fADs7Kpej7Bv3z58fX0JDw8HYPbs2axYsaLiza2scHJyAqCsrAy1Wm2RYx1KKyWxAbF8Pvhz1sWtY2bETFKup/B08tMMXz6cTw99SnZhdmOHWW/GlI+py3M0Wh1CgIu9ivO5hej++JvG7NOArayg7SC4b7F+qnPMc5B/EVb/Ff4TBr8+r68g0MDkok7pbmFUgnn55Zfp0qUL9913H/feey9du3bl5ZdfprCwkJEjR1Z5TmZmpqF7CyAwMJCMjIwqn9uzZ0+8vLxwcXFh2rRphuNr1qyhc+fODB06tEIL6lYJCQmEh4cbHtevXzfmJdVZgEsAz0Q/w8ZJG3mz75t4O3rz2eHPGLZiGE9tfopdWbvQiaZZx8qY9SB1eY5KaYX1Hy0UjVaHpkz//jToNGBXPxjwvD7R3LsIfLvC7k8hoTssHK2v8Fxm/vp6jVoDTpIamFELPh544AGGDBnCnj17APjoo4/w8fEBYN68eVWecyeT0/bs2UNBQQETJ04kOTmZwYMHM3r0aO69917s7e3ZsmULcXFxpKSkGFo85eLj44mPjzd8Xd5iMjc7azvGhoxlbMhYTlw9wY+pP7Lu/Do2pm8kyCWIKe2mMK7tOFxtXRskHlMwdj3InT5HaaWgjacjx7PysVEqUFlbNV7VAKW1fnfNDmPg6lk48K1+vGb5Q+Dopa991nUGtGht8ltrtDou3yihUK1p3BpwktRAjP5NPnToEOnp6YwdOxaFQsGpUzV3LQQEBFRosaSnp+Pv71/t852cnBg7dixr1qwBwNPT0zA7LSYmBn9/f1JTU40Nt0GFe4TzWu/X2Dh5I8/1eA6lQsm/9/+bQcsG8eL2FzmSe+SOEm5jqWk9SEyYp6Erq7Y1I1Vdp6WLLZ5ONvi42ZF1o9gyqgZ4hMCQ1/XbB0z8CjzDYPsH8N8u8H0cnFwDWk29b6PTCZJTc/jnmhO8vzGVY1n5nMq+WeF3Qi7qlJojowb5X3rpJY4ePUpKSgqpqalcunSJiRMnsnPnzmrP0Wq1hIaGsnbtWsMg/4gRI5g1a5bhOTk5OdjY2ODm5oZarWbChAmMGzeOhx9+mKysLEO15uPHjzNo0CBOnjyJu7t7jbFawoZjQgj2Z+/nx9Qf2Zi+kTJdGe1btGdyu8mMbjPaoqc6374eRN99Y0t2vlrfIrGxZmB7LxRCQVJq9WtGGmX3TFPITYX938DhJVCSB04++lZNtxngFlinS94+KeJU9k2OZOYR6e9KO2/nCq05c1VVkDPWJHOp9yyyqKgoDh48SNeuXTl4UF/VNjIykiNHjtR4XlJSEo899hhqtZqYmBjmz5/PunXrSExMZMGCBRw6dIhZs2ah1WrRarWMHDmSd955BysrK1544QVWr15t2OjsjTfeYPjw4fV6sY3hSvEVVp1ZxfJTy7lYcBFHlSOj24xmcrvJhLUIa+zwqlX+gbT/wjUSD2dVOWPs9g/Dqj7EmuwHm6YYjq+C37/RVwtAoS9LEz0LQofpu9qMuUwVu48KITidU8DZ3AIifF1wslUx4I+kvSnVtDPL5Iw1ydzqnWDuuecedu/eTZcuXTh48CBqtZro6GiOHj1q8mDry9ISTDmtTsvOrJ38eOpHtmZuRSd0RHpFMqXdFIYFD8PO2nTVgU3F2K2Zm/2HWPYJfaI5/AOo88C5FXR5ELo+aGjVVJdI84o0vL72OP7ulVut6dcKeXpwGD5uduw4c6XWqd91YcyUckmqj5o+c436M2zw4MG8/PLLFBUVkZiYyGeffcakSZNMGmRzp7RS0s+/H/38+3G58DIrTq/gp1M/8dKOl3hn3zuMDRnL5HaTCXELaexQDYzdlKx82q23ix0ejjYUa7SsPHgRoHl8iHmHw8h/w+B/wvGV8PtC2PoubP03ou1gjreK48f8CAo1VEquNU2ccLJV4eOm/8PCHNsFyG0IpMZm1G/XG2+8QZs2bYiMjOS7775j0qRJvPrqq+aOrdnycfQhPiqe9ZPW8+GAD4n0jGTJySWMXz2eGb/MYO25tai16sYO06hNye6qabc2DtBlGszZAI/uhB5z0V7YTcdtf+XZExOZkv8trUR2hTUtxhTSNNd2AXIbAqmxGdWCUSgUzJo1q8IAvVR/1lbWDAocxKDAQWTczGDFqRWsPLOS57c9zzt79a2aie0mNlpZGmM2Jcsr0tR76+UmOU7jHYFm2Du8VXIvXW9uptuVRHpmfk2PzG8479KDPSVj0LR+GJWNrWGm3KaTOVwtUONgY11hBp25tguQ2xBIja3GMZi5c+fWuLJ+/vz5ZgmqPix1DMZYGq2GpIwklp9azu5LuwGI9o5mcrvJDA4ajI3SppYrmFZtVYaNHaep6dqmGrupS6KqT3K7fXzFo/AMHbNXE567DruyfHQOXlh1mQZdp4NHSI33MtdYiRyDkcytzoP83377bY0XvrWumKVo6gnmVun56Sw/vZzVZ1ZzreQabrZujdaqMceHo6k+/OqSqEyR3KpLrhp1Ia1zkpjIRqzS/5jKH9xPv4Czw5gqt3s213YBchsCydzqPYusKWlOCaacRqthU8Ymlp9azp5L+moK3by7MTF0IkOChjT6DLS6fIjVp+VTfn55wqvLDCxTJbdar3PltL5awKGlUHQF7Nyg8336Vo13RI2vqz7bJJjiHEkyhkwwzUh6fjorTq9g1ZlVXCu5houNi75VEzqRtu5tGzW2O/kQM3b67u3Xub3lYWut5HJ+MSFeTjjY/DmkWFOiqm9yqyqeWpNrWam+ovOBb+HsZkDoKzt3na6v7GzrbPS9mu10cKlJkgmmGdJoNSRnJrP81HJ2Zum7YTp7dWZi6ESGBQ+z6GoBYNwCREdbleEDVCtElQs/bxSVkpSSQ9cgd8M2AOUyrxfxyugIXB1UFe57+UYJ729MJbBFxedXd46xr8foFsL1C/r6ZwcXwc0s/X41HeP0XWj+3avdhVOOp0iWSCaYZi7zZiY/nf6J1WdWk1Ocg5PKiZGtRzKx3UTCPRqm+GddGFNC5XJeMaEtnci+WUqhWsOxrHzaejkR2tIJhUKBVidISskGAYPCvbH648O5psWgVV2nqnPMTqeFM5v0rZpTv4KuDLza61s1kfeBo4fhqaZsdUmSKdV7oeVbb71V6Vj5lsndu3evX3RSvfk7+/N418f5a9Rf2Za5jRWnV7D89HJ+PPUjHVp0YGLoREa2GYmzTe3dMA3p1um7uTdLOJtbQKS/K6Et9RWz9XvJCFYcuMjgcG8cbeyAfM7lFmBjbUWwhyNKKwWhLZ05kH6dm8UaXB1sqqzUfPti0BKNjiOZeQCV6oHV9kFtsnESKyW0G6p/FOTA4aVw4DtY/wJseBXaj6Ss8wMU+vVHg6Le08Fr05DjNHJM6O5gVAtm2rRp7Nmzx1BJOTExkaioKM6dO8fkyZN57rnnGiJWo9yNLZiqXC68zKozq1h1ZhUXCy5ip7RjaPBQJoROoGvLrha1sVt13VZanWDLqVxKNFoGtW+JtdKKLadyUSAABTFhXlgpFBSVlnE2p5BWbnaU/LHGo7ap1FXVA6ttYoKptpOukRCQvgtx4Dt0x1ah1BZz3bolB1uM5FfVQFx9Q40eazJWQ47tyHGk5qfeXWT9+/dn7dq1uLi4APodLseMGcMvv/xCdHS0RX2gywRTkU7o2H1pNz+d/olN6Zso05UR7BLMhNAJjAkZg6e9Z2OHCFTdBVSi0ZKUmoO9tdKQTNKuFJJyOR+dgEHtW6KDWisR13VCwe1MtZ20MZJTc1i3/xSxmm10ubKGVgXHAThqE8VZ/zjSWw6kQKdqcmtl5DhS81PvLZNzcnIqbPTl6OhIdnY2Dg4O2NramiZKySysFFb09u3NezHvkTQ5iWejn0WpUPL+7+8zZNkQnkh6gq2ZWynTlTVqnFWVVNHqBJoyHX7u9oaxlSAPB9p4OaFQQPbNkgr7yqiUVrg6qCpUcs4r0qCyVlRb8qa8Hpgx3WKm2E7aGOXXcXP3INVvIv/rvJDvopayx/s+gsvOM/7cq8zdO4JBZ99heuA1+ret+x8JDVnq564qKyQBRo7BjBgxgsGDBzNlyhQAli9fzogRIygsLMTb29usAUqm427nzvSI6TwY/iCHcw+z8sxKfjn/C0kZSbS0b8m4tuOIaxtHgEtA7Rczg6pKqkzs6sfpnAJDuZOSMh021lb8fXgY3YM8quzDr6obxtvZhtM5Bfi42ldZ8qY2xhT+BEwyTlLVva46tmVn22dY0WIOL4WmY39sKT3Or0bx20o4FKHfsyZyCjjeWbIxtqCpKTTkvSTLYFSC+eCDD1i5ciXbtm0D4LHHHmP8+PEA/Prrr2YLTjIPhUJBVMsoolpG8ffuf2d92np+Ov0TXx79ki+Pfkl3n+7EtY1jcNBg7K3tGywuKysFsWEtK3R1KRUKwzqT2+t4VddnX1V159M5BYS2dCLnZmmV9cBqY8rtpOtzL1tbe+w7x6HqOhHys/QTAw4ugvXPw4ZXIGw4RD2g37vGiD1rGrJemayNdveR05Qlg7M3zvLT6Z9Ye24t10qu4axyZkTrEUwInUC4R3ijTgwwdtZRbdN5nx/ZHk2ZqNPspYYegzH6On9MDODgYv12AppCcPLWVwyIegC82pnuXvUkx2Can3oP8u/evZtXXnmFtLQ0ysr+7Ks/d+6c6aI0EZlg6k+j1bA1cys/nfmJ7Re3oxM6Qt1DiWsbx6g2o2hh16KxQ6xWTQP6dV1EWc6YVfumqv1V5+uoC+DEKnQHF2GVvkt/zL87RE2FiAlg72a6e9WBrI3W/NQ7wbRv354333yT7t27o1T++Vehn5+f6aI0EZlgTCunKIfEs4msOrOKC/kXsLayJtY/lrjQOHr79sbayritgxtKQyxINFe9MFNc59bxJ4eCNHrf/I17bq7HrugyWNtB+9H6ZNMmVr8OxwwxG0Oug2k+6p1gunfvzr59++74xsnJycTHx6NWq4mNjeWLL76okKAKCwsZMGAApaWlaDQa+vTpw6effoq1tTVlZWXMnTuXbdu2YWtry+eff06/fv1qvadMMOYhhOBgzkFWnlnJ+rT1FJcV42XvxeiQ0YxvO77R9qypyt3cDVPVa8/JK2RWq3Q6X10HJ9dAWQm4+Om70DpPBc/GrWEnNW31TjCvvvoqjo6OTJ48ucK0ZF9f32rP0el0tGvXjsTERMLDw5kyZQqjRo2qUOJfp9NRVFSEk5MTOp2OSZMmMXbsWGbOnMmCBQtISkpiyZIlHDlyhEmTJpGamlrrOIBMMOZXpClifdp6Vp1ZxYGcA4C+Dtq4tuMYHjy80SsG3K3dMEa13krz9eM0hxZD5h9/NPr3+KMLLa7KLjRJqkm9E0zr1q0rn6hQ1DgGs2fPHp599lm2bt0KwPr160lISCAxMbHK55eWlhIXF8eUKVOYMWMGI0aM4Nlnn2XgwIEA9OrVi4SEBLp27VpjrDLBNKwL+RdYfWY1q8+uJqcoB1ulLYMCBzGu7Th6+vREadV4M4Putm6YOx5/yj2ln4V2+H/6optKW+gwWt+qCRlQqQtNkqpS71pk58+fv+ObZmZmEhDw53qKwMBAMjIyqnxuz549SUlJYeTIkUybNq3G829PMAkJCSQkJBi+vn79+h3HKtVdkEsQj3d9nPioePZc2sOqM6vYeGEj686vw8fRhzFtxjCu7TiCXIIaPDb9wsv6J5amkqjueBqwVzsY/CoMfAnOb4FDS/RdaMdWgJOPfl1N1FRo2cFwSlN5LyTLUGOCuXLlCp6enmRlZVX5/Zq6yO5k9vOePXsoKChg4sSJJCcnM3jwYKPPj4+PJz4+3vB1eLjlVg9uzpRWSnr79aa3X2/yS/P59fyvrDqzyrC2pkvLLowLGcfQ4KGN3oVmrKZWN6u8GkJ14081Ft0MGah/lOTB8VX6Vs3O/+ofrTqji7yfXQ6x/HJe2yTeC8ky1JhgZs6cydq1a+nTpw8KhaLCh35tXWQBAQEVWizp6en4+/tX+3wnJyfGjh3LmjVrGDx4sOH80NBQo86XLIeLjQtTwqYwJWwK526cY/XZ1aw9u5bXdr3G23vf1nehhYyjZ6vG7UKrTVULNlcevAhgsZMFqqqGcCcLSrFzhW4zoNsMNLln0R76H7YnfsRq/XPcgxIvt96keo/mhHUvVh68SJlOV21FBUmqdQxGCMHFixfv+MNdq9USGhrK2rVrDYP8I0aMYNasWYbn5OTkYGNjg5ubG2q1mgkTJjBu3DgefvhhvvzySzZv3mwY5J8wYQKnT5+Wg/xNlFanZfel3aw+s5qkjCTUWjUtHVoyus1oxoWMo42b5cxCg6a//0p9urIq7RqqtMI5dz/DtcmEX9uIrbaQEqUzex1j+UnbF0VADxztbGSL5i5VrzEYhULBiBEjOHr06B3dVKlUsmDBAiZNmoRarSYmJoYHH3yQxMREEhMTWbBgAVlZWcyaNQutVotWq2XkyJHMmTMH0Leetm/fTtu2bbGxseGrr76yqBLz0p1RWinp49eHPn59yC/NZ33aehLPJPL1sa/5+tjXRHhEMDZkLCNaj8Ddzr2xw23ydbPqM/50e8vtRlEp6/KCuRz0FCEh/0fI9W0EZybSJ/9n+rOG6+f9Oeo5nOT8AUAPi23dSQ3PqFlkkydP5s0336Rdu5pLTlgC2YJpWtLz00k8m8ias2vIKszC2sqa/n79GRsyln7+/bBR2tR+ETNo6i2Yuqrqdd++a6gQsOVULu66a/RXb2WMYhvehSkAXHDshH//GSg7TQQHy634IJlOvacp9+7dm4MHD9K1a1ccHf/cEOq3334zXZQmIhNM06QTOn7P/p3Es4n8lvYbRWVFuNq6Mjx4OGNCxhDpGdngLdi7ccFmdVOd064UciD9OoPat8RWpSQpNQcroH0rF4I9HPEoPEOH3F8Izf4Ft7JcsFJBu2H6mWjthoO13Najuap3gtmyZUuVx2NiYuoXmRnIBNP0FZcVk5SexJpza9iVtQud0BHkEsSYNmMYHTIaP6eGKVHUnBdsVjdGU13L7dZdQ4tKyzielU+IlxOhLZ0Mib9Yo6VEXcorna5jfWwZnFgNpTf1EwfCx0PkvRDYC6yaX8vvblbvBAP6QfvTp08DEBoaWqHkiyWRCaZ5yS3KZd35daw5u4bU66kAdG3ZldEhoxkaNBRXW1ezx2Dpaz/uJD5jpl7X1HIr30ph/4VrJB7Oqrl1V1oEqevgyA9wZhMILbgGQKfJ+pbNLetrpKar3glmz5493Hfffbi76wdf8/LyWLp0KT169DBtpCYgE0zzlXotlbXn1rLu3DpyinNQWamIDYhldJvR9PPrh0pZtyrJTVVd1ukY0+1nlqrRhVfg2E9w9Mc/S9T4dIJOU6DTJHCpfk2dZNnqnWB69uxJQkIC0dHRAPz+++/89a9/Zc+ePaaN1ARkgmn+tDotey/vZe25tWy4sIHismJcbV0ZFjSM0SGjifKKuitmHN7pGNGdTlwwW9Xoq2fh6DJ9y+baOUABrfvpWzYdxsp6aE1MvRNM586dOXz4cK3HLIFMMHeXIk0RSRlJ/HzuZ3Zl7UIrtPg5+TGqzShGtRllUVWeTakus9zMuVdOnQgBFw/oWzXHVkBhrr4eWruh+pZN6FBQ2TVcPFKd1LsWma+vLwkJCcyaNQuFQsHXX39dY5kYSWooDioHRrcZzeg2o7lSfIX1aetZe3Yt84/MZ/6R+YR7hDOq9ShGtB6Bl4ORq9mbgLqs07G4LYsVCvDvpn8MfRPOJ8ORZZCyVl8TzdYVOozRd6G17i+LbzZBRrVgLl68yN/+9je2bNmCQqEgNjaWjz76SG44JlmstLw0fj7/M2vPriWzIBMrhRU9fHowsvVIBgcNbjL10KpT13U6TWLqdWkRnPoFjq6A07+BTqPfAjpigj7Z+HXTJyfJIphkFllTIROMdCshBEeuHOHncz+zPm0910quYWNlQ0xADKNaj2rUxZz1VZdk0eSmXhdfhxOJ+jGbtO2AAPdg6DhJn2zkTLRGV+8EExwczL333su0adOIjIw0eYCmJBOMVB2NTsPurN38fP5nktKTKC4rxlnlzJDgIYxsPZJo72iLLr55u/okC3NOvTbbtfOz9DPRji2HrIP6Yy0joNNE6DhRn3ikBlfvBJOdnc0PP/zAkiVLKCwsZOrUqUydOpWgoIbf46M2MsFIxijSFLElcwvrzq1j+8XtlIkyvOy9GN56OKNajyLcI7zJzESzlHU6Dbq9wdWz+okBR5fDFf36KPy76xNN+HhwaWXa+0nVMmkX2blz53jjjTf47rvv0Gq1JgnQlGSCke7UjZIbbEjfwLpz69ifvR+AQOdAhrcezsjWIwlxC2nkCJuG6rrsxnRuZb6S/kJA9jF9ojn2E+SlAwoI7gsdJ0CHceDoYdp7ShWYJMHk5ubyww8/sHjxYvLy8pg6dSovvfSSSQM1BZlgpPq4XHiZX8//yrrz6zh57SQA7dzbMaL1CEa0HtFgZWqamqomHQghOJ1TwNncAiJ8XXC0VZm3pL8QkLlf37I5vhIKLoNCqd/+OWICtB8l19iYQb0TzPDhw0lJSWHy5Mncf//9lbYttiQywUimkpaXxi9pv/DL+V84n6ffNryzV2dGtB7B0KChzWrac31VtcYm7UohKZfz0QkY1L4lOmi4GWs6LVzYqU82J1ZD8TVQ2kDbwfpkEzYcbJv2TEJLUe8Ek5SUxIABA5pEn7RMMJKpCSFIvZ7KuvPrWH9+PVmFWShQ0N2nO8NbD2dw4GCL2MOmMd3egtHqBFtO5aJAAApiwrywUigaZ7sDrQbOb9F3oZ1cC+o8sLbTL+TsOEH/XxvH2q8jVckkXWSnTp3i2LFjlJSUGI5NnTrVNBGakEwwkjkJITice5j1aetZn7ae3OJcrBXW9PTtyfDg4QwMHIiLjUtjh9kobh2DUUClkv7lGqVqQLkyNZxN0ieb1HVQWgAqB/2WAhFxEDoEVPYNH1cTVu8E85///IcffviBc+fOERsby4YNG4iNjWX16tUmD7a+ZIKRGopWp+VAzgF+Of8LGy5s4Ib6BiorFX18+zCs9TAGBAzAUXX3/GV867TpArWm2pL+da17ZnKaYji9AU6sgtRfQVMIKkcIG6FPNm0Hy1I1Rqh3gunUqRMHDhwgOjqaw4cPc/78eR5//HHWrFlT43nJycnEx8ejVquJjY3liy++qFDm/9ChQ8THx3Pjxg0UCgUPP/wwjz/+OAALFy7kmWeeISAgAIA+ffqQkJBQrxcrSeai0WnYe2kvv6b9yqb0TdwsvYmNlQ39/PsxPHg4/f3746CqXAOsOSpPFsaU9G/Qqc01KS2CMxv0kwNOrQdNEdg4/dGyGf9HspEtm6rUO8F0796dffv2ERUVxf79+7G2tqZTp04cPXq02nN0Oh3t2rUjMTGR8PBwpkyZwqhRo5gxY4bhOadOnUIIQVhYGPn5+XTr1o1ly5YRFRXFwoUL2b59OwsWLDDZi5WkhqDRatiZtZNf035lc8ZmCjWF2Cnt6O/fn2HBw+jn3w976+b/YWXMQlCLLF1TWqhPMidWwanfoKz4j2QzTL/GRnajVVDvYpfu7u7k5+czaNAgJk+ejKenJy1a1Lzf9r59+/D19SU8PByA2bNnk5CQUCHBtGvXzvBvFxcXOnToQEZGBlFRUcaEJUkWSaVUERMQQ0xADGqtmh0Xd7A+bT3JGcn8duE37K3tDcmmr1/fZptsrKwUxIa1NGxSVtXumRtP5hiSC+iLdXq72LHpZA592no2zsJRG0f94H/HCfpkc/o3fcsmZZ1+VprK8Y9kM06fbOQEgWrVmGCEECgUChITE7GxseHdd981rIP597//XeOFMzMzDd1bAIGBgWRkZFT7/LNnz7J//34WLlxoOLZmzRo6d+6Mt7c3b775Jt27d690XkJCQoWus+vXr9cYlyQ1JFulLQMDBzIwcCAlZSVsv7id9Wnr2ZK5hfVp6w3JZmjQ0GbbslEprSpVdoa6VYRucDaO+vGYiLg/k82J1XDqVzj+k36CQOiQP5LNMLB1atx4LUyNCaZbt24cOHCA+Ph4vvrqKwCmT59u1IXvpEDAjRs3GD9+PB999JGhZTR69Gjuvfde7O3t2bJlC3FxcaSkpODkVPEHGB8fT3x8vOHr8haTJFkaO2s7BgcNZnDQYIrLig3JZmvmVkOy6efXj6HBQ+nn16/Zj9lY3PYBtamQbIrgzMY/k82J1fqpz20H6zdNCxsOdubfztvS1ZhgSkpKSEpKYvv27ezatatS0ujdu3e15wYEBFRosaSnp+Pv71/peUVFRYwaNYq5c+cyefJkw3FPT0/Dv2NiYvD39yc1NZVu3brV/qokycLZW9szJGgIQ4KGGJLNhrQNJGfqu9HslHb08+/HkKAh9Pfv3yxno6mUVgzu0LLaMZjGrKtWKxsHCB+rf2hK9FOfy2ejpawFKxW0idV/P2zUXVuupsZB/pUrVzJ//ny2b99u2C7ZcKJCQVJSUrUX1mq1hIaGsnbtWsMg/4gRI5g1a5bhORqNhjFjxnDPPffw2muvVTg/KyvLsKnZ8ePHGTRoECdPnsTdveYFbXKQX2rKSspK2HFxB79d+I0tmVso1BRiY2VDb7/eDA0aSkxATLNaZ9Pktg+oTVmpflHnidWQ8rO+goBCCcF99C2b9qObXSHOes8ie+mll5g3b94d3zgpKYnHHnsMtVpNTEwM8+fPZ926dSQmJrJgwQIWL17M9OnT6dSpU4V7TZo0iRdeeIHVq1ejUqlQqVS88cYbDB8+vNZ7ygQjNRdqrZpdWbvYcGEDmzM2c7P0JtZW1tzT6h6GBA1hQMCAZlNBwFIqQpuUtgwu7ICTifoKAgWX9cf9e+h36uwwBlq0btwYTUBuOCZJTZxGq2H3pd1sTN9IUnoSN9Q3sFJYEe0dzeCgwQwKHERLBwvZkVKqTKeDzH1/JJtEuJGuP+7d6Y9kMxpahjfJnTplgpGkZqRMV8bv2b+z4cIGktKTyC3ORYGCzl6dDcnG37nyeKdkIYSAy0fg5Br9bp3l+9m4t/6zZeMXDVZNoyVX5wRTPk25KZEJRrqb6ISOw7mH2XBhA5subCKrMAuA9i3aMzhQP2OtjWubJvf/8V3lyml9sklZCxd/1x9z8oH2I/VjNsH9wNpyt/Wuc4Lp2rUrBw4cYPbs2YZpypZOJhjpbiWEIOVaij7ZpG/iXN45AIJdghkYOJBBgYPo6NkRK0XT+Mv4rpR3UT85IGUNpO0AoQVbV2g3VL+fTdvBFrfNQJ0TTHh4OJ988gmPPvooCxcuvKNpyo1FJhhJ0juXd45NFzaxKX0Tx68eB6ClfUsGBA5gYOBAuvt0R2XVCBWNG1mTmVBQdE2/sPPkGjizSV+yRmmjn/7cfhS0GwHO3o0dZd0TTH2mKTcWmWAkqbLLhZfZlL6JpPQkfs/+Ha3Q4mzjTIx/DAMDB9LHt4/FL+ysb2KwmMKadVFaBOeS9d1oqb/opz+jAP9ofbIJGwVe7Wq7ilk02jTlxiATjCTV7EbJDZIzk9mUvoldWbtQa9XYWNnQy7cXAwMHEuMfg4e95SwMrCkxaIUwOulYZGHNutCWQcYefVda6s9wPU1/3KMthI3UJxz/7mDVMJUQTDKLbP369WzevBmAQYMGMWTIENNFaEIywUiS8Yo0Rey6tIuk9CS2ZG4hT52HAgVdWnZhQMAAYgNiCXYNbpTYair7fzmvmNCWTmTfLDWqNXL7jpvlGmWHTVMSAnJO6hNNyjrIOqA/7uCh32ogbASEDDRrQc56J5h58+axZs0apk6dikKhYOnSpYwePZoXX3zR5MHWl0wwklQ3ZboyDmQfICkjic3pmw0z0lq7tmZAwAAGBAwg0ivS7JMEbm2xFKo1HMvKp+1tG5edyr7JyUv5DA73xvGPemY1tUbyijS8vvY4/u6VuwEbdYdNU8vP0tdGS/0Fzm0BrRqUtvpxm7Dh+nEbE1cSqHeCiYyMZN++fdja2gL6GmU9evTgyJEjJg3UFGSCkaT6E0Jw6vopQ7I5ee0kAB52HsQExBDrH8s9vveYpfpzbVsva3WCLadyKdFoGdS+JbZ/tEiKNVoKS8p4amg7XO1VlbYGaJYtmJqoC/Q10lJ/0Sed4mv6462i9F1pYcPBJ7LeizvrvR+MEAKrWxb9WFlZ3VG1ZEmSmhaFQkFYizDCWoTxaOdHuVx4mc0Zm9mcvpnEs4n8dPon7JR23ON7DwMCBtDfvz+e9p61X7gWt+8Ro9UJ7KyVKBCczy0ksIUDGq2OkjJ9BWaVtf5zSQhBdl4JRy/mcTPxOC4ONhW6zJp0Yc26snX6syCnTgsZe+HUL/qEk/yW/uHiD50mwpDXzRKCUQkmLi6OAQMGMG3aNACWLFnCxIkTzRKQJEmWx8fRh/vb38/97e+noLSAHVk7SM5IZmvmVpIzklGgoJNnJ/1Ga/4xtHNvV6fFnbfvEaO0UtDG05GUy/noBGjKdGh1Ak2ZjrZeTlj9cY8LV4s4npWPnbUVQZ6OqMt0rDx4EcDQZdY/1AuATSdzuFqgxsHGmrgufobjzZqVEoJ66R9DXoerZ//sSsu7aLbbGj3In5iYSHJyMgqFgtjYWMaMGWO2oOpDdpFJUsMp05VxMOcgyRnJbMncwoX8CwC0cmxFjH8MsQGxdPfpjo3SuJXoVXVlCSE4nVPA2dwCInxdcLJV0dLZhtM5Bfi42mOjtCIpJZvSMh0Rfq4Ee+gHtKvr/moy62Aaik5Xr7I0shaZJEkN4nzeeZIzkknOSOZQ7iF0Qoe9tT29fXsT4x9DP/9+tXalVTedeEznVnQP8sDBVolSoTCU+b9RVMrxrHw6+bsS1MKhQsuprgP4MgkZTyYYSZIa3I2SG2y7uI0tmVvYeXEnNzU3Aejo0ZH+Af2J8Y+hQ4sOlbrS7mSPGI1WR16Rhvc3nMLJzrreA/hNejFmI5EJRpKkRqXRaTiUc4gtGVvYenEr5/POA+Bl70U//3709+9Pr1a9KlQTuJNWhKkWUTabxZgNqN4JprCwEEfHprFlq0wwkmT50vPT2ZK5ha2ZW9mfvZ8yXRkqKxXR3tH09+9Pf//+BLoEGn09U+yMaclTmS25y67eCSYgIIARI0YwZ84cevToYfIATUkmGElqWgo1hezO2s3Wi1vZlrmN3OJcAIJcgujn149+fv3o5tMNW6VtrdeqzwexJS7GbApddvVeB3Pq1CmWLVvG//3f/5Gfn8+cOXN44IEHcHNzM2WckiTdhRxVjgwKGsSgoEHohI6UaylszdzK9ovbWXxyMYtOLsLe2p6ePj3p69eXfv798HXyrfJaKqUVrg51+wvfwVaJ/R9VAW5vwTjYWONg2zC1vW619XSuocvOw9GGYo220vRrS3bHYzA7d+5kypQpXLt2jSlTpvDaa68RHBxc5XOTk5OJj49HrVYTGxvLF198gVL55w/p0KFDxMfHc+PGDRQKBQ8//DCPP/44AGVlZcydO5dt27Zha2vL559/Tr9+/WqNT7ZgJKn5uF5ynZ1ZO9l2cRs7Lu7ghvoGoC9f09evL339+hLtHW30NOjaWNIYjCV32d2q3i0Y0CeL+fPns3//fh5++GGmT5/Oli1bGD58OCkpKZWer9PpmDNnDomJiYSHhzNlyhQWLVrEjBkzDM9xcHDg66+/JiwsjPz8fLp160b//v2Jiopi4cKFqNVqzpw5w5EjR5g0aRKpqalyZz5Juou427kzqs0oRrUZhVan5eiVo2y/uJ3tF7fz/Ynv+f7E99hb29PDp4ch4dRnu2hLWox5+6LTcvYqJVcL1BSptXVurTUUoxJMWFgY7dq14+GHH2bRokWGsjHBwcH8+uuvVZ6zb98+fH19CQ8PB2D27NkkJCRUSDDt2v25f4GLiwsdOnQgIyODqKgoVqxYwbPPPgvoa6F5eHhw8OBBunbtWrdXKklSk6a0UhLVMoqollE81uUxrhZfNbRudmbtZEvmFkC/g2cfvz708e1DtE/0HdVLs7JSEBvWkj5tPRt9UN0Su+zulFEJ5ssvv6R///4VjqWnpxMYGMjSpUurPCczM5OAgADD14GBgWRkZFR7j7Nnz7J//34WLlxY4/m3J5iEhAQSEhIMX1+/ft2YlyRJUhPnYe/BmJAxjAkZg1an5fjV4+y4uIPtWdtZmrKUxScXY2NlQ7RPNL19e9PHtw8hbiFG9YLUZyzHVJpD/TSjEsyTTz7JgQMHKhwbP358pWO3upOhnRs3bjB+/Hg++ugjWrRocUfnx8fHEx8fb/i6vMUkSdLdQ2mlJNIrkkivSB6NepQbJTfYfWk32y9uZ0fWDnZm7eQ93sPbwZvevr3p7debXq164WrrWud7GjNjrS6z2m49x5K67OqixgSTmZnJhQsXKCwsZOfOnYbjeXl5FBUV1XjhgICACi2W9PR0/P0r940WFRUxatQo5s6dy+TJkyudHxoaWuP5kiRJt3Ozc2N46+EMbz3csPXAzqyd7Mjawdpza1l5ZiVWCis6enSkt19vevv2pqNnR1RWtU9DNmbqcF2mF9d0jiV02dVFjQlm06ZNLFy4kKysrAqbizk7O/Ovf/2rxgtHR0eTmZnJiRMnCA8P56uvvmLChAkVnqPRaJgwYQJDhgwxzB4rN2HCBBYsWMDAgQM5cuQIubm5cvxFkqQ7VqYT+Ni14YEOoczqOIsiTRH7s/ez46K+ZfP54c/5/PDnOKmc6OHTQ9/C8e1NgEtAldczZupwXaYX13ZOY3fZ1YVR05SXLFnC1KlT7/jiSUlJPPbYY6jVamJiYpg/fz7r1q0jMTGRBQsWsHjxYqZPn06nTp0M57z00ktMmjQJjUbDnDlz2LFjBzY2Nnz22WfExMTUek85TVmSJDC+FZFVkMWurF3szNrJ7ku7yS/NB8DPyY9evr3o1aoXPVv1xNXW1aipw8AdTy9uKlOSq1LnlfxHjhwhMjKyQvfYrXr37m2aCE1IJhhJkqBua1q0Oi0nrp5gZ9ZOdl3axeGcw5SJMqwUVoS3CKeLVw9OnvMhwiMSpaJid1r5an/gjisCWGIVAWPVeR3Mxx9/zJdfflmhe6ycQqEgKSnJNBFKkiSZ0O07Y4J+/Yi3ix2bTubQp61nlS0CpZWSTl6d6OTVib90/ouhO21X1i52Ze3i+5SvATh4yRYf23B8bSPxtY3EXvhVmDp8p9OLm8OU5KrUmGC+/PJLADZv3twgwUiSJJmCqRYpOqgcDMU3AbILs1l48Dc2nN/OVc0JLqoPAqDCmU4e0SSejaFnq57VTi8e07nVH4P1VEhwzWFKclVqTDDVdY2Vs8QuMkmSJHO1CLwdvXm29wP09BrGxhPZ5KovcFNxEmF3ipS8PRzYpf9j3M/Jj0DXzpy+2RYnEYarjTttvRxJTsnll6OXqxwPaupTkqtS4xjMgAEDqj/RQrvI5BiMJElg/rpit69x0eg0HLtyjN2XdrPn0h4O5x6mTFcGgK99G5SlobR27EygQye0WttqY7Hk0vxVkRuOSZJ01zHFHjH1UaQp4kDOAXZe3E1i6lbydGmAQIEVnqoQvKwjcLMK599jxuJi2zT226qKnEUmSdJdq7FbBOUzxDxdtVxWH+eS+iiXSo9xoywTAJWVikivSHr49KC7T3c6e3U2WXXohlDnBDN37ly+/PLLKrvKZBeZJElS7apb43Kt5AoZxUcJ9Mti3+W9ZBboE46t0pYoryi6+3SnR6sedPToiEppmVOUQXaRSZIkNSpjxoMuFVxi7+W97L28l32X93Gp8BIA9tb2hoTT3ac7EZ4RRpW0aSj1TjBlZWXMnz+fLVv05bBjY2OZM2cOKpXlvMhyMsFIkmRp7nQ8SAhBZkEm+y/vZ9/lfey9vJfsomzgz4QT7RNNd5/ujd7CqXeCmTVrFjdv3mT69OkALFq0CAcHB0NpfUsiE4wkSZaqruNBQggyb2ayL1ufbPZf3m9IOHZKOzp7daabTzeivaOJ9IrEVmlr0vvXpN4Jpn379hV2rRRC0KFDhyp3smxsMsFIknQnGnsSQF3c2sLZn72f/Zf3k1WYBegnDXTy7ES0TzTdvLsR5RWFndL+jqs7G6veWyYHBARw+fJlfHx8AMjJySEwMLBeQUmSJDWmupTUtxQKhYIA5wACnAOIC40D4GLBRX7P/t3wmH9kPgDWCmv8HUPRFbcm2LEjAa4R6LTKWqs7m0KNCWbu3LkoFArs7e2JiIhg8ODBKBQKNm7cSJ8+fcwWlCRJkrnVpaS+JfNz8sPPyY+xIWMByCnK4UD2AfZe3s+vZ3ZwU7eO9Px1kK/A3ToAT+sOfHs4jHa+Y/B19jFLTDUmmL59+xr+PXHiRMO/R4wYYZZgJEmSGkJdi2E2JS0dWjK89XB6eQ8iL0O/DidbfZLs0pNcLj3BmZINCNbzt6TfWDFumVliqDHBzJgxwyw3lSRJakymKobZFJTXZRNaW4LsexBk3wOA/NJCsopOcl8XP7Pd26gxmOvXr/Pee+9x9OhRSkpKDMd/++03swUmSZJkLs21PH5VqqvUfO0m3N9lMLGB5usONCpFz5gxA09PT1JSUnj88cdxdHSU2xdLktRklX/oZueXUKzRAhgWPw7q0LLJd4/drn+oF3Fd/Cgu1ZJ5vYjiUm2DVGo26l3MyMjgqaeews7OjtGjR7N8+XKj9ohJTk4mIiKCtm3bMmfOHLRabaXnTJkyBS8vL9q2bVvh+MKFC/Hw8CAqKoqoqCji4+ONfEmSJEm1a6wP3cZgZaUgNqwlr4wJ55XREbwyJpzYsJZmny1nVBeZjY2+n9LZ2Zm0tDS8vLzIzc2t8RydTsecOXNITEwkPDycKVOmsGjRokrjOo888gjPPfccU6ZMqXSNuLg4FixYYOxrkSRJMlr5h26ftp4mWQfTFNbTqJRWDTq2ZFSCGT9+PNevX+e5554jOjoaIQRPP/10jefs27cPX19fwsPDAZg9ezYJCQmVEszAgQNJS0urW/SSJEn1VN8P3aa8nsbcjEowzz//PABjxowhKyuLkpISXFxcajwnMzOTgIAAw9eBgYFkZGTcUXBr1qyhc+fOeHt78+abb9K9e/dKz0lISCAhIcHw9fXr1+/oHpIkSfXR3NbTmJJRCaYuxS7rW6R59OjR3Hvvvdjb27Nlyxbi4uJISUnBycmpwvPi4+MrjM+Ut5gkSZLM7W5YT1MfRiWYuXPnVip2uWfPnhqLXQYEBFRosaSnp+Pv7290YJ6enoZ/x8TE4O/vT2pqKt26dTP6GpIkSeZ0N62nqQujEsyuXbsqFLYcM2YMHTp0qPGc6OhoMjMzOXHiBOHh4Xz11VdMmDDB6MCysrLw9fUF4Pjx46SlpdGmTRujz5ckSTK3u2k9TV0YlVrLi12WM6bYpVKpZMGCBUyaNImQkBCcnJx48MEHSUxMZM6cOYbnjRo1il69epGWloa/vz9vv/02AJ988gkRERFERUUxc+ZMFi5ciLu7e11eoyRJklk09fU0Gq2OvCINGq3OLNevdctkhULB5cuX2bFjR6Vil6tXrzZLUPUhy/VLktSQ7nQzMUtgyplvdS7XL4tdSpIk1czU62kaQkPNfLujYpfldcjs7OxMFoAkSVJz0NCLGOuqIWe+GXWV06dPc8899+Dv74+/vz+9e/fm9OnTJglAkiTJ0pl7rOJO1See8plvt05KAH2SKSoto0hduaRXXRk9Tfnvf/+7YRbYypUrmTt3LsnJySYLRJIkydJY2ip9U8TTkDPfjGrB3Lhxo8IU47i4OG7cuGGyICRJkixR+ViFg40Sf3cHHGz0Ww1vPV1zLUZLjqchZ74ZdSUXFxd+/vlnw9fr1q2rtVSMJElSU1bbWEVDd5eZMp6GqiRtVBfZ559/zgMPPMBDDz0E6NfFfP/99yYNRJIkyZJY2ip9U8bTUDPfak0wWq2WDRs2cODAAfLz81EoFDg7O5s8EEmSJEtiaav0a4tHZa0gr0hzR8nC3DPfar2yUqlk0aJFgL6rTCYXSZLuBpa2Sr+6eC7nFdPS2Ya31qXw+trj/HPNCZJTc9Dp6ldw2BSM6iIbPHgwr732GtOmTcPR0dFwvLxWmCRJUnNUPiax6WQOVwvUONhYN+qul1XFE9rSidM5Bfi42lvcdgE1loop17p168onKhScO3fOLEHVhywVI0mSqVnabpXl8aisFby1LgUHG2WlbrPiUi2vjAk3e7x1LhVT7vz58yYNSJIkqSkx51hFXZJXeTx5RRqLmohwuxoTzPnz53n66ac5c+YMUVFRvPfee3h7ezdUbJIkSc1WU1s0WRc1praHHnqIsLAw/v3vf+Ps7MyTTz7ZQGFJkiQ1b01t0WRd1NiCuXLlCv/6178AGDZsGF27dm2QoCRJkpozUxactLSJCLeqMcGoVCrDvxUKBQqFZe5tIEmS1JQ0xKJJjVbHzaKyRp2YUGOCOXToEDY2f74BZWVl2NjYIIRAoVBQWlpq9gAlSZKaG3OMnZQP/Ot0guTUHIso0FljWtPpdJSWlhoe5V9rNBqZXCRJkurInGMnllSg06ztpuTkZCIiImjbti1z5sxBq628z8CUKVPw8vKibdu2FY6XlZUxa9Ys2rZtS0REBNu2bTNnqJIkSQ3KHAUnLa1Ap9kSjE6nY86cOSxbtowzZ86Qn59vKDlzq0ceeYT169dXOr5w4ULUajVnzpxh6dKlzJ49GyPWhEqSJDUJ5WMnr4wJ55XREbwyJpzYsJb16sZqyM3EjGG2BLNv3z58fX0JDw8HYPbs2axYsaLS8wYOHEiLFi0qHV+xYgVz5swBIDIyEg8PDw4ePGiucCVJkhqFfuxEZZKB+FvHdm7VWOtizJZgMjMzCQgIMHwdGBhIRkaGyc9PSEggPDzc8Lh+/Xr9ApckSWqiLG1djFGlYuqivt1Zxp4fHx9PfHy84evyFpMkSdLdyJLWxZgtwQQEBFRocaSnp+Pv73/H54eGhtbpfEmSpLtRQ20mZlQs5rpwdHQ0mZmZhiqbX331FRMmTDD6/AkTJrBgwQIAjhw5Qm5urqwkIEmSZCRTju3UldnurFQqWbBgAZMmTSIkJAQnJycefPBBEhMTDYP3AKNGjaJXr16kpaXh7+/P22+/DcDMmTNRqVS0bduW++67j6+++kpWEpAkSWpCjNoPpimR+8FIkiQ1nJo+cxt/5xxJkiSpWWp2LRgXFxezTwa4fv067u7uZr2HqTW1mJtavND0Ym5q8YKMuSHcabyZmZnk5+dX+b1ml2AaQlPshmtqMTe1eKHpxdzU4gUZc0MwZbyyi0ySJEkyC5lgJEmSJLOQCaYObq0c0FQ0tZibWrzQ9GJuavGCjLkhmDJeOQYjSZIkmYVswUiSJElmIROMJEmSZBYywdQgIyODQYMG0aFDByIiInj++ecN33vuuedo27Yt7dq1q3Kfm8Y0dOhQoqKi6NSpE5MmTTLMUbfkmEHf92tt/Wf9VUuONzg4mIiICKKiooiKiuLo0aOAZcdcWFjIjBkzCAsLo3379nzxxReAZcack5NjeG+joqLw8fEhLi4OsMx4yy1atIjIyEiioqLo168fqampgGXH/Pnnn9OxY0ciIiL461//ath52CQxC6laWVlZYt++fUIIIdRqtejbt69YtWqV2LBhg+jXr58oKysTmZmZIiAgQNy8ebORo/3TjRs3DP9+4oknxKuvvmrxMW/dulVMnz5dKJVKIYSw+HiDgoJERkZGhWOWHvNf/vIX8c477wghhNDpdCI7O9viYy43aNAgsXjxYouOt7CwULRo0ULk5uYKIYT47LPPxKRJkyw65mPHjomQkBDDZ8bjjz8uvv32W5PFLFswNWjVqhXR0dEA2NjY0KVLF9LT01mxYgUzZ85EqVTi5+dHnz59+O233xo52j+5uroC+m2rS0pKUCgUFh2zWq3mueee47333jMcs+R4q2PJMd+8eZPExESefvppABQKBS1btrTomMtlZWWxf/9+xo8fb9Hx6nQ6hBAUFBQAkJeXR6tWrSw65hMnTtCzZ0/DZ8awYcP44YcfTBazTDBGunbtGqtWrWLIkCH13q2zIcTFxdGyZUtSU1N55plnLDrm119/ndmzZ+Pl9eeGSJYcb7kxY8YQFRXFiy++iEajseiYz507h7e3N4899hhdu3YlLi6OCxcuWHTM5ZYuXcr48eNxcHCw6HidnJz45JNP6NixI35+fnz77be88cYbFh1zZGQk27dvJysrC61Wy7Jly8jIyDBZzDLBGKG0tJRJkybxxBNP0L59+3rv1tkQVq5cSVZWFv7+/ixfvtxiYz5y5Ah79uxh1qxZFY5barzltm3bxsGDB9mxYwepqam89957Fh1zWVkZhw4dYtKkSRw4cIAxY8bw0EMPWXTM5RYtWsQDDzwAWPbvhUaj4dNPP2Xfvn1cvHiRSZMm8Y9//MOiYw4LC+PNN99k7Nix9O3bl4CAAJRKpclilgmmFlqtlqlTpxIVFcUzzzwD1H+3zoZiY2PDfffdx8qVKy025h07dnDixAlat25NcHAwWq2W4OBgvLy8LDLecuV/3Tk6OjJnzhx27txpse8xgL+/Px4eHgwePBiA++67j99//92iYwZ9F05ubi4DBw4ELPv/vUOHDiGEoEOHDoD+Pbb03wuABx54gP3797Nr1y66dOlC+/btTRezaYaKmq+HHnpIzJw5U+h0OsOx3377rcIAmL+/v8jPz2/EKP+Un58vsrKyhBBCaLVa8fDDD4vnn3/eomO+VfkgvyXHW1BQIPLy8oQQQpSVlYk5c+aIF154waJjFkKI/v37i99//10IIcSqVatE7969LT7m5557Tjz77LOGry053qysLOHh4SEyMzOFEEJ8+OGH4t5777XomIUQ4vLly0IIIfLy8kTPnj3Fjh07TBazde0p6O61Y8cOvv76azp27EiXLl0AeOihh3j88cfZsGED7dq1w8rKivfffx9nZ+dGjlbv5s2bjBs3DrVajU6no2fPnrz00ks4ODhYbMxVGTJkiMXGm52dzYQJE9DpdGi1Wnr16sWLL75o8e/xZ599xuzZsyksLMTNzY0FCxbQoUMHi41ZCMGSJUtYs2aN4Zgl/160atWKf/3rXwwZMgSVSoWXlxdff/01gYGBFhsz6Fta2dnZCCF49tln6d27N4BJYpalYiRJkiSzkGMwkiRJklnIBCNJkiSZhUwwkiRJklnIBCNJkiSZhUwwkiRJklnIBCNJDWDv3r0oFAoSExMbOxRJajAywUhSA/j+++/p378/ixYtauxQJKnByAQjSWZWVlbGihUr+O6779iyZQt5eXkAbNy4kfDwcLp06cLf//532rZtazjn448/pkePHnTu3Jk5c+ag0WgaK3xJqjOZYCTJzH799VeioqIICgpi5MiRLF++nJKSEmbOnMny5cs5ePAgOp3O8PykpCT27t3L7t27OXz4MHZ2dixYsKARX4Ek1Y0sFSNJZrZo0SKmTZsGwLRp05g3bx7R0dH4+voSHh4OwIMPPsiqVasAWLduHVu3bqVr164AlJSUYG9v3yixS1J9yAQjSWaUn5/P2rVr2blzJy+++CJCCC5fvsyNGzcqPE+hUBj+LYTgqaee4sknn2zYYCXJxGQXmSSZ0YoVKxgzZgzp6emkpaVx4cIF/va3v7Fz506ysrJISUkBqDD4P2LECL755htDErp+/Trnz59vjPAlqV5kgpEkM/r++++ZPHlyhWOTJ09m0aJFfP3118TFxdGlSxdKS0sN29YOHjyYRx55hP79+xMZGcmgQYPIzMxsjPAlqV5kNWVJaiQFBQU4OTkB8Oabb5KZmclnn33WyFFJkunIMRhJaiTfffcd8+fPR6PREBwczNdff93YIUmSSckWjCRJkmQWcgxGkiRJMguZYCRJkiSzkAlGkiRJMguZYCRJkiSzkAlGkiRJMguZYCRJkiSz+H/GC8dKFkQEUwAAAABJRU5ErkJggg==", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Solution\n", "\n", "grouped = gss.groupby('age')\n", "favor_by_age = grouped['grass'].mean()\n", "\n", "plt.plot(favor_by_age, 'o', alpha=0.5)\n", "plt.plot(df['age'], pred_male, label='Male')\n", "plt.plot(df['age'], pred_female, label='Female')\n", "\n", "plt.xlabel('Age')\n", "plt.ylabel('Probability of favoring legalization')\n", "plt.title('Legalization versus age, grouped by sex')\n", "plt.legend();" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Summary\n", "\n", "At this point, I'd like to summarize the topics we've covered so far, and make some connections that might clarify the big picture.\n", "A central theme of this book is **exploratory data analysis**, which is a process and set of tools for exploring a dataset, visualizing distributions, and discovering relationships between variables.\n", "The last four chapters demonstrate the steps of this process:\n", "\n", "- Chapter 7 is about importing and cleaning data, and checking for errors and other special conditions. This might not be the most exciting part of the process, but time spent understanding data can save you from embarrassing errors.\n", "\n", "- Chapter 8 is about exploring variables one at a time, visualizing distributions using PMFs, CDFs, and KDE, and choosing appropriate summary statistics.\n", "\n", "- In Chapter 9 we explored relationships between variables two at a time, using scatter plots and other visualizations; and we quantified those relationships using correlation and simple regression.\n", "\n", "- Finally, in this chapter, we explored multivariate relationships using multiple regression and logistic regression." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We moved through a lot of material quickly, but if you practice and apply these methods to other questions and datasets, you will learn more as you go.\n", "In the next chapter, we will move on to a new topic, resampling, which is a versatile tool for statistical inference." ] }, { "cell_type": "markdown", "metadata": { "tags": [ "remove-cell" ] }, "source": [ "Having finished this chapter, you might appreciate [this xkcd cartoon](https://xkcd.com/2048/).\n", "\n", "" ] }, { "cell_type": "markdown", "metadata": { "tags": [ "remove-cell" ] }, "source": [ "*Elements of Data Science*\n", "\n", "Copyright 2021 [Allen B. Downey](https://allendowney.com)\n", "\n", "License: [Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International](https://creativecommons.org/licenses/by-nc-sa/4.0/)" ] } ], "metadata": { "celltoolbar": "Tags", "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.14" } }, "nbformat": 4, "nbformat_minor": 2 }