nb-m180-evaluating_and_tunning_machine_learning_models_NEW.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "slide"
    }
   },
   "source": [
    "![data-x](https://raw.githubusercontent.com/afo/data-x-plaksha/master/imgsource/dx_logo.png)\n",
    "\n",
    "---\n",
    "\n",
    "# Data-X: Machine Learning Model Evaluation -- Regression.\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    "**Author List (in no particular order):** [Elias Castro Hernandez](https://www.linkedin.com/in/ehcastroh/), [Debbie Yuen](http://www.debbiecyuen.me/), [Arash Nourian](https://www.linkedin.com/in/arashnourian/), and [Ikhlaq Sidhu](https://ikhlaq-sidhu.com/) \n",
    "\n",
    "**Video Walkthrough:** To view walkthrough of this notebook, click [here]()\n",
    "\n",
    "**References and Additional Resources:** See end of this notebook for additional information related to TensorFlow and Keras.\n",
    "\n",
    "**License Agreement:** Feel free to do whatever you want with this code\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"model_thumbnail-01.png\" align=\"center\" width=\"60%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "    \n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "The **generalization** performance of a learning method is related to its capability (e.g. accuracy) on an independent test data set -- generally called the validation set.  Assessment of this performance is extremely important as it guides the approach to learning, and gives us a measure by which to evaluate and compare models. This module covers key terms, techniques, and metrics used to assess and improve a model’s performance -- with a strong focus on evaluation, as opposed to theory. Moreover, this model will focus on data with a _stationary_ distribution (no additional, 'on-line,' data needing to be considered). If interested on how to assess, improve, and validate non-stationary  machine learning models see [here](https://www.amazon.com/Machine-Learning-Non-Stationary-Environments-Introduction/dp/0262017091).\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    "<strong style=\"color:red\">KEY CONSIDERATION:</strong> Some of the following content may be written for machines running on Linux or Mac operating systems. If you are working on a Windows machine, you will need to enable the Linux Bash Shell, or adjust Shell commands to PowerShell syntax. A tutorial on how to enable the Linux Bash Shell on Windows 10 can be found [here](https://youtu.be/xzgwDbe7foQ)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "**Setup Environment**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Pyton 2 and 3 support\n",
    "from __future__ import division, print_function, unicode_literals\n",
    "\n",
    "# data manipulation\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "# data visualization \n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import altair as alt\n",
    "import missingno as msno\n",
    "\n",
    "# Machine learning section 1\n",
    "from sklearn import preprocessing, model_selection\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from sklearn.metrics import mean_squared_error, r2_score\n",
    "\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "# Hide warnings\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Initialize Virtual Environment**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Create Virtual Environment ##\n",
    "! python3 -m venv ./venv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Activate Virtual Environment ##\n",
    "! . ./venv/bin/activate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "## **REGRESSION** MODELS  \n",
    "\n",
    "<br>\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"least_squares_regression.png\" align=\"center\" width=\"40%\" padding=\"10px\"><br>\n",
    "\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    "We begin our discussion on model evaluation and complexity by examining the topic of [bias-varianced tradeoff](https://en.wikipedia.org/wiki/Bias%E2%80%93variance_tradeoff) within the context of supervised learning -- in particular using linear methods. Although the following modeling approaches may appear as pendantic, it is our firm belief that understanding complext machine learning topics within the scope of the more interpretable linear models is essential for the same concepts on more advanced more advanced models. Moreover, building models is extraordinarily simple when using packages like ScikitLearn and TensorFlow, while assessing a model and tuning it requires care and proper technique. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### BACKGROUND\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Suppose $X^T = \\left(X_1, X_2, \\dots, X_p\\right)$ is an input vector and we want to predict some real-valued output $Y$. The [linear regression](https://en.wikipedia.org/wiki/Linear_regression) model \n",
    "\n",
    "$$\\displaystyle{ g(X) = \\beta_0 + \\sum_{j=1}^p X_j \\beta_j + \\epsilon}$$\n",
    "\n",
    "assumes that the regression function is linear, or that function is sufficiently approximated by the regression function $\\displaystyle{E(Y|X)}$. In this model, $X_j~'s$ are variables, which can come from different sources, $\\beta_j~'s$ are unknown parameters to be approximated, and $\\epsilon$ is zero-mean noise with finite variance that is independent of $x,y$.  Suppose we split our data is sinto training and testing data. The training data $(x_1, y_1), (x_2, y_2), \\dots, (x_N, y_N)$ is then used to approximate the parameters $\\beta_j$. As such, a simple or multiple regression model is essentailly a hypothesis concerning the relationship among the dependent and independent variables. \n",
    "\n",
    "A popular approach of estimation is know as the method of [least squares](https://en.wikipedia.org/wiki/Least_squares). In [_least squares_](https://youtu.be/PaFPbb66DxQ?t=13) we pick coefficients $\\beta = \\left(\\beta_0, \\beta_1, \\dots, \\beta_p\\right)^T$, in an effort to minimize a metric measuring 'nearness.'\n",
    "\n",
    "It should be noted that there are more metrics for evaluating regression problems -- depending on input(s), output(s), and data type(s) -- that what is covered on this notebook. For more info, see the resources section at the end of this notebook."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **CASE STUDY:** LINEAR **REGRESSION**\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**About Data:** [Forest Fires Data Set](http://archive.ics.uci.edu/ml/datasets/Forest+Fires)\n",
    "\n",
    "1. X - x-axis spatial coordinate within the Montesinho park map: 1 to 9\n",
    "2. Y - y-axis spatial coordinate within the Montesinho park map: 2 to 9\n",
    "3. month - month of the year: 'jan' to 'dec'\n",
    "4. day - day of the week: 'mon' to 'sun'\n",
    "5. FFMC - FFMC index from the FWI system: 18.7 to 96.20\n",
    "6. DMC - DMC index from the FWI system: 1.1 to 291.3\n",
    "7. DC - DC index from the FWI system: 7.9 to 860.6\n",
    "8. ISI - ISI index from the FWI system: 0.0 to 56.10\n",
    "9. temp - temperature in Celsius degrees: 2.2 to 33.30\n",
    "10. RH - relative humidity in %: 15.0 to 100\n",
    "11. wind - wind speed in km/h: 0.40 to 9.40\n",
    "12. rain - outside rain in mm/m2 : 0.0 to 6.4\n",
    "13. area - the burned area of the forest (in ha): 0.00 to 1090.84\n",
    "(this output variable is very skewed towards 0.0, thus it may make\n",
    "sense to model with the logarithm transform).\n",
    "\n",
    "For more information, read Cortez and Morais, 2007."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>X</th>\n",
       "      <th>Y</th>\n",
       "      <th>month</th>\n",
       "      <th>day</th>\n",
       "      <th>FFMC</th>\n",
       "      <th>DMC</th>\n",
       "      <th>DC</th>\n",
       "      <th>ISI</th>\n",
       "      <th>temp</th>\n",
       "      <th>RH</th>\n",
       "      <th>wind</th>\n",
       "      <th>rain</th>\n",
       "      <th>area</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>mar</td>\n",
       "      <td>fri</td>\n",
       "      <td>86.2</td>\n",
       "      <td>26.2</td>\n",
       "      <td>94.3</td>\n",
       "      <td>5.1</td>\n",
       "      <td>8.2</td>\n",
       "      <td>51</td>\n",
       "      <td>6.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>oct</td>\n",
       "      <td>tue</td>\n",
       "      <td>90.6</td>\n",
       "      <td>35.4</td>\n",
       "      <td>669.1</td>\n",
       "      <td>6.7</td>\n",
       "      <td>18.0</td>\n",
       "      <td>33</td>\n",
       "      <td>0.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>oct</td>\n",
       "      <td>sat</td>\n",
       "      <td>90.6</td>\n",
       "      <td>43.7</td>\n",
       "      <td>686.9</td>\n",
       "      <td>6.7</td>\n",
       "      <td>14.6</td>\n",
       "      <td>33</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>mar</td>\n",
       "      <td>fri</td>\n",
       "      <td>91.7</td>\n",
       "      <td>33.3</td>\n",
       "      <td>77.5</td>\n",
       "      <td>9.0</td>\n",
       "      <td>8.3</td>\n",
       "      <td>97</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>mar</td>\n",
       "      <td>sun</td>\n",
       "      <td>89.3</td>\n",
       "      <td>51.3</td>\n",
       "      <td>102.2</td>\n",
       "      <td>9.6</td>\n",
       "      <td>11.4</td>\n",
       "      <td>99</td>\n",
       "      <td>1.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   X  Y month  day  FFMC   DMC     DC  ISI  temp  RH  wind  rain  area\n",
       "0  7  5   mar  fri  86.2  26.2   94.3  5.1   8.2  51   6.7   0.0   0.0\n",
       "1  7  4   oct  tue  90.6  35.4  669.1  6.7  18.0  33   0.9   0.0   0.0\n",
       "2  7  4   oct  sat  90.6  43.7  686.9  6.7  14.6  33   1.3   0.0   0.0\n",
       "3  8  6   mar  fri  91.7  33.3   77.5  9.0   8.3  97   4.0   0.2   0.0\n",
       "4  8  6   mar  sun  89.3  51.3  102.2  9.6  11.4  99   1.8   0.0   0.0"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get data\n",
    "url = 'http://archive.ics.uci.edu/ml/machine-learning-databases/forest-fires/forestfires.csv'\n",
    "data = pd.read_csv(url)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Clean Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>X</th>\n",
       "      <th>Y</th>\n",
       "      <th>FFMC</th>\n",
       "      <th>DMC</th>\n",
       "      <th>DC</th>\n",
       "      <th>ISI</th>\n",
       "      <th>temp</th>\n",
       "      <th>RH</th>\n",
       "      <th>wind</th>\n",
       "      <th>rain</th>\n",
       "      <th>...</th>\n",
       "      <th>month_nov</th>\n",
       "      <th>month_oct</th>\n",
       "      <th>month_sep</th>\n",
       "      <th>day_fri</th>\n",
       "      <th>day_mon</th>\n",
       "      <th>day_sat</th>\n",
       "      <th>day_sun</th>\n",
       "      <th>day_thu</th>\n",
       "      <th>day_tue</th>\n",
       "      <th>day_wed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>86.2</td>\n",
       "      <td>26.2</td>\n",
       "      <td>94.3</td>\n",
       "      <td>5.1</td>\n",
       "      <td>8.2</td>\n",
       "      <td>51</td>\n",
       "      <td>6.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>90.6</td>\n",
       "      <td>35.4</td>\n",
       "      <td>669.1</td>\n",
       "      <td>6.7</td>\n",
       "      <td>18.0</td>\n",
       "      <td>33</td>\n",
       "      <td>0.9</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>4</td>\n",
       "      <td>90.6</td>\n",
       "      <td>43.7</td>\n",
       "      <td>686.9</td>\n",
       "      <td>6.7</td>\n",
       "      <td>14.6</td>\n",
       "      <td>33</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>91.7</td>\n",
       "      <td>33.3</td>\n",
       "      <td>77.5</td>\n",
       "      <td>9.0</td>\n",
       "      <td>8.3</td>\n",
       "      <td>97</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>89.3</td>\n",
       "      <td>51.3</td>\n",
       "      <td>102.2</td>\n",
       "      <td>9.6</td>\n",
       "      <td>11.4</td>\n",
       "      <td>99</td>\n",
       "      <td>1.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 30 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   X  Y  FFMC   DMC     DC  ISI  temp  RH  wind  rain  ...  month_nov  \\\n",
       "0  7  5  86.2  26.2   94.3  5.1   8.2  51   6.7   0.0  ...          0   \n",
       "1  7  4  90.6  35.4  669.1  6.7  18.0  33   0.9   0.0  ...          0   \n",
       "2  7  4  90.6  43.7  686.9  6.7  14.6  33   1.3   0.0  ...          0   \n",
       "3  8  6  91.7  33.3   77.5  9.0   8.3  97   4.0   0.2  ...          0   \n",
       "4  8  6  89.3  51.3  102.2  9.6  11.4  99   1.8   0.0  ...          0   \n",
       "\n",
       "   month_oct  month_sep  day_fri  day_mon  day_sat  day_sun  day_thu  day_tue  \\\n",
       "0          0          0        1        0        0        0        0        0   \n",
       "1          1          0        0        0        0        0        0        1   \n",
       "2          1          0        0        0        1        0        0        0   \n",
       "3          0          0        1        0        0        0        0        0   \n",
       "4          0          0        0        0        0        1        0        0   \n",
       "\n",
       "   day_wed  \n",
       "0        0  \n",
       "1        0  \n",
       "2        0  \n",
       "3        0  \n",
       "4        0  \n",
       "\n",
       "[5 rows x 30 columns]"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# one hot encode categorical features\n",
    "df = pd.get_dummies(data)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Consider:** We just converted all the categorical variables to [one-hot](https://en.wikipedia.org/wiki/One-hot) encoding. What should be ringing alarm bells? Is that a pun?\n",
    "\n",
    "One hot encoding features are mutually exclusive. For days of the week, only one entry per column can be 1 and all the rest are 0s. So what's the problem? For one, weeks by convention start on Sunday, and we have features from Sunday through Saturday -- meaning that the encoding for Saturday is implicitly there when $Sunday=0, Monday=0, \\dots, Friday=0$. \n",
    "\n",
    "To prevent [multicollinearity](https://en.wikipedia.org/wiki/Multicollinearity) -- e.g. Saturday can be linearly predicted from the other days of the week -- we must drop one column from month and one from day -- it does not matter which month or day we choose.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop last month in year and last day in week\n",
    "df.drop(labels=['month_dec', 'day_sat'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X               8.00\n",
      "Y               7.00\n",
      "FFMC           77.50\n",
      "DMC           290.20\n",
      "DC            852.70\n",
      "ISI            56.10\n",
      "temp           31.10\n",
      "RH             85.00\n",
      "wind            9.00\n",
      "rain            6.40\n",
      "area         1090.84\n",
      "month_apr       1.00\n",
      "month_aug       1.00\n",
      "month_feb       1.00\n",
      "month_jan       1.00\n",
      "month_jul       1.00\n",
      "month_jun       1.00\n",
      "month_mar       1.00\n",
      "month_may       1.00\n",
      "month_nov       1.00\n",
      "month_oct       1.00\n",
      "month_sep       1.00\n",
      "day_fri         1.00\n",
      "day_mon         1.00\n",
      "day_sun         1.00\n",
      "day_thu         1.00\n",
      "day_tue         1.00\n",
      "day_wed         1.00\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# check data ranges\n",
    "print(df.max() - df.min())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Note:** The range in values for some features are orders of magnitude larger than others -- e.g. `area`. Generally we want to scale our data so that the range of values in each feature is roghly the same. This is particularly important when using [regularization](https://en.wikipedia.org/wiki/Regularization_(mathematics)) or [gradient descent](https://en.wikipedia.org/wiki/Gradient_descent). Two common ways of scaling data are [normalization](https://en.wikipedia.org/wiki/Normalization_(statistics)) and [standardization](https://en.wikipedia.org/wiki/Feature_scaling)\n",
    "\n",
    "To gain insight as to what our model is telling us, we will skip data scaling for now.\n",
    "    \n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Fit Model**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None,\n",
       "         normalize=False)"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# copy data to prevent pass by reference issues\n",
    "data = df.copy()\n",
    "\n",
    "# remove dependent variable\n",
    "target = data.pop('area')\n",
    "\n",
    "# fit model, with sklearn picking best intercept\n",
    "lr = LinearRegression(fit_intercept=True)\n",
    "lr.fit(data, target)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Evaluate Model**\n",
    "\n",
    "From the discussion below on metrics for evaluation regression models, we know that for our purposes $R^2$ and $RMSE$ will provide the most relevant information regarding our model's fit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE =  62.12143311792724\n",
      "R^2 =  0.04578209650808529\n",
      "OSR^2 =  0.04578209650808529\n"
     ]
    }
   ],
   "source": [
    "# root mean squared error\n",
    "predictions = lr.predict(data)\n",
    "mse = mean_squared_error(target, predictions)\n",
    "rmse = np.sqrt(mse)\n",
    "print(\"RMSE = \", rmse)\n",
    "\n",
    "# coefficient of determination using r2 built into lr\n",
    "r2 = lr.score(data, target)\n",
    "print(\"R^2 = \", r2)\n",
    "\n",
    "# Out of Sample R^2 using r2_score method\n",
    "osr2 = r2_score(target, predictions)\n",
    "print(\"OSR^2 = \", osr2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Interpretating Coefficient of Determination** \n",
    "\n",
    "Recall that [$R^2$](https://youtu.be/2AQKmw14mHM) is a metric of correlation between variables. In the above case, $R^2$ is essentially zero \n",
    "\n",
    "$$R^2 = 0.0458 \\Rightarrow R = \\sqrt{0.0458} = 0.21400$$\n",
    "\n",
    "From $R^2$ we see that rougly 5 percent of the variation in the data is explained by the current features and such we can deduce that there is essentially no linear relationship present between the features and the dependent variable.  However, our intuition and general knowledge should tell us that temperature and seasonality appears to be correlated with the likelihood of fires. From the discussion on metrics we know that $R^2$ is sensitive to outliers, so we should check for [data imbalance](https://www.analyticsvidhya.com/blog/2020/07/10-techniques-to-deal-with-class-imbalance-in-machine-learning/) and skewedness to see if we can explain and correct for the lack of correlation. We may also want to check for the [variance inflation factors](https://en.wikipedia.org/wiki/Variance_inflation_factor), which measures the relative variance of a predictor relative to all of the predictors.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x432 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# check for missing data: white space in bar indicates missing values.\n",
    "msno.bar(data, figsize=(14,6))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.FacetGrid at 0x7f5a032d8240>"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAagAAAGoCAYAAAATsnHAAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy86wFpkAAAACXBIWXMAAAsTAAALEwEAmpwYAAAVNUlEQVR4nO3dfZBddX2A8WdJQAxEkGoIkPr6xxerg1akARENYE0GHdPEFyo7Ki0iRasVrOlMiwyGDjMImk5DW6sQBEeBZmwbNIC1MJoYXyC17Wjb+VatcUwASbFGMDEhyfaPcxYva4YskrP3e/c+n5kd7v3dvbu/k83m4Zz7u+eMjI2NIUlSNQf1ewKSJO2LgZIklWSgJEklGShJUkkz+z2ByYiImcA8YHNm7u73fCRJ3RuIQNHE6ft33nlnv+chSTrwRvY16CE+SVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklTSUAVq3rx5jIyM9OVj3rx5/d58SRooM7v6whGxAFgN/Ec79C3gcuBG4EhgMzCamTsjYgmwDDgUWJmZq7qY05YtW7jooou6+NL7tWLFir58X0kaVF3vQX05Mxe0H+8BrgKuz8yTgU3AaETMbscXAacCyyLi8I7nJUkqbqoP8S0Abm1vrwEWAicBGzNzW2ZuBzYAp03xvCRJxXR2iK/1GxFxOzAb+BAwOzN3tI89AMwFjgG29jxnfFySNMS6DNR3gD8HbgaeDXwJGOl5fAQYA3ZNeN74uCRpiHUWqMzcAnymvfv9iLgfODYiZrWH8uYC9wL3AXN6njoXuKureUmSBkNnr0FFxO9GxGXt7WcARwPXAovbT1kKrAXuBk6IiCPaxRHzgfVdzUuSNBi6XCTxeeDFEbEB+BzwLmA5cEFE3AMcBdySmbuAS4F1NGFa3vM6lSRpSHV5iO9hYMk+Hlqwj89dTfOeKUmSgCE7k4QkaXAYKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklTSzC6/eEQ8Ffg2cDlwG3AjcCSwGRjNzJ0RsQRYBhwKrMzMVV3OSZI0GLreg7oE+HF7+yrg+sw8GdgEjEbE7HZ8EXAqsCwiDu94TpKkAdBZoCLieOAFwNp2aAFwa3t7DbAQOAnYmJnbMnM7sAE4ras5SZIGR5d7UFcDF/fcn52ZO9rbDwBzgWOArT2fMz4uSRpynQQqIt4GrMvMTT3Du3pujwBjE8Z6xyVJQ66rRRKvBZ4bEUuBecBOYEdEzGoP5c0F7gXuA+b0PG8ucFdHc5IkDZBOApWZZ4/fjojLaBZFnAgsBm4CltK8NnU3cEJEHAHsAeYDF3YxJ0nSYJnK90FdAVwQEfcARwG3ZOYu4FJgHbAeWN7zOpUkaYh1+j4ogMy8rOfugn08vhpY3fU8JEmDxTNJSJJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSppJldfeGImAV8EjgaOAxYDnwduBE4EtgMjGbmzohYAiwDDgVWZuaqruYlSRoMXe5BvR7YmJmvAt4AXA1cBVyfmScDm4DRiJjdji8CTgWWRcThHc5LkjQAOtuDysybe+7Oo9ljWgD8QTu2BvhDmlBtzMxtABGxATgNuL2ruUmS6ussUOMi4hvAXOAsYH1m7mgfeqAdPwbY2vOU8XFJ0hDrfJFEZs4HlgA3A7t7HhoBxoBdE54yPi5JGmKdBSoiXhYRzwLIzG+23+tn7eIJaPaS7gXuA+b0PHV8XJI0xLrcg3o58D6AiDgamA18HljcPr4UWAvcDZwQEUe0iyPmA+s7nJckaQB0Gai/BeZGxHrgc8C7gCuACyLiHuAo4JbM3AVcCqyjCdPyntepJElDqstVfDuBc/bx0IJ9fO5qYHVXc5EkDR7PJCFJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqaRJBSoiLtnH2IcP/HQkSWo87pkkImIp8BbglRFxQs9DBwO/SXMVXEmSDrjHDVRm/n1EfBO4Bvirnof2Av/V5cQkScNtv4f4MnMTcDbwNOBZwLOB59JcgFCSpE5M9mSxXwR+QHPZ9nFeVFCS1JnJBmpPZr6l05lIktRjsoFaExFnAV+h57Ltmbm9k1lJkobeZAP1LmBkwtgY8LwDOx1JkhqTClRmGiJJ0pSaVKAiYtU+hg/KzHMP7HQkSWpM9hDfZyc850TgsAM/HUmSGpM9xLd2wtCaiPhEB/ORJAmY/CG+iW/K/TXgpQd+OpIkNSZ7iO9NPbfHgG3A+Qd+OpIkNSZ7iO/3IuJ5wIuBPcC/ZuYPO52ZJGmoTfZyGx8A/g44E1gE/GNEXNjlxCRJw22yh/iWAPMzcw9ARBwMrAP+pquJSZKG22SvqDvCY08Ou7eDuUiS9KjJ7kHdDGyMiK/SxOrlwHWdzUqSNPT2d0XdQ4APAsuBNTRX0X0h8E0eewFDSZIOqP0d4ruK5kKFB2Xmpsz8h3bsp8DlXU9OkjS89neI7+TMnN87kJk7I+JiYEN305IkDbvJLpJ4jMwc+1WfK0nSZOwvMg9GxGkTByPidcDWbqYkSdL+D/G9F/hsRPwn8G/ADGA+zYUKJ56fT5KkA+Zx96Ay87s0K/duAH4O7AI+BrzUUx1Jkrq03/dBZeZe4I72Q5KkKeFCB0lSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklGShJUkkGSpJUkoGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSQZKklSSgZIklWSgJEklzezyi0fEFcDpwMHAlcCXgRuBI4HNwGhm7oyIJcAy4FBgZWau6nJekqT6OtuDiohXAi/JzFOA1wArgKuA6zPzZGATMBoRs9vxRcCpwLKIOLyreUmSBkOXh/i+Cry5vb0NOAQ4A7i1HVsDLAROAjZm5rbM3A5sAE7rcF6SpAHQ2SG+zNwNPNzePQ+4DXh9Zu5oxx4A5gLHAFt7njo+LkkaYp0vkoiIxcD5wPuAXT0PjQBjE8Z6xyVJQ6zTQEXEQuBSYFFm/gR4KCJmtQ/PBe4F7gPm9DxtfFySNMQ6O8QXEUcAHwXOyMwH2+E7gMXATcBSYC1wN3BC+/l7gPnAhV3NS5I0GLpcZn428HTglogYH3s7cENEXAwkcEtm7o6IS4F1wF5gec/rVJKkIdXlIomPAx/fx0ML9vG5q4HVXc1FkjR4PJOEJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKmtnlF4+IFwFrgBWZeU1EzAFuBI4ENgOjmbkzIpYAy4BDgZWZuarLeUmS6utsDyoiDgNWAnf2DF8FXJ+ZJwObgNGImN2OLwJOBZZFxOFdzUuSNBi6PMS3EzgLuLdnbAFwa3t7DbAQOAnYmJnbMnM7sAE4rcN5SZIGQGeH+DJzN7A7InqHZ2fmjvb2A8Bc4Bhga8/njI9LkobYVC+S2NVzewQYmzDWOy5JGmJTHaiHImJWe3suzeG/+4A5PZ8zPi5JGmJTHag7gMXt7aXAWuBu4ISIOKJdHDEfWD/F85IkFdPZa1ARcSLwEeA5wCMR8UZgFPh0RFwMJHBLZu6OiEuBdcBeYHnP61SSpCHV5SKJf6FZtTfRL41l5mpgdVdzkSQNHs8kIUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkoyUJKkkgyUJKkkAyVJKslASZJKMlCSpJIMlCSpJAMlSSrJQEmSSjJQkqSSDJQkqSQDJUkqyUBJkkqa2e8JDIuRkRFGRkb68r2PO+44Nm/e3JfvLUm/qjKBiojLgTOAQ4ELMnNjn6d0QI2NjXHRRRf15XuvWLGiL99Xkp6MEof4IuJ04GWZeSrwduCjfZ7StDK+99aPj3nz5vV78yUNqCp7UKcDawAy89sRcWxEzMrM7e3jMwDuv//+J/VNZs6cyfbt2/f/iR3o5/eeMWMG5513Xl++93XXXefhRUmP68wzz3wOsDkzd/eOj4yNjfVnRj0i4hPAHZn52fb+14BzMvP77f1XAOv7OEVJUreem5mbegeq7EHtmnB/BOgt5z3AacB9wJ6pmpQkacr80qGWKoG6D5jTc/+ZwI/G72TmTuArUz0pSVL/lFgkAdwOLAaIiJcC/5OZO/o7JUlSP5V4DQogIq4EfhvYDZyXmd/q85QkSX1UJlBdmu7vsQKIiCtoVkMeDFwJfBm4ETiS5tjuaGbujIglwDKaP4uVmbmqPzM+cCLiqcC3gcuB2xiC7Y6Ic4D307xe+0Ga12mn9XZHxOHAp4Cn02zPh4DvAZ8AZtH8Gbw7M8ci4kLgre34n2bmbf2Z9ZMTES+iWeG8IjOviYg5TPLnHBEzgL8GXkTz92R0fOHZoKhyiK8zw/Aeq4h4JfCSzDwFeA2wArgKuD4zTwY2AaMRMbsdXwScCixrf+kH3SXAj9vb036727m/n2ZbXgf8DkOw3cC5QGbmAuANwF/QxGlZZp5E8zr26RHxfOAC4FXAQuDqiOjPaVyehIg4DFgJ3Nkz/ER+zm8D9rb/9l1BE/SBMu0DxYT3WAHHRsSs/k7pgPsq8Ob29jbgEJo9xlvbsTU0v6gnARszc1v7HrMNNKsjB1ZEHA+8AFjbDi1g+m/3QmBtZv48M+/NzPMZju3+X36xmOoo4EHg+Zn5jXZsfLtfRfO2lUcy80c0i7COn+rJHgA7gbOAe3vGFjD5n/Oj//YBX2ifO1CGIVDHAFt77m8Fju7TXDqRmbsz8+H27nk0h7kO61lo8gAwl1/+sxgfH2RXAxf33J89BNv968AzI+L2iFgfEWcwHNt9C/CsiEjgLuADwP/1PD6ttrv9vZ64WOyJ/JwfHc/MR4AZ7WG/gTEMgdrfe6ymjYhYDJwPvI/Hbvf4Nk+rP4uIeBuwbsKb+6b9dgNPofmfrNcBvw98kmZx0bjput1vBTZlZgCvpnktptd03e5eT+Tv98TxgTMMgXrc91hNFxGxELgUWJSZPwEe6jmUOZfmMMHEP4vx8UH1WuCNEfF14B00iwV2DMF23w98LTP3ZOZ3gJ8CPxuC7T6F5i0pZOa/0yyAeEbP49N1u3s9kd/rR8cj4hDgkcwcqBMdDEOgpv17rCLiCJrFH2dl5oPt8B202w0spXmN5m7ghIg4on0RdT4DfAqpzDw7M3+rfcH4WppVfJ9nmm838M/AGREx0q7qms1wbPf3gJcBRMRxwEPAxog4pX18Cc12fxFYGBEHR8SxwFGZ+d/9mHAHnsjv9e00C2ig2dv+4tRO9ckblmXm0/o9VhHxTuAyoPeX8O3ADcBhQALnZubuiHgTzaq3vcCHM/OmKZ5uJyLiMppVTV8AbmKab3f7Mz+HJk4folliPa23u/3H9waavaZDgD+j2Zu8nuasOF/KzPe3n/temsOfe4E/zsy7+jLpJyEiTgQ+AjwHeATYAowCn2YSP+f29aZVwAuB7TTnNx2oMzcPRaAkSYNnGA7xSZIGkIGSJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKCkIiLC30eph++DkqZQRDwN+AxwOPBU4D00b7C9lebEp9fSXELiUJo3lr8jM38YEVfTnOrnKcDHMvPaPkxfmlL+H5s0tY4GPtVe0+hP2o8ZNJeHWA4sBz6amWcCfwlcEhFPoTmv2itoLqMwcNf1kX4VM/s9AWnIPAi8PiLeTbMH9bN2/J72vycBx0fEB2nC9UB7xdSjaM6vtpvHnhhUmrYMlDS1/gjYkpmjETEfuLId7700wtmZuWX8TntV6AXtxx6ak6RK056H+KSpdRTw3fb2m2hOetrrG/zi7PtnRMRb2uf8oL3o3FLgoPbyCdK05iIJaQq1Z6j+FPBD4BpgBfA84GmZ+XB7eYhP0iySGAPOpTks+E80Z6ReS3PJiYcy851TPX9pKhkoSVJJHuKTJJVkoCRJJRkoSVJJBkqSVJKBkiSVZKAkSSUZKElSSf8PG8gftXDbH+4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualize dependent variable\n",
    "sns.displot(target, height=6, binwidth=100, color='#424242')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Discussion:** As you can see, the response variable with highly skewed -- meanwhile [Least Squares Regression](https://en.wikipedia.org/wiki/Least_squares) assumes a normal distribution. Two straightforward ways for dealing with the imbalance are [resampling](https://machinelearningmastery.com/statistical-sampling-and-resampling/) and model selection. Sometimes, as an initial step, we can play with the training/test split to deal skewdness. This is because:\n",
    "\n",
    ">* A **larger test set** gives a more reliable estimate of accuracy -- i.e. higher bias, lower variance.\n",
    ">* A **larger training set** will lead to a more representative model -- but with poor generaliztions, i.e. higher out-of-sample variance.\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### BASELINE MODEL\n",
    "\n",
    "<br>\n",
    "\n",
    "**Playing with Train/Test Split**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# 50/50 split used to establish baseline\n",
    "X_train, X_test, y_train, y_test = train_test_split(data, target, shuffle=True,\n",
    "                                                    test_size=0.5, random_state=290)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Fit Model on Training Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None,\n",
       "         normalize=False)"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr_2 = LinearRegression(fit_intercept=True)\n",
    "lr_2.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Check in/out-of-sample error**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "def inSampleError(X_train, y_train, model):\n",
    "    '''\n",
    "        Computes in-sample-R^2 and RMSE. \n",
    "        @param X_train <type:DataFrame>\n",
    "        @param y_train <type:DataFrame>\n",
    "        @model <type:sklearn.linear_model>\n",
    "        @returns R2 <type:float> \n",
    "        @returns rmse <type:float>\n",
    "    '''\n",
    "    predictions = model.predict(X_train)\n",
    "    mse = mean_squared_error(y_train, predictions)\n",
    "    rmse = np.sqrt(mse)\n",
    "    return(model.score(X_train, y_train), rmse)\n",
    "    \n",
    "def outSampleError(X_test, y_test, model):\n",
    "    '''\n",
    "        Computes out-of-sample-R^2 and RMSE. \n",
    "        @param X_test <type:DataFrame>\n",
    "        @param y_test <type:DataFrame>\n",
    "        @model <type:sklearn.linear_model>\n",
    "        @returns R2 <type:float> \n",
    "        @returns rmse <type:float>\n",
    "    '''\n",
    "    predictions = model.predict(X_test)\n",
    "    mse = mean_squared_error(y_test, predictions)\n",
    "    rmse = np.sqrt(mse)\n",
    "    return(model.score(X_test, y_test), rmse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "# calculate in-sample error metrics\n",
    "is_r2, is_error = inSampleError(X_train, y_train, lr_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# calculate out-of-sample error metrics\n",
    "os_r2, os_error = outSampleError(X_test, y_test, lr_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "R^2_in =  0.07625005249125671\n",
      "R^2_out =  -0.1210187605707067 \n",
      "\n",
      "rmse_in =  69.32739834458923\n",
      "rmse_out =  56.91710873036987\n"
     ]
    }
   ],
   "source": [
    "print(\"R^2_in = \", is_r2)\n",
    "print(\"R^2_out = \", os_r2, \"\\n\")\n",
    "print(\"rmse_in = \", is_error)\n",
    "print(\"rmse_out = \", os_error)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Discussion:** Notice that $R^2_{in}$ is still pretty low. However, now we see that $R^2_{out}$ is lower that $R^2_{in}$ and also negative. Meanwhile, we see that $RMSE_{out}$ is higher that $RMSE_{in}$, which tells us that our model is [overfitting](https://elitedatascience.com/overfitting-in-machine-learning) the data -- keep in mind that the greater the gap between $RMSE_{in}$ and $RMSE_{out}$, the more the model overfits/underfits.  \n",
    "\n",
    "Since we know our model is overfitting, we can test different splits or use resampling to see if we can improve the model. Alternatively, we can leverage [cross-validation](https://en.wikipedia.org/wiki/Cross-validation_(statistics)) to find the penalty for some desireable [regularization](https://en.wikipedia.org/wiki/Regularization_(mathematics)) procedure(s) and thus find the best [tradeoff between bias and variance](https://en.wikipedia.org/wiki/Bias%E2%80%93variance_tradeoff). More on those topics below.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **MODEL** TUNING"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "# common 80/20 split on the data (train/test)\n",
    "X_train, X_test, y_train, y_test = train_test_split(data, target, shuffle=True,\n",
    "                                                    test_size=0.2, random_state=290)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Helper Function(s)**\n",
    "\n",
    "Detailed description and source of helper functions can be found [here](https://dziganto.github.io/cross-validation/data%20science/machine%20learning/model%20tuning/python/Model-Tuning-with-Validation-and-Cross-Validation/).\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "def trainingError(X_train, y_train, model):\n",
    "    '''returns in-sample root mean squared error for already fit model.'''\n",
    "    predictions = model.predict(X_train)\n",
    "    mse = mean_squared_error(y_train, predictions)\n",
    "    rmse = np.sqrt(mse)\n",
    "    return(rmse)\n",
    "    \n",
    "def validationError(X_test, y_test, model):\n",
    "    '''returns out-of-sample root mean squared error for already fit model.'''\n",
    "    predictions = model.predict(X_test)\n",
    "    mse = mean_squared_error(y_test, predictions)\n",
    "    rmse = np.sqrt(mse)\n",
    "    return(rmse)\n",
    "\n",
    "def errorMetrics(X_train, y_train, X_test, y_test, model):\n",
    "    '''fits model and returns the RMSE for in-sample error and out-of-sample error'''\n",
    "    model.fit(X_train, y_train)\n",
    "    train_error = trainingError(X_train, y_train, model)\n",
    "    validation_error = validationError(X_test, y_test, model)\n",
    "    return (train_error, validation_error)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "#### **BIAS-VARIANCE** TRADEOFF\n",
    "\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"bias_v_variance.png\" align=\"center\" width=\"40%\" padding=\"10px\"><br>\n",
    "    <br>\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    "We now take some time to highlight some of the key concepts in [bias](https://en.wikipedia.org/wiki/Bias_(statistics)), [variance](https://en.wikipedia.org/wiki/Variance), and [reducible/irreducible error](https://daviddalpiaz.github.io/r4sl/biasvariance-tradeoff.html).\n",
    "\n",
    "<br>\n",
    "\n",
    "Consider the previous functions. `errorMetrics( )` returns `train_error` and `validation_error`. Each error can be broken down into three components. Namely,\n",
    "\n",
    "<br>\n",
    "\n",
    "$$\\displaystyle{Total~Error~(model) = Variance~(model) + Bias~(model) + Variance~(irreducible ~error)}$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**BIAS** is a measure of how closely the functional form of the model captures the true relationship (i.e. the mapping function) between the independent (input) and dependent (output) variables. A model with high bias, which is **always positive**, is beneficial when the model matches the true underlying mapping function. However, large bias translates to poor [generalization](https://www.cs.toronto.edu/~rgrosse/courses/csc321_2018/readings/L09%20Generalization.pdf) of the model. Following are some ways to detect and deal with bias. \n",
    "\n",
    ">* **High Bias (source of error)** - will have a high training error. Validation error is similar in magnitude to training error.\n",
    ">> **Solution:** increasing the number of features, changing the model to one that can more closely represent the complexity of the relationship, and resampling techiniques -- such as [cross validation](https://en.wikipedia.org/wiki/Cross-validation_(statistics)) and [bootstrapping](https://en.wikipedia.org/wiki/Bootstrapping_(statistics)) -- are all approaches for reducing bias.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**VARIANCE** is the amount by which the model's performace changes when it is estimated (fit) on a different training data set. Variance is also always a **positive value**. Following are some ways to detect and deal with variance.\n",
    "\n",
    ">* **High Variance (source of error)** - will have a low training error, and very high validation error.\n",
    ">> **Solution:** [feature selection](https://en.wikipedia.org/wiki/Feature_selection), and [regularization](https://statweb.stanford.edu/~tibs/sta305files/Rudyregularization.pdf) are two common approaches.\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**ERROR** consists of both reducible and irreducible error(s) -- also known as [noise](https://people.eecs.berkeley.edu/~jrs/189/lec/12.pdf). The **reducible error** is a quantity that can be modulated during learning. Meanwhile, [irreducible error (Bayes Error Rate)](https://en.wikipedia.org/wiki/Bayes_error_rate) is the error caused by elements outside of modulation -- such as noise in the observations. Irreducible error serves as an, almost always, unknown bound on the accuracy of our prediction for the dependent variable(s). \n",
    "\n",
    "\n",
    ">* **Reducible Error (source of error)** - reducible error can be decomposed into **error due to squared bias** and **error due to variance**. \n",
    ">> **Solution 1:** since the portion of the reducible error due to squared bias has to do with the difference between predicted and true target values, techniques for dealing with high bias, such as [resampling](http://www.columbia.edu/~mh2078/MachineLearningORFE/ResamplingMethods.pdf), can influence this type of error while also decreasing model underfit.<br>\n",
    ">> **Solution 2:** on the other hand, the error due to variance is the amount by which the models predictions differ from one training set to the expected predicted values over all training sets, techniques for dealing with high variance -- such as [regularization](https://statweb.stanford.edu/~tibs/sta305files/Rudyregularization.pdf) -- can be used to reduce this error while also redusing overfit.\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **BIAS-VARIANCE** DECOMPOSITION\n",
    "\n",
    "\n",
    "To gain a deeper understanding of what sort of insights we can gain from total error, let us decompose the bias-variance relationship.\n",
    "\n",
    "From the above explanation we have that there are three sources of error. To derrive total error, let the inability of a hypothesis $h$ to fit $g$ be defined as the bias. Variance is then the error due to fitting noise in the data. Let our model then be\n",
    "\n",
    "$$\\displaystyle{X_i \\sim D, ~ \\epsilon_i \\sim D^{\\prime}, ~ y_i = g(X_i) + \\epsilon_i}$$\n",
    "\n",
    "where $h$ is a random variable, $\\epsilon_i \\sim D^{\\prime}$ has mean zero, and the sample points $X_i \\sim D$ come from an unknown probability distribution. \n",
    "\n",
    "We want to fit our hypothesis $h$ to $X,y$. Consider an arbitrary point $z \\in \\mathbb{R}^d$ (not necessarly a sample point) and suppose <br>\n",
    "\n",
    "$$\\displaystyle{\\gamma = g(z) + \\epsilon, ~~~~ \\epsilon \\in D^{\\prime}}$$\n",
    "\n",
    "For an arbitrary $z$ and $\\gamma$ random, the mean comes from $g$ and variance from $\\epsilon$ since <br>\n",
    "\n",
    "$$\\displaystyle{E[\\gamma] = g(z);~~~~Var(\\gamma) = Var(\\epsilon)}$$\n",
    "\n",
    "Then for a risk function $R$ we have that when loss, $L$ equal to the square error\n",
    "\n",
    "$$\\displaystyle{ R(h) = D[L(h(z), \\gamma)]}$$\n",
    "\n",
    "we have form some $h$, a random variable, when we integrage the loss over all possible values for weights, subject to the training data\n",
    "\n",
    "$$\\displaystyle{R(h) = E[(h(z) - \\gamma)^2] = E[h(z)^2] + E[\\gamma ^2] - 2E[\\gamma ~h(z)]}$$\n",
    "\n",
    "Note that $\\gamma$ and $h(z)$ are independent. Then\n",
    "\n",
    "$$\\displaystyle{R(h) = Var(h(z)) + E[h(z)]^2 + Var(\\gamma) + E[\\gamma]^2 -2E[\\gamma]E[h(z)] }$$\n",
    "\n",
    "and thus \n",
    "\n",
    "$$\\displaystyle{R(h) = \\left(E[h(z)]-E[\\gamma] \\right)^2 + Var(h(z)) + Var(\\gamma)  = \\left(E[h(z)] - g(z)\\right)^2 + Var(h(z)) + Var(\\epsilon) }$$\n",
    "\n",
    "<br>\n",
    "\n",
    "The above is known as the *bias-variance decomposition of the risk function*\n",
    "\n",
    "$$\\displaystyle{R(h) = Bias^2_{method} + Var_{method} + Var_{irreducible~error}}$$\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "<br>\n",
    "\n",
    "<br>\n",
    "\n",
    "To drive the point home, consider the following illistration of a bias-variance decomposition:\n",
    "<br>\n",
    "\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"bias-variance_decomposition.png\" align=\"center\" width=\"50%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "    50 fits of $g$ (20 samples each)\n",
    "</div>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **CONSEQUENCES** OF **BIAS-VARIANCE** DECOMPOSITION\n",
    "\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"validation_decision_curves.png\" align=\"center\" width=\"90%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "\n",
    ">* Ideally we want a model that has low bias and low variance -- the point where total error is minimized. This can be a very challenging task.<br>\n",
    ">* Underfitting $\\Leftrightarrow$ high bias<br>\n",
    ">* Overfitting $\\Leftrightarrow$ high variance.<br>\n",
    ">* Training error reflects bias but not variance.<br>\n",
    ">* Test error reflects both bias and variance. A reason why low training error can be obfuscate overfitting.<br>\n",
    ">* For many distributions, $Variance \\rightarrow 0$ as $n \\rightarrow \\infty$.<br>\n",
    ">* If $h$ can fit $g$ exactly, then for many distributions $Bias \\rightarrow 0$ as $n\\rightarrow \\infty$.<br>\n",
    ">* If $h$ fits $g$ poorly, then bias is large at \"most\" points.<br> \n",
    ">* Adding a good feature reduces bias; addiging a bad feature rarely increases it.<br>\n",
    ">* Adding a feature usually increases variance. As such, only add a feature if it reduces bias more than it increases variance.<br>\n",
    ">* Irreducible error cannot be reduced (mind blown).<br>\n",
    ">* Noise in a test set only affects $Var(\\epsilon)$.<br>\n",
    ">* Noise in a training set affects $Bias$ and $Var$  but not the error, $Var(\\epsilon)$.<br>\n",
    ">* Simple models such as linear and logistic regression, generally have a high bias and low variance.<br>\n",
    ">* Complex models, such as random forrests, tend to have low bias but high variance.<br>\n",
    ">* High bias/variance is not necesseraly a bad thing. Careful consideration should be given to the problem being addressed.<br>\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ORIGINAL ERROR\n",
      "--------------------------------------------------\n",
      "training error: 66.783 | testing error: 40.049\n",
      "test/train: 60.0%\n"
     ]
    }
   ],
   "source": [
    "lr_3 = LinearRegression(fit_intercept=True)\n",
    "\n",
    "# calculate error metrics\n",
    "train_error, test_error = errorMetrics(X_train, y_train, X_test, y_test, lr_3)\n",
    "\n",
    "# training and testing error to three significant figures\n",
    "print('ORIGINAL ERROR')\n",
    "print('-' * 50)\n",
    "print('training error: {:.3f} | testing error: {:.3f}'.format(float(train_error), float(test_error)))\n",
    "\n",
    "# test/train ratio as a percentage\n",
    "print('test/train: {:.1%}'.format(test_error/train_error))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Note:** if $\\frac{test}{train} \\le 100\\%$ we may be underfitting, while $\\frac{test}{train} > 100\\%$ may indicate overfitting.<br>\n",
    "\n",
    "<br>\n",
    "\n",
    "> **Underfitting** occurs when the training and validation errors are both substantial, but there also exists a small gap between them. If the model cannot reduce the training error, then the model may be too simple to capture the underlying distribution.<br>\n",
    "> **Overfitting** occurs when our model follows the training data more closely than the underlying, actual, distribution.  [Regularization](https://statweb.stanford.edu/~tibs/sta305files/Rudyregularization.pdf) is a collection of techniques used to combat [overfitting](https://en.wikipedia.org/wiki/Overfitting).<br>\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **MODEL** TUNING (CONTINUED)\n",
    "\n",
    "In the previous comparison between testing and training errors, we appear to be underfitting. However, the model was trained using a test/train split. To see how splitting the data can influence we now turn to a **Training | Validation | Testing** data split approach. However, it should be noted that there are a few drawbacks to splitting a data set in such a way. Namely:\n",
    "\n",
    "1. Data is generally scarce, and splitting the data further can lead to detrimental expectations.\n",
    "2. The performance on the validation set may be highly variable, dpending on how the data is split.\n",
    "3. The error in the validation set tends to over-estimate the test-error rate of the model fitted to the entire data set.\n",
    "\n",
    "Despite the aforementioned drawback, we now perform such as split to establish a baseline prior to discussing regularization and shrinkage methods."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training set: 0.6% | validation set: 0.2% | testing set 0.2%\n"
     ]
    }
   ],
   "source": [
    "# now we want to split the same data into training, testing, validation \n",
    "# This is done so that we have 'unseen' data to use when we compute out-of-sample error\n",
    "# we do this in three steps:\n",
    " \n",
    "# (1) split to original data into train/test (80/20)\n",
    "X_intermediate, X_test, y_intermediate, y_test = train_test_split(data,\n",
    "                                                                  target,\n",
    "                                                                  shuffle=True,\n",
    "                                                                  test_size=0.2,\n",
    "                                                                  random_state=290)\n",
    "\n",
    "# (2) then split train again into train/validation (75/25)\n",
    "X_train, X_validation, y_train, y_validation = train_test_split(X_intermediate,\n",
    "                                                                y_intermediate,\n",
    "                                                                shuffle=False,\n",
    "                                                                test_size=0.25,\n",
    "                                                                random_state=290)\n",
    "\n",
    "# (3) delete intermediate variables (optional)\n",
    "#del X_intermediate, y_intermediate\n",
    "\n",
    "# proportions\n",
    "print('training set: {}% | validation set: {}% | testing set {}%'.format(round(len(y_train)/len(target),2),\n",
    "                                                                         round(len(y_validation)/len(target),2),\n",
    "                                                                         round(len(y_test)/len(target),2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ORIGINAL ERROR\n",
      "--------------------------------------------------\n",
      "training error: 45.877 | validation error: 109.485\n",
      "validation/train: 87.3%\n"
     ]
    }
   ],
   "source": [
    "# calculate error metrics on train/validate\n",
    "train_error, validation_error = errorMetrics(X_train, y_train, X_validation, y_validation, lr_3)\n",
    "\n",
    "# training and testing error to three significant figures\n",
    "print('ORIGINAL ERROR')\n",
    "print('-' * 50)\n",
    "print('training error: {:.3f} | validation error: {:.3f}'.format(float(train_error), float(validation_error)))\n",
    "\n",
    "# test/train ratio as a percentage\n",
    "print('validation/train: {:.1%}'.format(test_error/train_error))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Discussion:** \n",
    "\n",
    "After reducing the amount of data available for training our model appears to be overfitting. Recall that in our new data split approach our initial data split yielded the training/test sets, then the resulting training data was subsequently split into the training/validation sets. Since we appear to be overfitting, there are several approaches we can take to reduce our validation error, and thus reduce variance and in turn decrease the overfit. One approach could be to reduce the number of features, but this task should be generally avoided as a first option or dismissed altogether -- for more details see the optional section on p-valies and variance inflation factors (vif) at the end of this notebook. \n",
    "\n",
    "Consider the following diagram:\n",
    "\n",
    "<br>\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"modulating_bias-variance.png\" align=\"center\" width=\"40%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "Since we cannont gather more data (total amount as opposed to split amounts), and we are ignoring the removing of features, we now turn our attention toward **resampling** and **regularization** methods. \n",
    "\n",
    "\n",
    "**Note:** techniques such as [subset selection](https://en.wikipedia.org/wiki/Feature_selection#Subset_selection), [Bayesian Information Criterion](https://en.wikipedia.org/wiki/Bayesian_information_criterion) and [Akaike Information Criterion](https://en.wikipedia.org/wiki/Akaike_information_criterion) can be used to deal with overfitting, but are not covered in this notebook.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### **RESAMPLING** METHODS\n",
    "\n",
    "<br>\n",
    "\n",
    "Resampling methods involve the process of (1) repeatedly drawing samples from the training data, (2) refitting the model of interest with each new sample, and (3) examining all of the fitted models to draw conclusions.  There are several approaches to resampling, however on this notebook we will focus on **cross validation**. For a comprehensive list of resampling methods with associated python examples, see [here](https://people.duke.edu/~ccc14/sta-663/ResamplingAndMonteCarloSimulations.html)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**CROSS VALIDATION** (AND HYPER PARAMETER TUNNING)\n",
    "\n",
    "To first build some intuition, consider the original data split (train/test) we performed at the start of this notebook. We took the data set and essentially performed a **2-fold cross-validation (hold-out-method)** when we split it into $training ~|~ validation ~|~ testing$ sets. A disadvantage to such approach is that the model's performance strongly depends on how the data is split, as well as the random seed used."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Why Cross-Validation**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 80/20 split - data drawn sequentially\n",
    "X_train_1a, X_validation_1a, y_train_1a, y_validation_1a = train_test_split(X_intermediate,\n",
    "                                                                            y_intermediate,\n",
    "                                                                            shuffle=False,\n",
    "                                                                            test_size=0.20,\n",
    "                                                                            random_state=290)\n",
    "\n",
    "# 80/20 split - data suffled prior to drawing \n",
    "X_train_1b, X_validation_1b, y_train_1b, y_validation_1b = train_test_split(X_intermediate,\n",
    "                                                                            y_intermediate,\n",
    "                                                                            shuffle=True,\n",
    "                                                                            test_size=0.20,\n",
    "                                                                            random_state=290)\n",
    "\n",
    "# 85/15 split - data drawn sequentially\n",
    "X_train_2a, X_validation_2a, y_train_2a, y_validation_2a = train_test_split(X_intermediate,\n",
    "                                                                            y_intermediate,\n",
    "                                                                            shuffle=False,\n",
    "                                                                            test_size=0.15,\n",
    "                                                                            random_state=290)\n",
    "\n",
    "# 90/10 split - data suffled prior to drawing \n",
    "X_train_3a, X_validation_3a, y_train_3a, y_validation_3a = train_test_split(X_intermediate,\n",
    "                                                                            y_intermediate,\n",
    "                                                                            shuffle=True,\n",
    "                                                                            test_size=0.10,\n",
    "                                                                            random_state=290)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "ERROR SENSITIVITY TO TRAIN|VALIDATE SPLIT\n",
      "----------------------------------------------------------------------\n",
      "80/20 split - data drawn sequentially: \n",
      "\t    training error: 72.789 | validation error: 37.049\n",
      "\n",
      "80/20 split - data suffled prior to drawing:  \n",
      "\t    training error: 73.903 | validation error: 24.114\n",
      "\n",
      "85/15 split - data drawn sequentially: \n",
      "\t    training error: 71.389 | validation error: 32.677\n",
      "\n",
      "90/10 split - data suffled prior to drawing: \n",
      "\t    training error: 70.127 | validation error: 21.160\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# calculate error metrics on train/validate\n",
    "train_error_1a, validation_error_1a = errorMetrics(X_train_1a, y_train_1a, X_validation_1a, y_validation_1a, lr_3)\n",
    "train_error_1b, validation_error_1b = errorMetrics(X_train_1b, y_train_1b, X_validation_1b, y_validation_1b, lr_3)\n",
    "train_error_2a, validation_error_2a = errorMetrics(X_train_2a, y_train_2a, X_validation_2a, y_validation_2a, lr_3)\n",
    "train_error_3a, validation_error_3a = errorMetrics(X_train_3a, y_train_3a, X_validation_3a, y_validation_3a, lr_3)\n",
    "\n",
    "# training and testing error to three significant figures\n",
    "print('\\n\\nERROR SENSITIVITY TO TRAIN|VALIDATE SPLIT')\n",
    "print('-' * 70)\n",
    "print('80/20 split - data drawn sequentially: \\n\\t    training error: {:.3f} | validation error: {:.3f}\\n'.format(float(train_error_1a), float(validation_error_1a)))\n",
    "print('80/20 split - data suffled prior to drawing:  \\n\\t    training error: {:.3f} | validation error: {:.3f}\\n'.format(float(train_error_1b), float(validation_error_1b)))\n",
    "print('85/15 split - data drawn sequentially: \\n\\t    training error: {:.3f} | validation error: {:.3f}\\n'.format(float(train_error_2a), float(validation_error_2a)))\n",
    "print('90/10 split - data suffled prior to drawing: \\n\\t    training error: {:.3f} | validation error: {:.3f}\\n'.format(float(train_error_3a), float(validation_error_3a)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Discussion:**\n",
    "\n",
    "Note the disparities in validation error in response to 5 percent change in the amount of available data for the training set. Cross-validation gives a way to deal with such disparities in our model's fit brough about by how we perform our data splits. Cross-validation is also used for:\n",
    "1. Picking which variables to include in a model (**feature selection**).\n",
    "2. Picking the prediction function to use based on error (**model assessment**).\n",
    "3. Picicking the parameters in the prediction function (e.g. hyperparameter tunning).\n",
    "4. Compare different predictors (**model selection**).\n",
    "\n",
    "While [Bootstrapping & Permutation Resampling](https://statweb.stanford.edu/~tibs/stat315a/Supplements/bootstrap.pdf) techniques have several benefits, for brevity we will focus entireley on cross-validation in this notebook -- in particular, **K-fold Cross-validation**. For a comprehensive explanation of how resampling affects accuracy and model selection, see [this](http://ai.stanford.edu/~ronnyk/accEst.pdf) research paper comparing different approaches.\n",
    "\n",
    "<br>\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"5-fold_cross-validation.png\" align=\"center\" width=\"40%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "    Example Process for Fitting Model Using 5-Fold Cross-Validation\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Computing Cross-Validation Scores Using $R^2$**\n",
    "\n",
    "Prior to examing cross-validation with regularization, we first want to establish a baseline. To understand what is going on in the background, we will first compute the cross-validation scores using [`cross_val_score`](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.cross_val_score.html) and the using [`KFold`](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.KFold.html) from the ScikitLearn library. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "CV (cross_val_score) scores: \n",
      "\t [-1.69652627e-02 -3.65222361e-01 -4.99759126e-01  6.53620680e-04\n",
      " -7.94678968e+00]\n",
      "\n",
      "CV (Kfold) scores: \n",
      "\t [-0.01104697 -0.95436741 -0.43480365 -0.28099057 -0.50994686]\n"
     ]
    }
   ],
   "source": [
    "# k-fold cross-validation\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.linear_model import LinearRegression\n",
    "\n",
    "# 80/20 split on the data (train/test) with suffling\n",
    "X_train, X_test, y_train, y_test = train_test_split(data, target, shuffle=True,\n",
    "                                                    test_size=0.2, random_state=135)\n",
    "\n",
    "# new linear model for clarity\n",
    "lm_4 = LinearRegression(fit_intercept=True)\n",
    "\n",
    "# 5-fold cross-validation with R^2 scores using cross_val_score\n",
    "scores_cvs = cross_val_score(lm_4, X_train, y_train, scoring='r2', cv=5)\n",
    "print(\"\\nCV (cross_val_score) scores: \\n\\t\", scores_cvs)\n",
    "\n",
    "# 5-fold cross-validation with R^2 scores using KFold\n",
    "folds = KFold(n_splits = 5, shuffle = True, random_state = 135)\n",
    "scores_KF = cross_val_score(lm_4, X_train, y_train, scoring='r2', cv=folds)\n",
    "print(\"\\nCV (Kfold) scores: \\n\\t\", scores_KF)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Note:** [work by Ron Kohave](http://ai.stanford.edu/~ronnyk/accEst.pdf) showed that for practical data (not skewed, inballanced, etc.), **10-fold corss-validation consistently leads to the best compromise between bias and variance** in most cases. \n",
    "    \n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Compute Best CV Score Using $-MSE$ on 10 CV fold**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "metadata": {},
   "outputs": [],
   "source": [
    "# iterate over various thresholds\n",
    "alphas = [1e-4, 1e-3, 1e-2, 1e-1, 1, 1e1]\n",
    "\n",
    "scores_neg_MSE = -cross_val_score(lm_4, X_train, y_train, scoring='neg_mean_squared_error', cv=10)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 1607.24767594 16096.82918195  1208.53995124   806.0764834\n",
      "   703.07870949  1257.99430898   479.88167429 28184.91198337\n",
      "   470.78656901   609.42573622]\n",
      "\n",
      " 470.78656900933606\n"
     ]
    }
   ],
   "source": [
    "print(scores_neg_MSE)\n",
    "print(\"\\n\", min(scores_neg_MSE))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "___\n",
    "\n",
    "\n",
    "#### **LEARNING** CURVES\n",
    "\n",
    "<table style=\"width:80%\">\n",
    "  <!-- left column -->\n",
    "  <th>\n",
    "      <div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "          <img src=\"validation_curve.png\" align=\"center\" width=\"70%\" padding=\"0px\"><br>\n",
    "      </div>\n",
    "  </th>\n",
    "  <!-- right column -->  \n",
    "  <th>\n",
    "      <div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "          <img src=\"learning_curve.png\" align=\"center\" width=\"70%\" padding=\"0px\"><br>\n",
    "      </div>\n",
    "  </th>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "We are starting to glimpse the power of cross-validation, in particular when it comes to hyperparameter optimization, and model selection.  We now take a brief aside to discuss the [diagnostic power of learning curves](https://machinelearningmastery.com/learning-curves-for-diagnosing-machine-learning-model-performance/). Despite the image, it should be noted that will not be covering validation curves (e.g. F-Measure Curve, ROC, etc.) in this notebook as they are most commonly used for classification. You should still make note of it, as it may be needed for your homework problems."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Learning Curves with High Bias**\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "  <img src=\"learning_curve_bias.png\" align=\"center\" width=\"20%\" padding=\"0px\"><br>\n",
    "</div>\n",
    "\n",
    "      \n",
    ">* Getting more training data may not, by itself, help much.<br>\n",
    ">* Learning curves will be **characterized by a small gap between curves, but both trending well above or bellow the desired accuracy**.<br>\n",
    ">* Will lead to **model underfitting**.<br>\n",
    ">* May be indicative of [systemic error](https://sciencing.com/difference-between-systematic-random-errors-8254711.html).<br>\n",
    ">* Adding polynomial features ($x_1, x_2, x_1^2, x_2^2, \\dots$) will reduce bias/underfit.<br>\n",
    ">* Adding additional features will reduce bias/underfit.<br>\n",
    ">* Loosening regularization will reduce bias/underfit."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Learning Curves with High Variance**\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "  <img src=\"learning_curve_variance.png\" align=\"center\" width=\"20%\" padding=\"0px\"><br>\n",
    "</div>\n",
    "\n",
    "      \n",
    ">* Getting more training data will likely help.<br>\n",
    ">* Learning curves will be **characterized by a wide gap between the two errors**.<br>\n",
    ">* Will lead to **model overfitting**.<br>\n",
    ">* Removing features will reduce variance/overfit.<br>\n",
    ">* Increasing regularization will reduce variance/overfit.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Visualizing Learning Curves**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import learning_curve\n",
    "\n",
    "# Create CV training and test scores for various training set sizes\n",
    "train_sizes, train_scores, test_scores  = learning_curve(lm_4, \n",
    "                                                         data, \n",
    "                                                         target, \n",
    "                                                         # Number of folds in cross-validation\n",
    "                                                         cv=10,\n",
    "                                                         # Evaluation metric\n",
    "                                                         scoring='neg_mean_absolute_error',\n",
    "                                                         # Use all computer cores\n",
    "                                                         n_jobs=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create mean and standard deviations of training errors\n",
    "train_mean = np.mean(train_scores, axis=1)\n",
    "train_std = np.std(train_scores, axis=1)\n",
    "\n",
    "# Create mean and standard deviations of test errors\n",
    "test_mean = np.mean(test_scores, axis=1)\n",
    "test_std = np.std(test_scores, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 320,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#create figure and axes\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "\n",
    "# Draw artists (lines)\n",
    "ax.plot(train_sizes, train_mean, color=\"#2F5E85\",  label=\"Training score\")\n",
    "ax.plot(train_sizes, test_mean, color=\"#F9634B\", label=\"Cross-validation score\")\n",
    "\n",
    "# draw accuracy bands\n",
    "ax.fill_between(train_sizes, train_mean - train_std, train_mean + train_std, color=\"gray\")\n",
    "ax.fill_between(train_sizes, test_mean - test_std, test_mean + test_std, color=\"gainsboro\")\n",
    "\n",
    "# header\n",
    "ax.set_title(\"LEARNING CURVES\", size=18, color=\"#434343\")\n",
    "# create artist labels\n",
    "ax.set_xlabel(r'Training Data Size', size=16, color=\"#434343\")\n",
    "ax.set_ylabel(r'Model Score ($-MAE$)', size=16, color=\"#434343\")\n",
    "\n",
    "# legend\n",
    "plt.legend(loc=\"best\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "___\n",
    "\n",
    "\n",
    "#### **SHRINKAGE** METHODS -- **REGRESSION**\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"l1_l2_contour_simple.png\" align=\"center\" width=\"70%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "</div>\n",
    "\n",
    "\n",
    "[Regularization](https://en.wikipedia.org/wiki/Regularization_(mathematics)) is the purposeful modulation of the bias estimator in an effort to reduce the total error. That is, both L1 and L2 regularization penalize features that contribute little to the objective function's output -- this is known as [shrinkage](https://datacadamia.com/data_mining/shrinkage). There are several ways of performing shrinkage but this notebook will focus exclusively on [Lasso (L1)](https://aswani.ieor.berkeley.edu/teaching/SP15/265/lecture_notes/ieor265_lec6.pdf) and [Ridge (L2)](https://aswani.ieor.berkeley.edu/teaching/SP17/265/lecture_notes/ieor265_lec5.pdf) regression. For a quick reminder of what Lasso regression is, see [here](https://youtu.be/NGf0voTMlcs). For a quick reminder of Ridge regression is, see [here](https://youtu.be/Q81RR3yKn30). For a math heavy explanation of regression, see [here](https://ocw.mit.edu/courses/mathematics/18-650-statistics-for-applications-fall-2016/lecture-videos/). \n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**LASSO** AND **RIDGE** REGRESSION"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Suppose that we have a independent paris of measurements $(x_i, y_i)$ for $i=1, \\dots, n$, where $x_i\\in \\mathbb{R}^p$ and $y_i \\in \\mathbb{R}$. If our linear model with *noiseles* measurements is discribed by the following model:\n",
    "\n",
    "$$\\displaystyle{y_i = \\sum_{j=1}^p \\beta_j x_i^j + \\epsilon_i = x_i^{\\prime} \\beta}$$\n",
    "\n",
    "In the case that $p>n$, relatively speaking, then [ordinary least squares regression (OLS)](https://en.wikipedia.org/wiki/Ordinary_least_squares) will have a high estimation error. Overfitting can be dealth with using both [convex and non-convex](https://jmlr.csail.mit.edu/papers/volume15/aravkin14a/aravkin14a.pdf) approaches, under the assumption that the measurements, $y_i$, are noiseless. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "Now suppose we have noisy measurements, and $\\epsilon_i$ are iid, zero-mean, and bounded random variables. Then\n",
    "\n",
    "$$\\displaystyle{y_i = \\sum_{j=1}^p \\beta_j x_i^j + \\epsilon_i}$$\n",
    "\n",
    "A common way of estimating parameters in such as setting is given by:\n",
    "\n",
    "$$\\displaystyle{\\hat{\\beta} = \\arg \\min_{\\beta} \\{||Y -X\\beta ||_2^2 ~~:~~ ||\\beta||_1 \\le \\mu \\}}$$\n",
    "\n",
    "<br>\n",
    "\n",
    "This approach is known as **Lasso Regression (L1)** as is often written as:\n",
    "\n",
    "$$\\displaystyle{\\hat{\\beta} = \\arg \\min_{\\beta} ||Y-X\\beta||_2^2 + \\lambda ||\\beta||_1  }$$\n",
    "\n",
    "Thus, given some optimal minimizer $\\lambda$, the two above optimization problems are identical -- for vectos $X \\in \\mathbb{R}^{n \\times p}$ and $Y \\in \\mathbb{R}^n$ .  This hints to the fact that addition of some bias can improve the estimation error."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "In a similar fashion we can take the OLS estimators and add statistical shrinkage.\n",
    "\n",
    "$$\\displaystyle{\\hat{\\beta} = \\arg \\min_{\\beta} ||Y-X\\beta||_2^2 + \\lambda||\\beta||_2^2  }$$\n",
    "\n",
    "Where $\\lambda \\ge 0$. \n",
    "\n",
    "Now consider a one-dimensional case where $x_i \\in \\mathbb{R}$. We have:\n",
    "\n",
    "$$\\displaystyle{ \\hat{\\beta} = \\arg \\min \\sum_{i=1}^n (y_i-x_i \\cdot \\beta)^2 + \\lambda \\beta^2} $$\n",
    "\n",
    "$$\\displaystyle{  = \\arg \\min \\sum_{i=1}^n y_i^2 -2y_ix_i\\beta + x_i^2 \\beta^2 + \\lambda \\beta^2} $$\n",
    "\n",
    "$$\\displaystyle{ = \\arg \\min \\left(\\sum_{i=1}^n -2y_ix_i\\right)\\beta + \\left(\\lambda + \\sum_{i=1}^n x_i^2\\right)\\beta^2 + \\sum_{i=1}^n y_i^2} $$\n",
    "\n",
    "Evaluating the objective function at zero we have:\n",
    "\n",
    "\n",
    "$$\\displaystyle{\\left(\\sum_{i=1}^n -2y_ix_i\\right) + 2 \\left(\\lambda + \\sum_{i=1}^n x_i^2\\right)\\beta=0  ~~~~ \\Longrightarrow ~~~~ \\hat{\\beta} = \\frac{\\left(\\sum_{i=1}^n y_ix_i\\right)}{\\left(\\lambda + \\sum_{i=1}^n x_i^2 \\right)}} $$\n",
    "\n",
    "<br>\n",
    "\n",
    "Note that if $\\lambda = 0$ then the previous equation is the generic OLS estimate, while $\\lambda=\\infty$ yield no solution. For $\\lambda>0$ the denominator is large and will shrink toward zero as $\\lambda$ gets larger. \n",
    "\n",
    "Consequently, choosing the right $\\lambda$ will lead to a lower estimation error when imposing **Ridge Regression (L2)**. \n",
    "\n",
    "Lasso and Ridge tend to significantly outperform OLS because the introducion of a small amount of bias, via $\\lambda$ selection, pays off in a large decrease in variance.\n",
    "\n",
    "For full derivation steps see [here](https://math.bu.edu/people/cgineste/classes/ma575/p/w14_1.pdf). For a quick video of how they differ, see [here](https://youtu.be/Xm2C_gTAl8c)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**GEOMETRIC** INTERPRETATION OF **REGRESSION**\n",
    "\n",
    "<br>\n",
    "\n",
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"l1_l2_contour_detailed.png\" align=\"center\" width=\"70%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "    Code for creating these plots found at end of notebook\n",
    "</div>\n",
    "\n",
    "<br>\n",
    "\n",
    "In the above visualization, the ellipses correspond to the contours of the residual sum of squares (RSS). The smaller the ellipse, the smaller the RSS, with RSS being minimized at OLS estimates. The circle is Ridge regression, while the diamon is lasso regression. The goal is to reduce both the ellipse and the circle or diamond during L2, or L1 regularization, respectively. The ridge (or lasso) estimate is given by the point at which the ellipse and circle touch (or diamond for lasso). For more information on regression shrinkage methods, see [here](https://online.stat.psu.edu/stat508/book/export/html/732)\n",
    "\n",
    "\n",
    "**Note:** The above illustration provides a visual explanation for why we perform [feature scaling](cikit-learn.org/stable/auto_examples/preprocessing/plot_scaling_importance.html) prior to training our models. Loosely stated, if you consider the 'distance' our model must travel for each 'ring,' clearly the wider the ring and the more disparate the sizes of the rings, the longer it takes our model extract relationships on the way to the minima."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**About Lasso Regression (L1)**\n",
    "\n",
    ">* Lasso will make irrelevant feature potentially non-existent -- can penalize them with weights equal to zero.<br>\n",
    ">* Lasso is biased toward providing [sparse solutions](http://www.stat.cmu.edu/~ryantibs/statml/lectures/sparsity.pdf) -- extinguishes a feature's influence by driving them to zero.<br>\n",
    ">* Lasso is computationally more expensive than Ridge regression.<br>\n",
    ">* Lasso is sensitive to correlations between features.<br>\n",
    ">* Lasso depends on the algorithm used.<br>\n",
    ">* Lasso provides much more aggressive regularization because the intersection of the constraint function and the cost function occurs at the “vertices” of the diamond where features will tend toward zero. <br>\n",
    ">* The aforementioned behavior make Lasso atype of automatic feature selection."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**About Ridge Regression (L2)**\n",
    "\n",
    ">* If $\\lambda$ is picked \"well\", then regularization helps avoid overfitting.<br>\n",
    ">* We can use cross-validation to pick $\\lambda$.<br>\n",
    ">* Ridge finds featues that contribute little to the output and penanilzes them -- assigns them small non-zero weights.<br>\n",
    ">* Ridge is a shrinkage method that is biased towards a [dense model](https://datacadamia.com/data_mining/shrinkage) -- penalizes features without extinguishing them altogher.<br>\n",
    ">* Ridge regression provides a less aggressive form of regularization because the contraint function and cost function can approach but never meet at vertices. As such the coefficients tend to zero in the limit only, meaning they are not extinguished.\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Compute Optimal Tunning Parameter for Lasso and Ridge**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 335,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import cross_val_score, KFold\n",
    "from sklearn.linear_model import Lasso, Ridge, RidgeCV\n",
    "\n",
    "# hyperparams\n",
    "alphas = [1e-4, 1e-3, 1e-2, 1e-1, 1, 1e1, 1e2, 1e3]\n",
    "\n",
    "val_errors_l1 = []\n",
    "val_errors_l2 = []\n",
    "\n",
    "# fit iteratively to find optimal lambda\n",
    "for alpha in alphas:\n",
    "    lasso = Lasso(alpha=alpha, fit_intercept=True, random_state=290)\n",
    "    ridge = Ridge(alpha=alpha, fit_intercept=True, random_state=290)\n",
    "    errors_l1 = np.sum(-cross_val_score(lasso, \n",
    "                                        data, \n",
    "                                        y=target, \n",
    "                                        scoring='neg_mean_squared_error', \n",
    "                                        cv=10, \n",
    "                                        n_jobs=-1))\n",
    "    val_errors_l1.append(np.sqrt(errors_l1))\n",
    "    errors_l2 = np.sum(-cross_val_score(ridge, \n",
    "                                        data, \n",
    "                                        y=target, \n",
    "                                        scoring='neg_mean_squared_error', \n",
    "                                        cv=10, \n",
    "                                        n_jobs=-1))\n",
    "    val_errors_l2.append(np.sqrt(errors_l2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "L1 Regression best lambda: 100.0\n",
      "L2 Regression best lambda: 1000.0\n"
     ]
    }
   ],
   "source": [
    "print('L1 Regression best lambda: {}'.format(alphas[np.argmin(val_errors_l1)]))\n",
    "print('L2 Regression best lambda: {}'.format(alphas[np.argmin(val_errors_l2)]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Computing Training/Validation Errors Using Lasso and Ridge Regression (w/out Cross Validation)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 337,
   "metadata": {},
   "outputs": [],
   "source": [
    "# declared at start, import added here for additional clarity\n",
    "from sklearn.metrics import mean_squared_error\n",
    "\n",
    "def evaluateModel(alphas, reg_model, X_train, y_train, X_validation, y_validation, X_test, y_test):\n",
    "    \"\"\"computes RMSE given alphas, and regularization model\"\"\"\n",
    "    for alpha in alphas:\n",
    "        # instantiate and fit model\n",
    "        model = reg_model(alpha=alpha, fit_intercept=True, random_state=99)\n",
    "        model.fit(X_train, y_train)\n",
    "        # calculate errors\n",
    "        validation_error = mean_squared_error(y_validation, model.predict(X_validation))\n",
    "        test_error = mean_squared_error(y_test, model.predict(X_test))\n",
    "        print('alpha: {:7} | train error: {:5} | val error: {:6} | test error: {}'.format(\n",
    "            alpha,\n",
    "            round(np.sqrt(train_error),3),\n",
    "            round(np.sqrt(validation_error),3),\n",
    "            round(np.sqrt(test_error),3)))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 338,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "LASSO ERRORS -- RMSE\n",
      "------------------------------------------------------------------------------------------\n",
      "alpha:  0.0001 | train error: 6.773 | val error: 109.485 | test error: 39.757\n",
      "alpha:   0.001 | train error: 6.773 | val error: 109.486 | test error: 39.758\n",
      "alpha:    0.01 | train error: 6.773 | val error: 109.491 | test error: 39.766\n",
      "alpha:     0.1 | train error: 6.773 | val error: 109.598 | test error: 39.869\n",
      "alpha:       1 | train error: 6.773 | val error: 109.581 | test error: 39.261\n",
      "alpha:    10.0 | train error: 6.773 | val error: 109.053 | test error: 37.108\n",
      "alpha:   100.0 | train error: 6.773 | val error: 109.053 | test error: 37.108\n",
      "alpha:  1000.0 | train error: 6.773 | val error: 109.053 | test error: 37.108\n"
     ]
    }
   ],
   "source": [
    "# training and testing error to three significant figures\n",
    "print('\\n\\nLASSO ERRORS -- RMSE')\n",
    "print('-' * 90)\n",
    "evaluateModel(alphas, Lasso,  X_train, y_train, X_validation, y_validation, X_test, y_test)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 339,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "RIDGE ERRORS -- MSE\n",
      "------------------------------------------------------------------------------------------\n",
      "alpha:  0.0001 | train error: 6.773 | val error: 109.485 | test error: 39.757\n",
      "alpha:   0.001 | train error: 6.773 | val error: 109.485 | test error: 39.757\n",
      "alpha:    0.01 | train error: 6.773 | val error: 109.486 | test error: 39.757\n",
      "alpha:     0.1 | train error: 6.773 | val error: 109.488 | test error: 39.761\n",
      "alpha:       1 | train error: 6.773 | val error: 109.508 | test error: 39.791\n",
      "alpha:    10.0 | train error: 6.773 | val error: 109.593 | test error: 39.909\n",
      "alpha:   100.0 | train error: 6.773 | val error: 109.462 | test error: 39.405\n",
      "alpha:  1000.0 | train error: 6.773 | val error: 109.114 | test error: 37.697\n"
     ]
    }
   ],
   "source": [
    "# training and testing error to three significant figures\n",
    "print('\\n\\nRIDGE ERRORS -- MSE')\n",
    "print('-' * 90)\n",
    "evaluateModel(alphas, Ridge,  X_train, y_train, X_validation, y_validation, X_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Computing 10-Fold Cross-Validation $R^2$ and $MSE$ on vanilla, Lasso and Ridge Regressions Using Standardized Features**\n",
    "\n",
    "**Note:** we are comparing the three regression on a $(training ~|~ testing)$ set that did not undergo the usual splitting process. This is because we are going to let ScikitLearn handle the k-fold cross-validation. We are also passing the alphas as a list, since ScikitLearn can find and auto-utilize the optimal tunning parameter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 439,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load additional packages \n",
    "from sklearn.metrics import mean_absolute_error, mean_squared_error,r2_score\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.linear_model import LassoCV, Ridge, LinearRegression\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "\n",
    "%matplotlib inline\n",
    "plt.style.use('seaborn-white')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 452,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standardize features\n",
    "scaler = StandardScaler()\n",
    "data_std = scaler.fit_transform(data)\n",
    "\n",
    "# Training | Validation | Testing \n",
    "# (1) split to original data into train/test (80/20)\n",
    "X_intermediate, X_test, y_intermediate, y_test = train_test_split(data_std,\n",
    "                                                                  target,\n",
    "                                                                  shuffle=True,\n",
    "                                                                  test_size=0.2,\n",
    "                                                                  random_state=290)\n",
    "\n",
    "# (2) then split train again into train/validation (75/25)\n",
    "X_train, X_validate, y_train, y_validate = train_test_split(X_intermediate,\n",
    "                                                                y_intermediate,\n",
    "                                                                shuffle=False,\n",
    "                                                                test_size=0.25,\n",
    "                                                                random_state=290)\n",
    "\n",
    "# (3) delete intermediate variables (optional)\n",
    "del X_intermediate, y_intermediate\n",
    "\n",
    "\n",
    "# options for tunning parameters\n",
    "alphas = np.linspace(1e-4, 1e4, 1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Compare Regression Metrics**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 453,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluateFittedModel(reg_model, X_train, y_train, X_validate, y_validate, X_test, y_test, alphas, cv=2, plot_error=False):\n",
    "    \"\"\"computes and print MSE and RMSE given a fitted model\"\"\"\n",
    "    \n",
    "    # fit model\n",
    "    if reg_model is not LinearRegression:\n",
    "        model = reg_model(alphas=alphas, cv=cv).fit(X_train, y_train)\n",
    "    else:\n",
    "        model = reg_model().fit(X_train,y_train)\n",
    "        \n",
    "    # metrics\n",
    "    r2_validate = r2_score(y_validate,model.predict(X_validate))*100\n",
    "    r2_test = r2_score(y_test,model.predict(X_test))*100\n",
    "    mse_validate = mean_squared_error(y_validate, model.predict(X_validate))\n",
    "    mse_test = mean_squared_error(y_test, model.predict(X_test))\n",
    "    mae_validate = mean_absolute_error(y_validate, model.predict(X_validate))\n",
    "    mae_test = mean_absolute_error(y_test, model.predict(X_test))\n",
    "    print('validation R^2: {:6} | testing R^2: {}'.format(\n",
    "        round(r2_validate,3),\n",
    "        round(r2_test,3)))\n",
    "    print('validation MSE: {:6} | testing MSE: {}'.format(\n",
    "        round(mse_validate,3),\n",
    "        round(mse_test,3)))\n",
    "    print('validation RMSE: {:6} | testing RMSE: {}'.format(\n",
    "        round(np.sqrt(mse_validate),3),\n",
    "        round(np.sqrt(mse_validate),3)))\n",
    "    print('validation MAE: {:6} | testing MAE: {}'.format(\n",
    "        round(mae_validate,3),\n",
    "        round(mae_test,3)))\n",
    "    print()\n",
    "    \n",
    "    # Plot\n",
    "    if plot_error == True and reg_model is LassoCV:\n",
    "        \n",
    "        #create figure and axes\n",
    "        fig = plt.figure(figsize = (17,7))\n",
    "        \n",
    "        ax = fig.add_subplot(1, 2, 1)\n",
    "        # plot mean squared error of fold\n",
    "        ax.plot(model.alphas_, model.mse_path_, lw=2, alpha=0.5, color='#F9634B')\n",
    "        ax.plot(model.alphas_, model.mse_path_.mean(axis=-1),label='Average MSE across the folds', linewidth=1, alpha=.7, color=\"#2F5E85\")\n",
    "        ax.set_xlabel('$\\lambda$')\n",
    "        ax.set_ylabel('Mean Squared Error ($MSE$)')\n",
    "        ax.set_title('Mean Squared Error Per Fold')\n",
    "        # plot best tunning parameter\n",
    "        ax.axvline(model.alpha_, color=\"#434343\", linestyle='--', alpha=.5, label='Optimal Tunning Parameter')\n",
    "        \n",
    "         # legend\n",
    "        plt.legend(loc=\"best\")\n",
    "        \n",
    "        ax = fig.add_subplot(1, 2, 2)\n",
    "        # plot root mean squared error of fold\n",
    "        ax.plot(model.alphas_, np.sqrt(model.mse_path_), lw=2, alpha=0.5, color='#F9634B')\n",
    "        ax.plot(model.alphas_, np.sqrt(model.mse_path_.mean(axis=-1)), label='Average RMSE across the folds', linewidth=1, alpha=.7, color=\"#2F5E85\")\n",
    "        ax.set_xlabel('$\\lambda$')\n",
    "        ax.set_ylabel('Root Mean Squared Error ($RMSE$)')\n",
    "        ax.set_title('Root Mean Squared Error Per Fold')\n",
    "        # plot best tunning parameter\n",
    "        ax.axvline(model.alpha_, color=\"#434343\", linestyle='--', alpha=.5, label='Optimal Tunning Parameter')\n",
    "        \n",
    "        # legend\n",
    "        plt.legend(loc=\"best\")\n",
    "\n",
    "        plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 454,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Ordinary Least Squares\n",
      "----------------------------------------------------------------------------------------------------\n",
      "validation R^2: -1.373 | testing R^2: -14.97\n",
      "validation MSE: 11987.036 | testing MSE: 1580.614\n",
      "validation RMSE: 109.485 | testing RMSE: 109.485\n",
      "validation MAE: 31.252 | testing MAE: 20.004\n",
      "\n",
      "\n",
      "\n",
      "Cross-Validation Lasso\n",
      "----------------------------------------------------------------------------------------------------\n",
      "validation R^2: -0.574 | testing R^2: -0.16\n",
      "validation MSE: 11892.498 | testing MSE: 1377.017\n",
      "validation RMSE: 109.053 | testing RMSE: 109.053\n",
      "validation MAE: 23.448 | testing MAE: 16.07\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x504 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Cross-Validation Ridge\n",
      "----------------------------------------------------------------------------------------------------\n",
      "validation R^2: -0.565 | testing R^2: -0.665\n",
      "validation MSE: 11891.508 | testing MSE: 1383.954\n",
      "validation RMSE: 109.048 | testing RMSE: 109.048\n",
      "validation MAE:   23.8 | testing MAE: 16.154\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# training and testing error to three significant figures\n",
    "print('\\n\\nOrdinary Least Squares')\n",
    "print('-' * 100)\n",
    "evaluateFittedModel(LinearRegression,  X_train, y_train, X_validation, y_validation, X_test, y_test, alphas)\n",
    "print('\\n\\nCross-Validation Lasso')\n",
    "print('-' * 100)\n",
    "evaluateFittedModel(LassoCV,  X_train, y_train, X_validation, y_validation, X_test, y_test, alphas, cv=10, plot_error=True)\n",
    "print('\\n\\nCross-Validation Ridge')\n",
    "print('-' * 100)\n",
    "evaluateFittedModel(RidgeCV,  X_train, y_train, X_validation, y_validation, X_test, y_test, alphas, cv=5, plot_error=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "**Discussion:**\n",
    "\n",
    "The low $R^2$ accross all models is telling us that our modeling approach fails to explain any variability of the response data around its mean. However, this does not mean we have a bad model. $R^2$ cannot determine whether the coefficient estimates and predictions are biased -- which is why visualizations are so helpful in evaluating the accuracy of our models. For a more comprehensive explanation of $R^2$ in application, see [here](https://blog.minitab.com/blog/adventures-in-statistics-2/regression-analysis-how-do-i-interpret-r-squared-and-assess-the-goodness-of-fit).\n",
    "    \n",
    "For $MSE$ note the extremely high values of MSE -- two orders of magnitue larger than MAE. Recall that MSE squares the error terms, sums them, and finally takes the average. Our model may be still have outlier values that are inflating our MSE measure.\n",
    "\n",
    "$RMSE$ is consistent throughout our models with some differences in the significant figures. RMSE is a reliable way of comparing models, since it indicates the abslute fit of the model to the data while also providing an error metric that is in  the same units as the response variable. RMSE can be thought of as the standard deviation (i.e. unexplained variance). When comparing models, the model with the lowest RMSE tends to be the best.\n",
    "\n",
    "When it comes to $MAE$, RMSE should always be larger that MAE particularly as sample size increases. MAE is less sensitive to outliers, and commonly used as an evaluation metric when modeling time-dependent data. MAE is less biased than MSE, clearly, but MAE may not reflect the model's performance when dealing with large errors.\n",
    "        \n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "## References and Additional Resources\n",
    "___\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "#### METRICS\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### **[Residual Sum of Squares (RSS)](https://en.wikipedia.org/wiki/Residual_sum_of_squares)**\n",
    "\n",
    "$$\\displaystyle{RSS(\\beta) = \\sum_{i=1}^N \\left( y_i - f(x_i) \\right) = \\sum_{i=1}^N \\left(y_i-\\beta_0 - \\sum_{j=1}^p x_{ij}\\beta_j \\right)^2}$$\n",
    "\n",
    "About: \n",
    ">* A [residual](https://en.wikipedia.org/wiki/Errors_and_residuals), is equal to the difference between the observed and predicted value. $e_i = y_ - \\hat{y}_i$ <br>\n",
    ">* RSS is a measure of the amount of variance in the data that is not explained by the regression model.<br>\n",
    ">* RSS seems like a reasonable metic if each trainging observation $(x_i, y_i)$ represents a random draw from the underlying population -- independent, and indentically distributed.<br>\n",
    ">* If the $x_i~'s$ are not drawn randomly then so long as the $y_i~'s$ are conditionally independent given the inputs $x_i$ then we can assume RSS is still a good metric.<br>\n",
    ">* RSS seeks to minimize the distance between actual and predicted data. That is, the lower RSS the better the regression model ability to explain the data.<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "##### **[Root Mean Squared Error (RMSE)](https://en.wikipedia.org/wiki/Root-mean-square_deviation)**\n",
    "\n",
    "Consider the [Mean Squared Error (MSE)](https://en.wikipedia.org/wiki/Mean_squared_error) of an estimator $\\tilde{\\theta}$ estimating $\\theta$. We have\n",
    "\n",
    "$$\\displaystyle{MSE(\\tilde{\\theta}) = E\\left(\\tilde{\\theta} - \\theta\\right)^2 = Var\\left(\\tilde{\\theta}\\right) + \\left[E\\left(\\tilde{\\theta}\\right) - \\theta\\right]^2}$$\n",
    "\n",
    "Then for $y_i \\in \\theta$ and $\\hat{y}_i \\in \\hat{\\theta}$ for $i=1,2,\\dots$\n",
    "\n",
    "\n",
    "$$\\displaystyle{RMSE(\\hat{\\theta}) = \\sqrt{MSE(\\hat{\\theta})} = \\sqrt{E[(\\hat{\\theta}-\\theta)^2]} = \\sqrt{\\frac{\\sum_i(\\hat{y}_i - y_i)^2}{n}}}$$\n",
    "\n",
    "\n",
    "About: \n",
    ">* [Gauss-Markok Theorem](https://math.unm.edu/~fletcher/JPG/mch4.pdf) implies that the least squares estimator has the smallest mean square error -- across all linear estimators with no bias. However, there may exist a biased estimator that has a larger reduction in variance.<br>\n",
    ">* RMSE is the [Euclidean distance](https://en.wikipedia.org/wiki/Euclidean_distance) between the vector of true values and the vector of predicted values, averaged by the square roon of the number of data points.<br>\n",
    ">* Because RMSE and MSE are an average, they sensitive to large outliers.\n",
    ">* Because RMSE and MSE are squared, they penalize small errors which leads to over-estimation of poor model performance.\n",
    ">* RMSE is preferred over MSE when large errors are undesired."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "##### **[Mean Absolute Error (MAE)](https://en.wikipedia.org/wiki/Mean_absolute_error)**\n",
    "\n",
    "\n",
    "$$\\displaystyle{MAE = \\frac{1}{n} \\sum_{i=1}^n \\left|\\hat{y}_i - y_i\\right|}$$\n",
    "\n",
    "\n",
    "About: \n",
    ">* MAE is an arithmetic average of the absolute errors $\\left|e_i\\right|=\\left|\\hat{y}_-y_i\\right|$ where $\\hat{y}_i$ is the prediction and $y_i$ is the actual value.<br>\n",
    ">* MAE uses same scale as date being measured. As such it cannot be used to compare between different scales.<br>\n",
    ">* MAE is a common measure of forecast error in time series analysis."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "##### **[Coefficient of Determination ($R^2$)](https://en.wikipedia.org/wiki/Coefficient_of_determination)**\n",
    "\n",
    "Let $e_i = y_i - f(x_i)$ be a vector of residuals. If $\\bar{y}$ is the mean of the observed data\n",
    "\n",
    "$$\\displaystyle{\\bar{y}=\\frac{1}{n}\\sum_{i=1}^n y_i}$$\n",
    "\n",
    "Then the variablity of the data is captured by the [Residual Sum of Squares (RSS)](https://en.wikipedia.org/wiki/Residual_sum_of_squares) and the [Total Sum of Squares (SST)](https://en.wikipedia.org/wiki/Total_sum_of_squares). Namely,\n",
    "\n",
    "$$\\displaystyle{RSS =\\sum_{i=1}^n \\left(y_i - f(x_i)\\right)^2 = \\sum_{i=1}^n(y_i - \\hat{y}_i)^2}$$ <br>\n",
    "\n",
    "$$\\displaystyle{SST = \\sum_i \\left(y_i - \\bar{y}\\right)^2}$$\n",
    "\n",
    "Then the [coefficient of determination](https://youtu.be/xxFYro8QuXA) is given by \n",
    "\n",
    "\n",
    "$$\\displaystyle{R^2 = 1- \\frac{SSR}{SST} = 1-\\frac{\\sum_{i=1}^n(y_i -\\hat{y}_i)^2}{\\sum_{i=1}^n (y_i-\\bar{y}_i)^2}}$$\n",
    "\n",
    "\n",
    "About: \n",
    ">* $R^2$ is a highly interpretable, scale free, score that is always less than or equal to 1, that provides a meassure for how well the regression predictions approximate the real data points.<br>\n",
    ">* $R^2$ of 1 implies that the predictions perfectly fit the data, while $R^2=0$ implies no linear relationship between the dependent and independent variables..<br>\n",
    ">* When expressed as a percentage, $R^2$ gives the percentage of total sum of squares that can be explained using the regression equation.\n",
    ">* A negative $R^2$ score, $-\\infty<R^2<0$, may be due to: (1) Choosing an incorrect model that does not fit the data (numerator high); (2) There is a high number of large outlier data points causing bloating of the denominator; (3) A missing intercept, which serves as the baseline when tracking the dependent variable (seen as a large nominator).  \n",
    ">* Values of $R^2$ between 0 and 1, $0\\le R^2\\le1$, occur when the model fist the data worse than a horizontal hyperplane.  This is indicative of either using an incorrect model or the existance of non-sensical constraints.<br> \n",
    ">* One possible drawback for using $R^2$ is that $R^2$ [increases monotonically](https://en.wikipedia.org/wiki/Monotonic_function) with the number of features. A [kitchen sink regression](https://en.wikipedia.org/wiki/Kitchen_sink_regression) can become a tempting approach for incraseing $R^2$.\n",
    ">* An alternate to $R^2$ is [$adjusted~ R^2$](https://en.wikipedia.org/wiki/Coefficient_of_determination#Adjusted_R2), which essentially the same as $R^2$ but with a penalty added as more variables are included in the model.\n",
    "\n",
    "<br>\n",
    "\n",
    "**Note:** If $X_{n \\times p}$ is the feature vector and $rank(X) = p$ then for a given n-vector of responses $y$, $RSS(\\beta)$ is given by \n",
    "\n",
    "<br>\n",
    "\n",
    "$$\\displaystyle{\\min_{\\beta \\in \\mathbb{R}^n} RSS(\\beta) = \\hat{\\beta} \\left(X^TX\\right)^{-1} X^Ty }$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "##### **[Quantiles of Error](https://en.wikipedia.org/wiki/Quantile_regression)**\n",
    "\n",
    "Quantiles (aka percentiles) are a metric that is less susceptible to large outliers. Which percentile to use as an evaluation metric will depend on the problem. In general the [Median Absolute Percentage Error (MAPE)](https://en.wikipedia.org/wiki/Mean_absolute_percentage_error) is a good quantile to begin with. \n",
    "\n",
    "$$\\displaystyle{MAPE = median\\left(\\left|\\frac{\\hat{y}_i - y_i}{y_i}\\right|\\right)}$$\n",
    "\n",
    "\n",
    "About: \n",
    ">* MAPE gives a relative measure for error. For example computing the 90th percentile of the absolute percent error would give an indication of \"almost worst case\" behavior."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "[**Variance Inflation Factors (VIF)**](https://en.wikipedia.org/wiki/Variance_inflation_factor) is a metric that gives us a feature's individual cotribution to total variance.\n",
    "\n",
    "$$\\displaystyle{ VIF_i = \\frac{1}{1-R^2_i} }$$\n",
    "\n",
    "About:\n",
    ">* $VIF > 10$ --> definetly a problem with that feature<br>\n",
    ">* $VIF > ~~5$  --> there may be a problem with the feature<br>\n",
    ">* $VIF \\le ~~5$ --> the feature is, probably, ok<br>\n",
    "\n",
    ">* To deal with multicollinearity you can drop variables with high VIF. \n",
    ">* If more tha one variable have similar VIF values, keep the one you like.\n",
    ">* Repeat iteratively (meaning recalculate VIF values and see if high VIF values remain)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "#### Install Python Development Environment"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Install/Update pip3**\n",
    "\n",
    "https://pip.pypa.io/en/stable/installing/\n",
    "\n",
    "```bash\n",
    "    # TensorFlow requires pip version >= 19.0\n",
    "    $ pip install --upgrade pip \n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Install/Update Python 3**\n",
    "\n",
    "https://www.python.org/downloads/\n",
    "\n",
    "```bash\n",
    "    # TensorFlow requires Python 3.5-3.8 \n",
    "    $ sudo apt-get update && sudo apt-get install python3-dev python3-pip python3-venv python-virtualenv\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "#### Create The Validation Curve Plot\n",
    "\n",
    "Based on: https://stats.stackexchange.com/users/192854/xavier-bourret-sicotte"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create data\n",
    "x = np.linspace(0, 1, 1000)\n",
    "y1 = -(x - 0.5) ** 2\n",
    "y2 = y1 - 0.33 + np.exp(x - 1)\n",
    "\n",
    "#create figure and axes\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "\n",
    "# plot artists (curves)\n",
    "ax.plot(x, y2, lw=8, alpha=0.7, color='#2F5E85')\n",
    "ax.plot(x, y1, lw=8, alpha=0.7, color='#F9634B')\n",
    "\n",
    "# add text to figure\n",
    "ax.text(0.17, 0.05, \"Training Score\", rotation=50, size=16, color='#2F5E85')\n",
    "ax.text(0.2, -0.05, \"Validation Score\", rotation=20, size=16, color='#F9634B')\n",
    "\n",
    "ax.text(0.02, 0.03, r'$\\longleftarrow$ High Bias', size=16, rotation=90, va='center', color=\"#434343\")\n",
    "ax.text(0.98, 0.12, r'$\\longleftarrow$ High Variance $\\longrightarrow$', size=16, rotation=90, ha='right', va='center', color=\"#434343\")\n",
    "ax.text(0.48, -0.08, 'Best$\\\\longrightarrow$\\nModel', size=16, rotation=90, va='center', color=\"#434343\")\n",
    "\n",
    "# modify axes limits\n",
    "ax.set_xlim(0, 1)\n",
    "ax.set_ylim(-0.3, 0.5)\n",
    "\n",
    "# create artist labels\n",
    "ax.set_xlabel(r'Model Complexity $\\longrightarrow$', size=16, color=\"#434343\")\n",
    "ax.set_ylabel(r'Model Score (e.g. Accuracy) $\\longrightarrow$', size=16, color=\"#434343\")\n",
    "\n",
    "# set formating on major ticker\n",
    "ax.xaxis.set_major_formatter(plt.NullFormatter())\n",
    "ax.yaxis.set_major_formatter(plt.NullFormatter())\n",
    "\n",
    "# header\n",
    "ax.set_title(\"VALIDATION CURVES\", size=18, color=\"#434343\")\n",
    "# save image\n",
    "fig.savefig('validation_curve.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "#### Create The Learning Curve Plot\n",
    "\n",
    "Based on: https://stats.stackexchange.com/users/192854/xavier-bourret-sicotte"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create data\n",
    "N = np.linspace(0, 1, 1000)\n",
    "y1 = 0.75 + 0.2 * np.exp(-4 * N)\n",
    "y2 = 0.7 - 0.6 * np.exp(-4 * N)\n",
    "\n",
    "#create figure and axes\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "\n",
    "# plot artists (curves)\n",
    "ax.plot(x, y1, lw=8, alpha=0.7, color='#2F5E85')\n",
    "ax.plot(x, y2, lw=8, alpha=0.7, color='#F9634B')\n",
    "\n",
    "# plot artist (horizontal line) -- Desired Accuracy\n",
    "ax.axhline(y=.725, color=\"#96A3B0\", linestyle='--')\n",
    "\n",
    "# add text to figure\n",
    "ax.text(0.15, 0.82, \"Training Score\", rotation=-15, size=16, color='#2F5E85')\n",
    "ax.text(0.15, 0.48, \"Validation Score\", rotation=30, size=16, color='#F9634B')\n",
    "ax.text(0.15, 0.73, \"Desired Score\", rotation=0, size=12, color='#96A3B0')\n",
    "\n",
    "ax.text(0.98, 0.5, r'Good Fit $\\longrightarrow$', size=16, rotation=90, ha='right', va='center', color=\"#434343\")\n",
    "ax.text(0.02, 0.55, r'$\\longleftarrow$ High Variance $\\longrightarrow$', size=16, rotation=90, va='center', color=\"#434343\")\n",
    "\n",
    "# modify axes limits\n",
    "ax.set_xlim(0, 1)\n",
    "ax.set_ylim(0, 1)\n",
    "\n",
    "# create artist labels\n",
    "ax.set_xlabel(r'Training Data Size $\\longrightarrow$', size=16, color=\"#434343\")\n",
    "ax.set_ylabel(r'Model Score (e.g. RMSE) $\\longrightarrow$', size=16, color=\"#434343\")\n",
    "\n",
    "# set formating on major ticker\n",
    "ax.xaxis.set_major_formatter(plt.NullFormatter())\n",
    "ax.yaxis.set_major_formatter(plt.NullFormatter())\n",
    "\n",
    "# header\n",
    "ax.set_title(\"LEARNING CURVES\", size=18, color=\"#434343\")\n",
    "\n",
    "# save image\n",
    "fig.savefig('learning_curve.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<br>\n",
    "\n",
    "#### Visualize OLS, L1, L2, and Solution Paths\n",
    "\n",
    "Based on: https://stats.stackexchange.com/users/192854/xavier-bourret-sicotte"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 438,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn import linear_model\n",
    "\n",
    "%matplotlib inline\n",
    "plt.style.use('seaborn-white')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Simulate dataset**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Creating the dataset (as previously)\n",
    "x = np.linspace(0,1,40)\n",
    "noise = 1*np.random.uniform(size = 40)\n",
    "y = np.sin(x * 1.5 * np.pi ) \n",
    "y_noise = (y + noise).reshape(-1,1)\n",
    "\n",
    "#Subtracting the mean so that the y's are centered\n",
    "y_noise = y_noise - y_noise.mean()\n",
    "X = np.vstack((2*x,x**2)).T\n",
    "\n",
    "#Nornalizing the design matrix to facilitate visualization\n",
    "X = X / np.linalg.norm(X,axis = 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Helper Functions**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def costfunction(X,y,theta):\n",
    "    '''OLS cost function'''\n",
    "    #Initialisation of useful values \n",
    "    m = np.size(y)\n",
    "    \n",
    "    #Cost function in vectorized form\n",
    "    h = X @ theta\n",
    "    J = float((1./(2*m)) * (h - y).T @ (h - y));    \n",
    "    return (J);\n",
    "\n",
    "def closed_form_solution(X,y):\n",
    "    '''Linear regression closed form solution'''\n",
    "    return (np.linalg.inv(X.T @ X) @ X.T @ y)\n",
    "    \n",
    "def closed_form_reg_solution(X,y,lamda = 10): \n",
    "    '''Ridge regression closed form solution'''\n",
    "    m,n = X.shape\n",
    "    I = np.eye((n))\n",
    "    return ((np.linalg.inv(X.T @ X + lamda * I) @ X.T @ y)[:,0])\n",
    "\n",
    "def cost_l1(x,y):\n",
    "    '''Lasso (L1) cost function'''\n",
    "    return (np.abs(x) + np.abs(y))\n",
    "\n",
    "def cost_l2(x,y):\n",
    "    '''Ridge (L2) cost functiom'''\n",
    "    return (x**2 + y**2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Compute Closed Form Solutions as Functions of $\\lambda$**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Lasso\n",
    "lambda_range = np.logspace(0,5,num = 400)/4000\n",
    "theta_0_list_reg_l1 = []\n",
    "theta_1_list_reg_l1 = []\n",
    "\n",
    "for l in lambda_range:\n",
    "    model_sk_reg = linear_model.Lasso(alpha=l, fit_intercept=False)\n",
    "    model_sk_reg.fit(X,y_noise)\n",
    "    t0, t1 = model_sk_reg.coef_\n",
    "    theta_0_list_reg_l1.append(t0)\n",
    "    theta_1_list_reg_l1.append(t1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Ridge\n",
    "lambda_range = np.logspace(0,5,num = 200)/2000\n",
    "theta_0_list_reg_l2 = []\n",
    "theta_1_list_reg_l2 = []\n",
    "\n",
    "for l in lambda_range:\n",
    "    t0, t1 = closed_form_reg_solution(X,y_noise,l)\n",
    "    theta_0_list_reg_l2.append(t0)\n",
    "    theta_1_list_reg_l2.append(t1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "**Plot Contour Plots and L1, L2 Paths**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ehch/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:28: MatplotlibDeprecationWarning: Adding an axes using the same arguments as a previous axes currently reuses the earlier instance.  In a future version, a new instance will always be created and returned.  Meanwhile, this warning can be suppressed, and the future behavior ensured, by passing a unique label to each axes instance.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1224x504 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Setup of meshgrid of theta values\n",
    "xx, yy = np.meshgrid(np.linspace(-3,17,100),np.linspace(-17,3,100))\n",
    "\n",
    "#Computing the cost function for each theta combination\n",
    "zz_l2 = np.array(  [cost_l2(xi, yi )for xi, yi in zip(np.ravel(xx), np.ravel(yy)) ] ) #L2 function\n",
    "\n",
    "zz_l1 = np.array(  [cost_l1(xi, yi )for xi, yi in zip(np.ravel(xx), np.ravel(yy)) ] ) #L1 function\n",
    "\n",
    "zz_ls = np.array(  [costfunction(X, y_noise.reshape(-1,1),np.array([t0,t1]).reshape(-1,1)) \n",
    "                     for t0, t1 in zip(np.ravel(xx), np.ravel(yy)) ] ) #least square cost function\n",
    "\n",
    "#Reshaping the cost values    \n",
    "Z_l2 = zz_l2.reshape(xx.shape)\n",
    "Z_ls = zz_ls.reshape(xx.shape)\n",
    "Z_l1 = zz_l1.reshape(xx.shape)\n",
    "\n",
    "#Defining the global min of each function\n",
    "min_ls = np.linalg.inv(X.T@X)@X.T@y_noise\n",
    "min_l2 = np.array([0.,0.])\n",
    "min_l1 = np.array([0.,0.])\n",
    "\n",
    "# setting figure and axes\n",
    "fig = plt.figure(figsize = (17,7))\n",
    "ax = fig.add_subplot(1, 2, 1)\n",
    "\n",
    "##################### LASSO #####################\n",
    "#Plotting the contours - L1 \n",
    "ax = fig.add_subplot(1, 2, 1)\n",
    "ax.contour(xx, yy, Z_l1, levels = [.5,.75,1.25,2.25,3.5,5,6.5,8,10,13,16,20],  cmap = 'bone')\n",
    "ax.contour(xx, yy, Z_ls, levels = [.05,.06,.07,.09,.115,.135], cmap = 'Blues')\n",
    "ax.set_xlabel('theta 1')\n",
    "ax.set_ylabel('theta 2')\n",
    "ax.set_title('Lasso solution as a function of $\\\\lambda$ - OLS and $L_1$ ', color = '#434343')\n",
    "\n",
    "#Plotting the minimum - L1\n",
    "ax.plot(min_ls[0],min_ls[1], marker = 'x', color = '#2F5E85', markersize = 10)\n",
    "ax.plot(0,0, marker = 'o', color = '#2F5E85', markersize = 10)\n",
    "\n",
    "#Plotting the path of L1 regularized minimum\n",
    "ax.plot(theta_0_list_reg_l1,theta_1_list_reg_l1, linestyle = 'none', marker = 'o', color = '#F9634B', alpha = .2)\n",
    "\n",
    "# add text to figure OLS contour\n",
    "ax.text(13.00, -14.00, \"OLS Estimate\", rotation=-40, size=12, color=\"#2F5E85\", alpha=.8)\n",
    "ax.text(14.90, -14.30, r'$\\longrightarrow$', size=38, rotation=-90, ha='right', va='center', alpha=.5, color=\"#2F5E85\")\n",
    "\n",
    "# add text to lasso contour\n",
    "ax.text(5.55, -6.35, \"Lasso Estimate\", rotation=43, size=12, color=\"#F9634B\", alpha=.8)\n",
    "ax.text(3.35, -7.10, \"*\", rotation=50, size=44, color=\"#F9634B\", alpha=1)\n",
    "ax.text(7.5, -4.70, r'$\\longrightarrow$', size=44, rotation=210, ha='right', va='center', alpha=.5, color=\"#F9634B\")\n",
    "##################### LASSO #####################\n",
    "\n",
    "##################### RIDGE #####################\n",
    "#Plotting the contours - L2 \n",
    "ax = fig.add_subplot(1, 2, 2)\n",
    "ax.contour(xx, yy, Z_l2, levels = [.2,.5,1.25,2.5,5,10,15,30,50,90,150,250],  cmap = 'bone')\n",
    "ax.contour(xx, yy, Z_ls, levels = [.05,.06,.07,.09,.115,.137], cmap = 'Blues')\n",
    "ax.set_xlabel('theta 1')\n",
    "ax.set_ylabel('theta 2')\n",
    "ax.set_title('Ridge solution as a function of $\\\\lambda$ - OLS and $L_2$ ', color = '#434343')\n",
    "\n",
    "#Plotting the minimum - L2 \n",
    "ax.plot(min_ls[0],min_ls[1], marker = 'x', color = '#2F5E85', markersize = 10)\n",
    "ax.plot(0,0, marker = 'o', color = '#2F5E85', markersize = 10)\n",
    "\n",
    "#Plotting the path of L2 regularized minimum\n",
    "ax.plot(theta_0_list_reg_l2,theta_1_list_reg_l2, linestyle = 'none', marker = 'o', color = '#F9634B', alpha = .2)\n",
    "\n",
    "# add text to figure OLS contour\n",
    "ax.text(13.00, -14.00, \"OLS Estimate\", rotation=-40, size=12, color=\"#2F5E85\", alpha=.8)\n",
    "ax.text(14.90, -14.30, r'$\\longrightarrow$', size=38, rotation=-90, ha='right', va='center', alpha=.5, color=\"#2F5E85\")\n",
    "\n",
    "# add text to Ridge contour\n",
    "ax.text(5.60, -6.30, \"Ridge Estimate\", rotation=50, size=12, color=\"#F9634B\", alpha=.8)\n",
    "ax.text(3.20, -6.90, \"*\", rotation=50, size=44, color=\"#F9634B\", alpha=1)\n",
    "ax.text(7.5, -4.70, r'$\\longrightarrow$', size=44, rotation=210, ha='right', va='center', alpha=.5, color=\"#F9634B\")\n",
    "##################### RIDGE #####################\n",
    "\n",
    "# save image\n",
    "fig.savefig('l1_l2_contour_detailed.png')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "## Additional Content and Sources\n",
    "\n",
    "> [Elements of Statistical Learnig by Trevor Hastie, Robert Tirnshirani, and Jerome Friedman](https://web.stanford.edu/~hastie/ElemStatLearn/)<br>\n",
    "> [Evaluating Machine Learning Models - AWS Developer's Guide](https://docs.aws.amazon.com/machine-learning/latest/dg/evaluating_models.html)<br>\n",
    "> [Machine Learning Crash Course - Google Machine Developer's Guide](https://developers.google.com/machine-learning/crash-course)<br>\n",
    "> [Evaluating Machine Learning Models by O'Reilley](https://www.oreilly.com/content/evaluating-machine-learning-models/)<br>\n",
    "> [DATA BLOG by Xavier Bourret Sicotte](https://xavierbourretsicotte.github.io/)<br>\n",
    "> [Model Tuning by David Ziganto](https://dziganto.github.io/data%20science/machine%20learning/model%20tuning/python/Model-Tuning-Train-Test-Split/)<br>\n",
    "> [Machine Learning for Operations Research & Financial Engineering by Martin Haugh](https://martin-haugh.github.io/teaching/ml-orfe/)<br>\n",
    "> [Forest Firre Impact Prediction (Stats and ML) by Vishnu and Koutharapu](https://www.kaggle.com/psvishnu/forestfire-impact-prediction-stats-and-ml)<br>\n",
    "> [Nuts and Bolts of Building Applications using Deep Learning](https://youtu.be/F1ka6a13S9I)<br>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div align=\"center\" style=\"font-size:12px; font-family:FreeMono; font-weight: 100; font-stretch:ultra-condensed; line-height: 1.0; color:#2A2C2B\">\n",
    "    <img src=\"DataX_icon_wide_logo.png\" align=\"center\" width=\"50%\" padding=\"0px\"><br>\n",
    "    <br>\n",
    "</div>\n",
    "\n",
    "___"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "celltoolbar": "Slideshow",
  "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": "3.8.8"
  },
  "livereveal": {
   "scroll": true
  },
  "rise": {
   "enable_chalkboard": true,
   "height": "90%",
   "width": "100%"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}