Quiz/m7/dependent_labels_solution.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Overlapping Labels\n",
    "\n",
    "In this exercise, we're going to **simulate** the issue we encounter when training machine learning models using targets that are returns that _overlap in time_. The core issue is that these labels are correlated and in fact _redundant_. We'll see what impact this has on our machine learning algorithm, and we'll have the opportunity to implement some of the solutions to the problem that we described in the lectures, and will later encounter in the project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "\n",
    "%matplotlib inline\n",
    "plt.style.use('ggplot')\n",
    "plt.rcParams['figure.figsize'] = (14, 8)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load data\n",
    "Let's load up a small dataset of artificial \"toy\" data that we can play with. The columns in these data are `A`, `B`, `C` and `D`. We will use the `E` column as targets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('dependent_labels_dataset.csv', index_col = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>E</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>t</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>0.780067</td>\n",
       "      <td>-1.139543</td>\n",
       "      <td>0.298152</td>\n",
       "      <td>0.412867</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.0</th>\n",
       "      <td>0.521061</td>\n",
       "      <td>-0.680001</td>\n",
       "      <td>0.148710</td>\n",
       "      <td>0.461035</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.0</th>\n",
       "      <td>0.425336</td>\n",
       "      <td>-0.620970</td>\n",
       "      <td>0.099168</td>\n",
       "      <td>0.409145</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4.0</th>\n",
       "      <td>0.696707</td>\n",
       "      <td>-0.666513</td>\n",
       "      <td>0.099526</td>\n",
       "      <td>0.477195</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5.0</th>\n",
       "      <td>-0.059698</td>\n",
       "      <td>-0.220574</td>\n",
       "      <td>0.249784</td>\n",
       "      <td>0.465186</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            A         B         C         D  E\n",
       "t                                             \n",
       "1.0  0.780067 -1.139543  0.298152  0.412867  3\n",
       "2.0  0.521061 -0.680001  0.148710  0.461035  3\n",
       "3.0  0.425336 -0.620970  0.099168  0.409145  3\n",
       "4.0  0.696707 -0.666513  0.099526  0.477195  3\n",
       "5.0 -0.059698 -0.220574  0.249784  0.465186  3"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fa4bbd958d0>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa4bbd95208>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000, 5)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create redundancy in the data\n",
    "In order to illustrate the effect of redundancy in the data, we are going to deliberately create an extreme version of this condition by **duplicating each row in the data 5 times**, using the function `create_redundant_data` below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_duplicates = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_redundant_data(df, num_duplicates):\n",
    "    \"\"\"\n",
    "    From the existing dataset, create a new dataset in which every original row is exactly duplicated `num_duplicates` times. \n",
    "    Re-order this new dataset according to the order of the original index.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    df : pandas DataFrame\n",
    "        The original dataset.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    redundant_df : pandas DataFrame\n",
    "        The new, redundant dataset.\n",
    "    \"\"\"\n",
    "    \n",
    "    redundant_df = df.copy()\n",
    "    for i in range(num_duplicates-1):\n",
    "        redundant_df = redundant_df.append(df)\n",
    "    redundant_df.sort_index(axis=0, level='t', inplace=True)\n",
    "    return redundant_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5000, 5)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "redundant_df = create_redundant_data(df, num_duplicates)\n",
    "redundant_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create a function for scoring the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model_score(m, X_train, y_train, X_valid, y_valid):\n",
    "    '''\n",
    "    Take in the model and training and validation datasets, and return the training accuracy score, validation\n",
    "    accuracy score, and out-of-bag score. Furthermore, print each of these results.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    m : RandomForestClassifier instance\n",
    "        The trained model.\n",
    "        \n",
    "    X_train : pandas DataFrame\n",
    "        The training features.\n",
    "        \n",
    "    y_train : pandas Series\n",
    "        The training labels.\n",
    "        \n",
    "    X_valid : pandas DataFrame\n",
    "        The validation features.\n",
    "        \n",
    "    y_valid : pandas Series\n",
    "        The validation labels.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    train_score : float\n",
    "        The mean training accuracy.\n",
    "        \n",
    "    valid_score : float\n",
    "        The mean validation accuracy.\n",
    "        \n",
    "    oob_score : float\n",
    "        The out-of-bag score.\n",
    "    \n",
    "    '''\n",
    "\n",
    "    train_score = m.score(X_train, y_train)\n",
    "    valid_score = m.score(X_valid, y_valid)\n",
    "    oob_score = m.oob_score_\n",
    "    print(\"train: %f, oob: %f, valid: %f\" % (train_score, oob_score, valid_score))\n",
    "    return train_score, valid_score, oob_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split data into train, valid and test sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_splits(df, features, target, split_valid=0.20, split_test=0.20):\n",
    "    temp = df.dropna()\n",
    "    \n",
    "    X = temp[features].copy()\n",
    "    y = temp[target].copy()\n",
    "\n",
    "    train_end = int(X.shape[0]*(1-split_valid-split_test))\n",
    "    valid_end = train_end + int(X.shape[0]*split_valid)\n",
    "\n",
    "    X_train, X_valid, X_test = X.iloc[:train_end,], X.iloc[(train_end+1):valid_end,], X.iloc[(valid_end+1):]\n",
    "    y_train, y_valid, y_test = y.iloc[:train_end,], y.iloc[(train_end+1):valid_end,], y.iloc[(valid_end+1):]\n",
    "    \n",
    "    return X, X_train, X_valid, X_test, y_train, y_valid, y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = ['A', 'B', 'C', 'D']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "X, X_train, X_valid, X_test, y_train, y_valid, y_test = make_splits(\n",
    "    redundant_df,\n",
    "    features,\n",
    "    'E'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train one tree, take a look at it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def instantiate_and_fit_one_tree(X_train, y_train):\n",
    "    \"\"\"\n",
    "    Instantiate a single decision tree and fit it on the training data. Return the fitted classifier.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    X_train : pandas DataFrame\n",
    "        The training features.\n",
    "        \n",
    "    y_train : pandas Series\n",
    "        The training labels.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    clf : DecisionTreeClassifier\n",
    "        The fitted classifier instance.\n",
    "    \"\"\"\n",
    "    \n",
    "    # you can do this with a DecisionTreeClassifier or with a RandomForestClassifier, with n_estimators=1\n",
    "    clf = RandomForestClassifier(\n",
    "    n_estimators=1,\n",
    "    max_depth=3,\n",
    "    max_features=None,\n",
    "    bootstrap=True,\n",
    "    criterion='entropy',\n",
    "    random_state=0\n",
    "    )\n",
    "    \n",
    "    clf.fit(X_train.values, y_train)\n",
    "    return clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "clf = instantiate_and_fit_one_tree(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting graphviz\n",
      "  Downloading https://files.pythonhosted.org/packages/1f/e2/ef2581b5b86625657afd32030f90cf2717456c1d2b711ba074bf007c0f1a/graphviz-0.10.1-py2.py3-none-any.whl\n",
      "Installing collected packages: graphviz\n",
      "Successfully installed graphviz-0.10.1\n",
      "\u001b[33mYou are using pip version 9.0.1, however version 19.0.3 is available.\n",
      "You should consider upgrading via the 'pip install --upgrade pip' command.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "! pip install graphviz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import graphviz\n",
    "from sklearn.tree import export_graphviz\n",
    "\n",
    "def export_graph(classifier, feature_names):\n",
    "    \"\"\"\n",
    "    First, export the dot data from the fitted classifier. Then, create a graphviz Source object from the dot data,\n",
    "    and return it.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    classifier : DecisionTreeClassifier\n",
    "        The single decision tree you created and fit above.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    graph : graphviz Source object\n",
    "        The Source object created with the graph information.\n",
    "    \"\"\"\n",
    "    \n",
    "    dot_data = export_graphviz(\n",
    "    clf.estimators_[0],\n",
    "    out_file=None,\n",
    "    feature_names=features,\n",
    "    filled=True, rounded=True,\n",
    "    special_characters=True,\n",
    "    rotate=False\n",
    "    )\n",
    "    graph = graphviz.Source(dot_data);\n",
    "    return graph\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
       "<!-- Generated by graphviz version 2.38.0 (20140413.2041)\n",
       " -->\n",
       "<!-- Title: Tree Pages: 1 -->\n",
       "<svg width=\"1595pt\" height=\"373pt\"\n",
       " viewBox=\"0.00 0.00 1594.50 373.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 369)\">\n",
       "<title>Tree</title>\n",
       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-369 1590.5,-369 1590.5,4 -4,4\"/>\n",
       "<!-- 0 -->\n",
       "<g id=\"node1\" class=\"node\"><title>0</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.047059\" stroke=\"black\" d=\"M887.5,-365C887.5,-365 690.5,-365 690.5,-365 684.5,-365 678.5,-359 678.5,-353 678.5,-353 678.5,-309 678.5,-309 678.5,-303 684.5,-297 690.5,-297 690.5,-297 887.5,-297 887.5,-297 893.5,-297 899.5,-303 899.5,-309 899.5,-309 899.5,-353 899.5,-353 899.5,-359 893.5,-365 887.5,-365\"/>\n",
       "<text text-anchor=\"start\" x=\"749.5\" y=\"-349.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">D ≤ &#45;0.419</text>\n",
       "<text text-anchor=\"start\" x=\"731\" y=\"-334.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.991</text>\n",
       "<text text-anchor=\"start\" x=\"733\" y=\"-319.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 1896</text>\n",
       "<text text-anchor=\"start\" x=\"686.5\" y=\"-304.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [768, 875, 658, 699]</text>\n",
       "</g>\n",
       "<!-- 1 -->\n",
       "<g id=\"node2\" class=\"node\"><title>1</title>\n",
       "<path fill=\"#e58139\" fill-opacity=\"0.325490\" stroke=\"black\" d=\"M723,-261C723,-261 535,-261 535,-261 529,-261 523,-255 523,-249 523,-249 523,-205 523,-205 523,-199 529,-193 535,-193 535,-193 723,-193 723,-193 729,-193 735,-199 735,-205 735,-205 735,-249 735,-249 735,-255 729,-261 723,-261\"/>\n",
       "<text text-anchor=\"start\" x=\"590.5\" y=\"-245.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">A ≤ &#45;0.853</text>\n",
       "<text text-anchor=\"start\" x=\"571\" y=\"-230.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.518</text>\n",
       "<text text-anchor=\"start\" x=\"577.5\" y=\"-215.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 811</text>\n",
       "<text text-anchor=\"start\" x=\"531\" y=\"-200.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [678, 389, 181, 25]</text>\n",
       "</g>\n",
       "<!-- 0&#45;&gt;1 -->\n",
       "<g id=\"edge1\" class=\"edge\"><title>0&#45;&gt;1</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M737.053,-296.884C721.885,-287.214 705.178,-276.563 689.529,-266.587\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"691.274,-263.549 680.96,-261.125 687.511,-269.452 691.274,-263.549\"/>\n",
       "<text text-anchor=\"middle\" x=\"686.364\" y=\"-281.834\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">True</text>\n",
       "</g>\n",
       "<!-- 8 -->\n",
       "<g id=\"node9\" class=\"node\"><title>8</title>\n",
       "<path fill=\"#d739e5\" fill-opacity=\"0.152941\" stroke=\"black\" d=\"M1043,-261C1043,-261 855,-261 855,-261 849,-261 843,-255 843,-249 843,-249 843,-205 843,-205 843,-199 849,-193 855,-193 855,-193 1043,-193 1043,-193 1049,-193 1055,-199 1055,-205 1055,-205 1055,-249 1055,-249 1055,-255 1049,-261 1043,-261\"/>\n",
       "<text text-anchor=\"start\" x=\"910.5\" y=\"-245.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">A ≤ &#45;0.629</text>\n",
       "<text text-anchor=\"start\" x=\"891\" y=\"-230.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.779</text>\n",
       "<text text-anchor=\"start\" x=\"893\" y=\"-215.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 1085</text>\n",
       "<text text-anchor=\"start\" x=\"851\" y=\"-200.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [90, 486, 477, 674]</text>\n",
       "</g>\n",
       "<!-- 0&#45;&gt;8 -->\n",
       "<g id=\"edge8\" class=\"edge\"><title>0&#45;&gt;8</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M840.947,-296.884C856.115,-287.214 872.822,-276.563 888.471,-266.587\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"890.489,-269.452 897.04,-261.125 886.726,-263.549 890.489,-269.452\"/>\n",
       "<text text-anchor=\"middle\" x=\"891.636\" y=\"-281.834\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">False</text>\n",
       "</g>\n",
       "<!-- 2 -->\n",
       "<g id=\"node3\" class=\"node\"><title>2</title>\n",
       "<path fill=\"#e58139\" fill-opacity=\"0.917647\" stroke=\"black\" d=\"M349,-157C349,-157 197,-157 197,-157 191,-157 185,-151 185,-145 185,-145 185,-101 185,-101 185,-95 191,-89 197,-89 197,-89 349,-89 349,-89 355,-89 361,-95 361,-101 361,-101 361,-145 361,-145 361,-151 355,-157 349,-157\"/>\n",
       "<text text-anchor=\"start\" x=\"237\" y=\"-141.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">B ≤ 0.537</text>\n",
       "<text text-anchor=\"start\" x=\"215\" y=\"-126.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 0.391</text>\n",
       "<text text-anchor=\"start\" x=\"221.5\" y=\"-111.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 326</text>\n",
       "<text text-anchor=\"start\" x=\"193\" y=\"-96.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [481, 40, 0, 0]</text>\n",
       "</g>\n",
       "<!-- 1&#45;&gt;2 -->\n",
       "<g id=\"edge2\" class=\"edge\"><title>1&#45;&gt;2</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M522.67,-195.535C474.564,-181.751 418.072,-165.566 371.004,-152.08\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"371.687,-148.635 361.109,-149.245 369.759,-155.364 371.687,-148.635\"/>\n",
       "</g>\n",
       "<!-- 5 -->\n",
       "<g id=\"node6\" class=\"node\"><title>5</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.274510\" stroke=\"black\" d=\"M723,-157C723,-157 535,-157 535,-157 529,-157 523,-151 523,-145 523,-145 523,-101 523,-101 523,-95 529,-89 535,-89 535,-89 723,-89 723,-89 729,-89 735,-95 735,-101 735,-101 735,-145 735,-145 735,-151 729,-157 723,-157\"/>\n",
       "<text text-anchor=\"start\" x=\"590.5\" y=\"-141.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">C ≤ &#45;1.713</text>\n",
       "<text text-anchor=\"start\" x=\"571\" y=\"-126.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.678</text>\n",
       "<text text-anchor=\"start\" x=\"577.5\" y=\"-111.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 485</text>\n",
       "<text text-anchor=\"start\" x=\"531\" y=\"-96.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [197, 349, 181, 25]</text>\n",
       "</g>\n",
       "<!-- 1&#45;&gt;5 -->\n",
       "<g id=\"edge5\" class=\"edge\"><title>1&#45;&gt;5</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M629,-192.884C629,-184.778 629,-175.982 629,-167.472\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"632.5,-167.299 629,-157.299 625.5,-167.299 632.5,-167.299\"/>\n",
       "</g>\n",
       "<!-- 3 -->\n",
       "<g id=\"node4\" class=\"node\"><title>3</title>\n",
       "<path fill=\"#e58139\" fill-opacity=\"0.976471\" stroke=\"black\" d=\"M164,-53C164,-53 12,-53 12,-53 6,-53 1.42109e-14,-47 1.42109e-14,-41 1.42109e-14,-41 1.42109e-14,-12 1.42109e-14,-12 1.42109e-14,-6 6,-0 12,-0 12,-0 164,-0 164,-0 170,-0 176,-6 176,-12 176,-12 176,-41 176,-41 176,-47 170,-53 164,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"30\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 0.167</text>\n",
       "<text text-anchor=\"start\" x=\"36.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 308</text>\n",
       "<text text-anchor=\"start\" x=\"8\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [474, 12, 0, 0]</text>\n",
       "</g>\n",
       "<!-- 2&#45;&gt;3 -->\n",
       "<g id=\"edge3\" class=\"edge\"><title>2&#45;&gt;3</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M208.296,-88.9485C188.417,-78.794 166.649,-67.675 147.161,-57.7203\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"148.588,-54.5188 138.09,-53.0867 145.404,-60.7526 148.588,-54.5188\"/>\n",
       "</g>\n",
       "<!-- 4 -->\n",
       "<g id=\"node5\" class=\"node\"><title>4</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.749020\" stroke=\"black\" d=\"M340,-53C340,-53 206,-53 206,-53 200,-53 194,-47 194,-41 194,-41 194,-12 194,-12 194,-6 200,-0 206,-0 206,-0 340,-0 340,-0 346,-0 352,-6 352,-12 352,-12 352,-41 352,-41 352,-47 346,-53 340,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"215\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 0.722</text>\n",
       "<text text-anchor=\"start\" x=\"226\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 18</text>\n",
       "<text text-anchor=\"start\" x=\"202\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [7, 28, 0, 0]</text>\n",
       "</g>\n",
       "<!-- 2&#45;&gt;4 -->\n",
       "<g id=\"edge4\" class=\"edge\"><title>2&#45;&gt;4</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M273,-88.9485C273,-80.7153 273,-71.848 273,-63.4814\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"276.5,-63.2367 273,-53.2367 269.5,-63.2367 276.5,-63.2367\"/>\n",
       "</g>\n",
       "<!-- 6 -->\n",
       "<g id=\"node7\" class=\"node\"><title>6</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.309804\" stroke=\"black\" d=\"M552,-53C552,-53 382,-53 382,-53 376,-53 370,-47 370,-41 370,-41 370,-12 370,-12 370,-6 376,-0 382,-0 382,-0 552,-0 552,-0 558,-0 564,-6 564,-12 564,-12 564,-41 564,-41 564,-47 558,-53 552,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"409\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.144</text>\n",
       "<text text-anchor=\"start\" x=\"415.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 266</text>\n",
       "<text text-anchor=\"start\" x=\"378\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [161, 239, 13, 0]</text>\n",
       "</g>\n",
       "<!-- 5&#45;&gt;6 -->\n",
       "<g id=\"edge6\" class=\"edge\"><title>5&#45;&gt;6</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M572.34,-88.9485C555.169,-78.9318 536.389,-67.9767 519.502,-58.1263\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"521.264,-55.1022 510.863,-53.0867 517.737,-61.1487 521.264,-55.1022\"/>\n",
       "</g>\n",
       "<!-- 7 -->\n",
       "<g id=\"node8\" class=\"node\"><title>7</title>\n",
       "<path fill=\"#399de5\" fill-opacity=\"0.254902\" stroke=\"black\" d=\"M772.5,-53C772.5,-53 593.5,-53 593.5,-53 587.5,-53 581.5,-47 581.5,-41 581.5,-41 581.5,-12 581.5,-12 581.5,-6 587.5,-0 593.5,-0 593.5,-0 772.5,-0 772.5,-0 778.5,-0 784.5,-6 784.5,-12 784.5,-12 784.5,-41 784.5,-41 784.5,-47 778.5,-53 772.5,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"629.5\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.65</text>\n",
       "<text text-anchor=\"start\" x=\"631.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 219</text>\n",
       "<text text-anchor=\"start\" x=\"589.5\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [36, 110, 168, 25]</text>\n",
       "</g>\n",
       "<!-- 5&#45;&gt;7 -->\n",
       "<g id=\"edge7\" class=\"edge\"><title>5&#45;&gt;7</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M647.887,-88.9485C652.853,-80.2579 658.222,-70.8608 663.233,-62.0917\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"666.371,-63.6557 668.293,-53.2367 660.293,-60.1826 666.371,-63.6557\"/>\n",
       "</g>\n",
       "<!-- 9 -->\n",
       "<g id=\"node10\" class=\"node\"><title>9</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.423529\" stroke=\"black\" d=\"M1043,-157C1043,-157 855,-157 855,-157 849,-157 843,-151 843,-145 843,-145 843,-101 843,-101 843,-95 849,-89 855,-89 855,-89 1043,-89 1043,-89 1049,-89 1055,-95 1055,-101 1055,-101 1055,-145 1055,-145 1055,-151 1049,-157 1043,-157\"/>\n",
       "<text text-anchor=\"start\" x=\"910.5\" y=\"-141.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">C ≤ &#45;0.669</text>\n",
       "<text text-anchor=\"start\" x=\"891\" y=\"-126.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.729</text>\n",
       "<text text-anchor=\"start\" x=\"897.5\" y=\"-111.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 437</text>\n",
       "<text text-anchor=\"start\" x=\"851\" y=\"-96.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [90, 365, 130, 100]</text>\n",
       "</g>\n",
       "<!-- 8&#45;&gt;9 -->\n",
       "<g id=\"edge9\" class=\"edge\"><title>8&#45;&gt;9</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M949,-192.884C949,-184.778 949,-175.982 949,-167.472\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"952.5,-167.299 949,-157.299 945.5,-167.299 952.5,-167.299\"/>\n",
       "</g>\n",
       "<!-- 12 -->\n",
       "<g id=\"node13\" class=\"node\"><title>12</title>\n",
       "<path fill=\"#d739e5\" fill-opacity=\"0.325490\" stroke=\"black\" d=\"M1376.5,-157C1376.5,-157 1197.5,-157 1197.5,-157 1191.5,-157 1185.5,-151 1185.5,-145 1185.5,-145 1185.5,-101 1185.5,-101 1185.5,-95 1191.5,-89 1197.5,-89 1197.5,-89 1376.5,-89 1376.5,-89 1382.5,-89 1388.5,-95 1388.5,-101 1388.5,-101 1388.5,-145 1388.5,-145 1388.5,-151 1382.5,-157 1376.5,-157\"/>\n",
       "<text text-anchor=\"start\" x=\"1248.5\" y=\"-141.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">B ≤ &#45;0.894</text>\n",
       "<text text-anchor=\"start\" x=\"1229\" y=\"-126.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.363</text>\n",
       "<text text-anchor=\"start\" x=\"1235.5\" y=\"-111.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 648</text>\n",
       "<text text-anchor=\"start\" x=\"1193.5\" y=\"-96.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [0, 121, 347, 574]</text>\n",
       "</g>\n",
       "<!-- 8&#45;&gt;12 -->\n",
       "<g id=\"edge12\" class=\"edge\"><title>8&#45;&gt;12</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M1055.02,-194.006C1093.39,-182.427 1136.82,-169.321 1175.64,-157.607\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"1176.84,-160.9 1185.4,-154.66 1174.82,-154.198 1176.84,-160.9\"/>\n",
       "</g>\n",
       "<!-- 10 -->\n",
       "<g id=\"node11\" class=\"node\"><title>10</title>\n",
       "<path fill=\"#47e539\" fill-opacity=\"0.592157\" stroke=\"black\" d=\"M975.5,-53C975.5,-53 814.5,-53 814.5,-53 808.5,-53 802.5,-47 802.5,-41 802.5,-41 802.5,-12 802.5,-12 802.5,-6 808.5,-0 814.5,-0 814.5,-0 975.5,-0 975.5,-0 981.5,-0 987.5,-6 987.5,-12 987.5,-12 987.5,-41 987.5,-41 987.5,-47 981.5,-53 975.5,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"837\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.297</text>\n",
       "<text text-anchor=\"start\" x=\"843.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 355</text>\n",
       "<text text-anchor=\"start\" x=\"810.5\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [90, 365, 94, 3]</text>\n",
       "</g>\n",
       "<!-- 9&#45;&gt;10 -->\n",
       "<g id=\"edge10\" class=\"edge\"><title>9&#45;&gt;10</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M930.113,-88.9485C925.147,-80.2579 919.778,-70.8608 914.767,-62.0917\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"917.707,-60.1826 909.707,-53.2367 911.629,-63.6557 917.707,-60.1826\"/>\n",
       "</g>\n",
       "<!-- 11 -->\n",
       "<g id=\"node12\" class=\"node\"><title>11</title>\n",
       "<path fill=\"#d739e5\" fill-opacity=\"0.627451\" stroke=\"black\" d=\"M1160.5,-53C1160.5,-53 1017.5,-53 1017.5,-53 1011.5,-53 1005.5,-47 1005.5,-41 1005.5,-41 1005.5,-12 1005.5,-12 1005.5,-6 1011.5,-0 1017.5,-0 1017.5,-0 1160.5,-0 1160.5,-0 1166.5,-0 1172.5,-6 1172.5,-12 1172.5,-12 1172.5,-41 1172.5,-41 1172.5,-47 1166.5,-53 1160.5,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"1031\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 0.842</text>\n",
       "<text text-anchor=\"start\" x=\"1042\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 82</text>\n",
       "<text text-anchor=\"start\" x=\"1013.5\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [0, 0, 36, 97]</text>\n",
       "</g>\n",
       "<!-- 9&#45;&gt;11 -->\n",
       "<g id=\"edge11\" class=\"edge\"><title>9&#45;&gt;11</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M997.965,-88.9485C1012.33,-79.2516 1028,-68.6752 1042.23,-59.073\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"1044.54,-61.7325 1050.87,-53.2367 1040.62,-55.9305 1044.54,-61.7325\"/>\n",
       "</g>\n",
       "<!-- 13 -->\n",
       "<g id=\"node14\" class=\"node\"><title>13</title>\n",
       "<path fill=\"#399de5\" fill-opacity=\"0.301961\" stroke=\"black\" d=\"M1372,-53C1372,-53 1202,-53 1202,-53 1196,-53 1190,-47 1190,-41 1190,-41 1190,-12 1190,-12 1190,-6 1196,-0 1202,-0 1202,-0 1372,-0 1372,-0 1378,-0 1384,-6 1384,-12 1384,-12 1384,-41 1384,-41 1384,-47 1378,-53 1372,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"1229\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 1.368</text>\n",
       "<text text-anchor=\"start\" x=\"1235.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 214</text>\n",
       "<text text-anchor=\"start\" x=\"1198\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [0, 119, 188, 40]</text>\n",
       "</g>\n",
       "<!-- 12&#45;&gt;13 -->\n",
       "<g id=\"edge13\" class=\"edge\"><title>12&#45;&gt;13</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M1287,-88.9485C1287,-80.7153 1287,-71.848 1287,-63.4814\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"1290.5,-63.2367 1287,-53.2367 1283.5,-63.2367 1290.5,-63.2367\"/>\n",
       "</g>\n",
       "<!-- 14 -->\n",
       "<g id=\"node15\" class=\"node\"><title>14</title>\n",
       "<path fill=\"#d739e5\" fill-opacity=\"0.698039\" stroke=\"black\" d=\"M1574.5,-53C1574.5,-53 1413.5,-53 1413.5,-53 1407.5,-53 1401.5,-47 1401.5,-41 1401.5,-41 1401.5,-12 1401.5,-12 1401.5,-6 1407.5,-0 1413.5,-0 1413.5,-0 1574.5,-0 1574.5,-0 1580.5,-0 1586.5,-6 1586.5,-12 1586.5,-12 1586.5,-41 1586.5,-41 1586.5,-47 1580.5,-53 1574.5,-53\"/>\n",
       "<text text-anchor=\"start\" x=\"1436\" y=\"-37.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">entropy = 0.803</text>\n",
       "<text text-anchor=\"start\" x=\"1442.5\" y=\"-22.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">samples = 434</text>\n",
       "<text text-anchor=\"start\" x=\"1409.5\" y=\"-7.8\" font-family=\"Helvetica,sans-Serif\" font-size=\"14.00\">value = [0, 2, 159, 534]</text>\n",
       "</g>\n",
       "<!-- 12&#45;&gt;14 -->\n",
       "<g id=\"edge14\" class=\"edge\"><title>12&#45;&gt;14</title>\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M1359.4,-88.9485C1381.94,-78.6562 1406.66,-67.3729 1428.69,-57.3155\"/>\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"1430.31,-60.4236 1437.95,-53.0867 1427.4,-54.0558 1430.31,-60.4236\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>\n"
      ],
      "text/plain": [
       "<graphviz.files.Source at 0x7fa4ae34b160>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the single decision tree graph in the notebook\n",
    "graph = export_graph(clf, features)\n",
    "graph"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Redundant labels\n",
    "Let's see what happens when we train a full random forest on these redundant data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def instantiate_and_fit_a_rf(n_estimators, X_train, y_train, min_samples_leaf=5):\n",
    "    \"\"\"\n",
    "    Instantiate a random forest classifier and fit it on the training data. Return the fitted classifier.\n",
    "    Make sure you use bootstrapping and calculate an out-of-bag score. Set `random_state` equal to an integer so that\n",
    "    when you fit the model again you get the same result. \n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    n_estimators : int\n",
    "        The number of trees.\n",
    "    \n",
    "    X_train : pandas DataFrame\n",
    "        The training features.\n",
    "        \n",
    "    y_train : pandas Series\n",
    "        The training labels.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    clf : RandomForestClassifier\n",
    "        The fitted classifier instance.\n",
    "    \"\"\"\n",
    "    \n",
    "    clf = RandomForestClassifier(\n",
    "        n_estimators=n_estimators,\n",
    "        max_features='sqrt',\n",
    "        min_samples_leaf=min_samples_leaf,\n",
    "        bootstrap=True,\n",
    "        oob_score=True,\n",
    "        n_jobs=-1,\n",
    "        criterion='entropy',\n",
    "        verbose=0,\n",
    "        random_state=0\n",
    "    )\n",
    "    clf.fit(X_train, y_train)\n",
    "    return clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3382e88f3e3d4773a3ca650a281582e1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=4), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: 1.000000, oob: 0.975667, valid: 0.664665\n",
      "train: 1.000000, oob: 0.999667, valid: 0.684685\n",
      "train: 1.000000, oob: 0.999333, valid: 0.649650\n",
      "train: 1.000000, oob: 1.000000, valid: 0.664665\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_score = []\n",
    "valid_score = []\n",
    "oob_score = []\n",
    "tree_sizes = [10, 50, 100, 250]\n",
    "\n",
    "for trees in tqdm(tree_sizes):\n",
    "    clf_red = instantiate_and_fit_a_rf(trees, X_train, y_train)\n",
    "    \n",
    "    tr, va, oob = model_score(clf_red, X_train, y_train, X_valid, y_valid)\n",
    "    train_score.append(tr); valid_score.append(va); oob_score.append(oob)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_results(tree_sizes, train_score, oob_score, valid_score, title, y_range):\n",
    "    plt.plot(tree_sizes, train_score, 'xb-');\n",
    "    plt.plot(tree_sizes, oob_score, 'xg-');\n",
    "    plt.plot(tree_sizes, valid_score, 'xr-');\n",
    "    plt.title(title);\n",
    "    plt.xlabel('Number of Trees');\n",
    "    plt.ylabel('Accuracy')\n",
    "    plt.legend(['train','oob', 'valid'])\n",
    "    plt.ylim(y_range[0], y_range[1]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_range = []\n",
    "y_range.append(0.45)\n",
    "y_range.append(1.005)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa4a9a28748>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_results(tree_sizes, train_score, oob_score, valid_score, 'Random Forest with Redundant Data (accuracy): train, validation, oob', y_range)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What you can see from this run is that the OOB score virtually tracks the training score because the OOB data contains the same information as the training data. The validation score, calculated on unseen data, is much lower."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Solution #1: use every 5th row\n",
    "In this section, implement the first solution to the redundancy issue by sub-sampling every 5th row of the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_subsampled_dataset(num_duplicates, X_train, y_train):\n",
    "    \"\"\"\n",
    "    Create the sub-sampled dataset according to the first solution proposed to the problem of overlapping labels. \n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    num_duplicates : int\n",
    "        The number of duplications made earlier.\n",
    "    \n",
    "    X_train : pandas DataFrame\n",
    "        The training features.\n",
    "        \n",
    "    y_train : pandas Series\n",
    "        The training labels.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    X_train_sub : pandas DataFrame\n",
    "        The training features, subsampled.\n",
    "        \n",
    "    y_train_sub : pandas Series\n",
    "        The training labels, subsampled.\n",
    "    \"\"\"\n",
    "    \n",
    "    X_train_sub = X_train[::5]\n",
    "    y_train_sub = y_train[::5]\n",
    "    \n",
    "    return X_train_sub, y_train_sub"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_sub, y_train_sub = create_subsampled_dataset(5, X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "443fdd4c92454e8383c471b22298c609",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=4), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: 0.946667, oob: 0.781667, valid: 0.629630\n",
      "train: 0.965000, oob: 0.835000, valid: 0.659660\n",
      "train: 0.966667, oob: 0.841667, valid: 0.669670\n",
      "train: 0.960000, oob: 0.850000, valid: 0.664665\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_score = []\n",
    "valid_score = []\n",
    "oob_score = []\n",
    "tree_sizes = [10, 50, 100, 250]\n",
    "\n",
    "for trees in tqdm(tree_sizes):\n",
    "    clf_sub = instantiate_and_fit_a_rf(trees, X_train_sub, y_train_sub)\n",
    "    \n",
    "    tr, va, oob = model_score(clf_sub, X_train_sub, y_train_sub, X_valid, y_valid)\n",
    "    train_score.append(tr); valid_score.append(va); oob_score.append(oob)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa4a91d2f60>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_results(tree_sizes, train_score, oob_score, valid_score, 'Random Forest on Redundant Data, with Subsampling (accuracy): train, validation, oob', y_range)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the case of our artificial dataset, this completely removes the redundancy."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Solution #2: reduce bag size\n",
    "In this section, implement the second solution we proposed—to use `BaggingClassifier` to randomly draw a smaller fraction of the original dataset's rows when creating each tree's dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import BaggingClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_clf = DecisionTreeClassifier(\n",
    "    criterion='entropy',\n",
    "    max_features='sqrt',\n",
    "    min_samples_leaf=5\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "def instantiate_and_fit_a_BaggingClassifier(n_estimators, base_estimator, X_train, y_train, max_samples = 0.2):\n",
    "    \"\"\"\n",
    "    Instantiate a Bagging Classifier and fit it on the training data. Return the fitted classifier.\n",
    "    Make sure you use bootstrapping and calculate an out-of-bag score. Set `random_state` equal to an integer so that\n",
    "    when you fit the model again you get the same result. \n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    n_estimators : int\n",
    "        The number of trees.\n",
    "        \n",
    "    base_estimator : DecisionTreeClassifier\n",
    "        The base estimator of the BaggingClassifier.\n",
    "        \n",
    "    max_samples : float\n",
    "        The percentage of the number of original rows to draw for each bag.\n",
    "    \n",
    "    X_train : pandas DataFrame\n",
    "        The training features.\n",
    "        \n",
    "    y_train : pandas Series\n",
    "        The training labels.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    clf : BaggingClassifier\n",
    "        The fitted classifier instance.\n",
    "    \"\"\"\n",
    "    \n",
    "    clf = BaggingClassifier(\n",
    "        base_estimator = base_estimator,\n",
    "        n_estimators = n_estimators,\n",
    "        max_samples = max_samples,\n",
    "        bootstrap=True,\n",
    "        oob_score=True,\n",
    "        n_jobs=-1,\n",
    "        verbose=0,\n",
    "        random_state=0\n",
    "    )\n",
    "    clf.fit(X_train, y_train)\n",
    "    return clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b51caba111594b4cbace218a4c023988",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=4), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: 0.956667, oob: 0.930333, valid: 0.604605\n",
      "train: 0.970000, oob: 0.959333, valid: 0.659660\n",
      "train: 0.976667, oob: 0.956333, valid: 0.654655\n",
      "train: 0.976667, oob: 0.962333, valid: 0.659660\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_score = []\n",
    "valid_score = []\n",
    "oob_score = []\n",
    "tree_sizes = [10, 50, 100, 250]\n",
    "\n",
    "for trees in tqdm(tree_sizes):\n",
    "    clf_bag = instantiate_and_fit_a_BaggingClassifier(trees, base_clf, X_train, y_train)\n",
    "    \n",
    "    tr, va, oob = model_score(clf_bag, X_train, y_train, X_valid, y_valid)\n",
    "    train_score.append(tr); valid_score.append(va); oob_score.append(oob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa4a919ea20>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_results(tree_sizes, train_score, oob_score, valid_score, 'Random Forest on Redundant Data, with Reduced Bag Size (accuracy): train, validation, oob', y_range)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, in the case of this small artificial dataset, this helped but didn't fully resolve the problem. The OOB score still vastly overestimates the validation score."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Solution #3: bagged non-overlapping labels\n",
    "Finally, let's look at the last solution we proposed to this problem. We'll do this one for you for this small exercise, but you'll have the opportunity to work on it yourself later. We will fit separate models on each non-redundant subset of data, and then ensemble them together."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import VotingClassifier\n",
    "from sklearn.base import clone\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.utils.validation import has_fit_parameter, check_is_fitted\n",
    "from sklearn.utils.metaestimators import _BaseComposition\n",
    "from sklearn.utils import Bunch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "class NoOverlapVoter(VotingClassifier):\n",
    "    def __init__(self, base_estimator, overlap_increment=5):\n",
    "        self.est_list = []\n",
    "        for i in range(overlap_increment):\n",
    "            self.est_list.append(('clf'+str(i), clone(base_estimator)))\n",
    "        self.overlap_increment = overlap_increment\n",
    "        super().__init__(\n",
    "            self.est_list,\n",
    "            voting='soft'\n",
    "        )\n",
    "    \n",
    "    @property\n",
    "    def oob_score_(self):\n",
    "        oob = 0\n",
    "        for clf in self.estimators_:\n",
    "            oob = oob + clf.oob_score_\n",
    "        return oob / len(self.estimators_)\n",
    "    \n",
    "    def fit(self, X, y, sample_weight=None):\n",
    "        names, clfs = zip(*self.estimators)\n",
    "        self._validate_names(names)\n",
    "\n",
    "        self.le_ = LabelEncoder().fit(y)\n",
    "        self.classes_ = self.le_.classes_\n",
    "        self.estimators_ = []\n",
    "\n",
    "        transformed_y = self.le_.transform(y)\n",
    "\n",
    "        self.estimators_ = []\n",
    "        for i in range(self.overlap_increment):\n",
    "            self.estimators_.append(\n",
    "                clfs[i].fit(X[i::self.overlap_increment], transformed_y[i::self.overlap_increment])\n",
    "            )\n",
    "        \n",
    "        self.named_estimators_ = Bunch(**dict())\n",
    "        for k, e in zip(self.estimators, self.estimators_):\n",
    "            self.named_estimators_[k[0]] = e\n",
    "        return self"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5a3c6cc0076048a19cc648c607b82c84",
       "version_major": 2,
       "version_minor": 0
      },
      "text/html": [
       "<p>Failed to display Jupyter Widget of type <code>HBox</code>.</p>\n",
       "<p>\n",
       "  If you're reading this message in the Jupyter Notebook or JupyterLab Notebook, it may mean\n",
       "  that the widgets JavaScript is still loading. If this message persists, it\n",
       "  likely means that the widgets JavaScript library is either not installed or\n",
       "  not enabled. See the <a href=\"https://ipywidgets.readthedocs.io/en/stable/user_install.html\">Jupyter\n",
       "  Widgets Documentation</a> for setup instructions.\n",
       "</p>\n",
       "<p>\n",
       "  If you're reading this message in another frontend (for example, a static\n",
       "  rendering on GitHub or <a href=\"https://nbviewer.jupyter.org/\">NBViewer</a>),\n",
       "  it may mean that your frontend doesn't currently support widgets.\n",
       "</p>\n"
      ],
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=4), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:453: UserWarning: Some inputs do not have OOB scores. This probably means too few trees were used to compute any reliable oob estimates.\n",
      "  warn(\"Some inputs do not have OOB scores. \"\n",
      "/opt/conda/lib/python3.6/site-packages/sklearn/ensemble/forest.py:458: RuntimeWarning: invalid value encountered in true_divide\n",
      "  predictions[k].sum(axis=1)[:, np.newaxis])\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: 0.946667, oob: 0.781667, valid: 0.629630\n",
      "train: 0.965000, oob: 0.835000, valid: 0.659660\n",
      "train: 0.966667, oob: 0.841667, valid: 0.669670\n",
      "train: 0.960000, oob: 0.850000, valid: 0.664665\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_score = []\n",
    "valid_score = []\n",
    "oob_score = []\n",
    "tree_sizes = [10, 50, 100, 250]\n",
    "\n",
    "for trees in tqdm(tree_sizes):\n",
    "    clf = RandomForestClassifier(\n",
    "        n_estimators=trees,\n",
    "        max_features='sqrt',\n",
    "        min_samples_leaf=5,\n",
    "        bootstrap=True,\n",
    "        oob_score=True,\n",
    "        n_jobs=-1,\n",
    "        criterion='entropy',\n",
    "        verbose=0,\n",
    "        random_state=0\n",
    "    )\n",
    "    \n",
    "    clf_nov = NoOverlapVoter(clf)\n",
    "    clf_nov.fit(X_train.reset_index()[['A','B','C','D']], y_train)\n",
    "    t, v, o = model_score(clf_nov, X_train, y_train, X_valid, y_valid)\n",
    "    train_score.append(t); valid_score.append(v); oob_score.append(o)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa4a8035ba8>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_results(tree_sizes, train_score, oob_score, valid_score, 'Random Forest on Redundant Data, with NoOverlapVoter (accuracy): train, validation, oob', y_range)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So in the case of this artificial dataset, this method performs as well as the first method, but this makes sense because each of the separate classifiers has the exact same information as the others. Combining them together should not yield much performance enhancement. You can then compare this result to the result that you'll see in the project, when this method is applied to real financial data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Check the performance on test set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Finally, let's look at the performance of the third method on the held-out test set. For this, we'll train on all of the previous training and validation data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "X, X_train, X_valid, X_test, y_train, y_valid, y_test = make_splits(\n",
    "    redundant_df,\n",
    "    features,\n",
    "    'E',\n",
    "    split_valid = 0\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train: 0.963750, oob: 0.856250, valid: 0.694695\n"
     ]
    }
   ],
   "source": [
    "clf = RandomForestClassifier(\n",
    "        n_estimators=150,\n",
    "        max_features='sqrt',\n",
    "        min_samples_leaf=5,\n",
    "        bootstrap=True,\n",
    "        oob_score=True,\n",
    "        n_jobs=-1,\n",
    "        criterion='entropy',\n",
    "        verbose=False,\n",
    "        random_state=0\n",
    "    )\n",
    "    \n",
    "clf_nov = NoOverlapVoter(clf)\n",
    "clf_nov.fit(X_train.reset_index()[['A','B','C','D']], y_train)\n",
    "t, v, o = model_score(clf_nov, X_train, y_train, X_test, y_test)\n",
    "train_score.append(t); valid_score.append(v); oob_score.append(o)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "We see that the result on the held-out test set has improved somewhat."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}