credCardConfIntV1g_import.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# First Part is from \"How To Deal With Imbalanced Classification, Without Re-balancing the Data\"  by David Rosen  \n",
    "\n",
    "https://towardsdatascience.com/how-to-deal-with-imbalanced-classification-without-re-balancing-the-data-8a3c02353fe3\n",
    "## Second Part is down below"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2021-09-08 12:27:36.673224\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 284807 entries, 0 to 284806\n",
      "Data columns (total 31 columns):\n",
      " #   Column  Non-Null Count   Dtype  \n",
      "---  ------  --------------   -----  \n",
      " 0   Time    284807 non-null  float64\n",
      " 1   V1      284807 non-null  float64\n",
      " 2   V2      284807 non-null  float64\n",
      " 3   V3      284807 non-null  float64\n",
      " 4   V4      284807 non-null  float64\n",
      " 5   V5      284807 non-null  float64\n",
      " 6   V6      284807 non-null  float64\n",
      " 7   V7      284807 non-null  float64\n",
      " 8   V8      284807 non-null  float64\n",
      " 9   V9      284807 non-null  float64\n",
      " 10  V10     284807 non-null  float64\n",
      " 11  V11     284807 non-null  float64\n",
      " 12  V12     284807 non-null  float64\n",
      " 13  V13     284807 non-null  float64\n",
      " 14  V14     284807 non-null  float64\n",
      " 15  V15     284807 non-null  float64\n",
      " 16  V16     284807 non-null  float64\n",
      " 17  V17     284807 non-null  float64\n",
      " 18  V18     284807 non-null  float64\n",
      " 19  V19     284807 non-null  float64\n",
      " 20  V20     284807 non-null  float64\n",
      " 21  V21     284807 non-null  float64\n",
      " 22  V22     284807 non-null  float64\n",
      " 23  V23     284807 non-null  float64\n",
      " 24  V24     284807 non-null  float64\n",
      " 25  V25     284807 non-null  float64\n",
      " 26  V26     284807 non-null  float64\n",
      " 27  V27     284807 non-null  float64\n",
      " 28  V28     284807 non-null  float64\n",
      " 29  Amount  284807 non-null  float64\n",
      " 30  Class   284807 non-null  int64  \n",
      "dtypes: float64(30), int64(1)\n",
      "memory usage: 67.4 MB\n"
     ]
    }
   ],
   "source": [
    "import datetime\n",
    "print(datetime.datetime.now())\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import re\n",
    "from sklearn import metrics, model_selection\n",
    "from sklearn.experimental import enable_hist_gradient_boosting\n",
    "from sklearn.ensemble import HistGradientBoostingClassifier\n",
    "df=pd.read_csv('creditcard.csv')\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(492, 0.001727485630620034)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Class.sum(),df.Class.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## split into train and test and train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "Xtrain, Xtest, ytrain, ytest = model_selection.train_test_split(\n",
    "        df.loc[:,'V1':'Amount'], df.Class,  stratify=df.Class, \n",
    "        test_size=0.3, random_state=42)\n",
    "gbc=HistGradientBoostingClassifier(learning_rate=0.01, \n",
    "        max_iter=2000, max_leaf_nodes=6, validation_fraction=0.2, \n",
    "        n_iter_no_change=15, random_state=42).fit(Xtrain,ytrain)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[pandas.core.frame.DataFrame,\n",
       " pandas.core.frame.DataFrame,\n",
       " pandas.core.series.Series,\n",
       " pandas.core.series.Series]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[type(x) for x in (Xtrain, Xtest, ytrain, ytest)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(249927    0\n",
       " 214082    0\n",
       " 106005    0\n",
       " 58619     0\n",
       " 191638    0\n",
       "          ..\n",
       " 3441      0\n",
       " 233802    0\n",
       " 85418     0\n",
       " 29062     0\n",
       " 17677     0\n",
       " Name: Class, Length: 199364, dtype: int64,\n",
       " 186882    0\n",
       " 165754    0\n",
       " 235285    0\n",
       " 101271    0\n",
       " 5832      0\n",
       "          ..\n",
       " 64725     0\n",
       " 16688     0\n",
       " 183868    0\n",
       " 45011     0\n",
       " 150527    0\n",
       " Name: Class, Length: 85443, dtype: int64)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ytrain,ytest"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## apply the full model to the test data as a hard classifier with (implicit) decision threshold 0.5\n",
    "\n",
    "In other words, this classifier tells us which class is more probable than the other according to the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>Pred 0(Approve as Legit)</th>\n",
       "      <th>Pred 1(Deny as Fraud)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>True 0(Legit)</th>\n",
       "      <td>TN = 85285 (TNR = 99.99%)</td>\n",
       "      <td>FP = 10 (FPR = 0.01%)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>True 1(Fraud)</th>\n",
       "      <td>FN = 42 (FNR = 28.38%)</td>\n",
       "      <td>TP = 106 (TPR = 71.62%)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                Pred 0(Approve as Legit)    Pred 1(Deny as Fraud)\n",
       "True 0(Legit)  TN = 85285 (TNR = 99.99%)    FP = 10 (FPR = 0.01%)\n",
       "True 1(Fraud)     FN = 42 (FNR = 28.38%)  TP = 106 (TPR = 71.62%)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hardpredtst=gbc.predict(Xtest)\n",
    "def conf_matrix(y,pred):\n",
    "    ((tn, fp), (fn, tp)) = metrics.confusion_matrix(y, pred)\n",
    "    ((tnr,fpr),(fnr,tpr))= metrics.confusion_matrix(y, pred, \n",
    "            normalize='true')\n",
    "    return pd.DataFrame([[f'TN = {tn} (TNR = {tnr:1.2%})', \n",
    "                                f'FP = {fp} (FPR = {fpr:1.2%})'], \n",
    "                         [f'FN = {fn} (FNR = {fnr:1.2%})', \n",
    "                                f'TP = {tp} (TPR = {tpr:1.2%})']],\n",
    "            index=['True 0(Legit)', 'True 1(Fraud)'], \n",
    "            columns=['Pred 0(Approve as Legit)', \n",
    "                            'Pred 1(Deny as Fraud)'])\n",
    "conf_matrix(ytest,hardpredtst)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The true pos rate (Recall of class 1 (frauds)) is too low -- we want to identify more than 72% of frauds as such"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ultimately it is a business decision what threshold to apply when deciding whether to approve a transaction, as it is a trade-off between false approvals and false rejections."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## plot the true positive rate and false positive rate as a function of threshold to manage the trade-off"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "predtst=gbc.predict_proba(Xtest)[:,1]\n",
    "fpr, tpr, thresholds = metrics.roc_curve(ytest, predtst)\n",
    "dfplot=pd.DataFrame({'Threshold':thresholds, \n",
    "        'False Positive Rate':fpr, \n",
    "        'False Negative Rate': 1.-tpr})\n",
    "ax=dfplot.plot(x='Threshold', y=['False Positive Rate',\n",
    "        'False Negative Rate'], figsize=(10,6))\n",
    "ax.plot([0.00035,0.00035],[0,0.1]) #mark example thresh.\n",
    "ax.set_xbound(0,0.0008); ax.set_ybound(0,0.3) #zoom in"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## choose a threshold of 0.00035 (0.035%) (vertical green line shown) which will give us a false negative (approved fraud) rate of about 9% and a false positive (rejected legitimate) rate of about 6%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Pred 0(Approve as Legit)</th>\n",
       "      <th>Pred 1(Deny as Fraud)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>True 0(Legit)</th>\n",
       "      <td>TN = 80388 (TNR = 94.25%)</td>\n",
       "      <td>FP = 4907 (FPR = 5.75%)</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>True 1(Fraud)</th>\n",
       "      <td>FN = 14 (FNR = 9.46%)</td>\n",
       "      <td>TP = 134 (TPR = 90.54%)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                Pred 0(Approve as Legit)    Pred 1(Deny as Fraud)\n",
       "True 0(Legit)  TN = 80388 (TNR = 94.25%)  FP = 4907 (FPR = 5.75%)\n",
       "True 1(Fraud)      FN = 14 (FNR = 9.46%)  TP = 134 (TPR = 90.54%)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hardpredtst_tuned_thresh = np.where(predtst >= 0.00035, 1, 0)\n",
    "conf_matrix(ytest, hardpredtst_tuned_thresh)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sure enough, False neg rate of 9.46% (i.e. identified/denied 90.54% of our frauds) and false pos rt of 5.75% (i.e. still approved 94.25% of our legitimate transactions)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# This second part is for \"Get confidence intervals for model performance, like “92.4% [89.6% to 94.7%]”, for metrics like Accuracy, Recall, Precision, F1, and others\", for any metric such as Accuracy, Recall, Precision, Specificity, F1 and others\" by David Rosen \n",
    "https://towardsdatascience.com/get-confidence-intervals-for-any-model-performance-metrics-in-machine-learning-f9e72a3becb2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Original test set and two generated bootstrap sample datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from boot_conf_intervals_ml import specificity_score, \\\n",
    "         make_boot_df, raw_metric_samples, ci\n",
    "# boot_conf_intervals_ml.py just concatenates the three gists below\n",
    "\n",
    "# gist make_df_and_boot.py \n",
    "# https://gist.github.com/DavidRosen/3aeedbd2ccb73e636e7eb22657373b2b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th colspan=\"2\" halign=\"left\">Orig</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Boot0</th>\n",
       "      <th colspan=\"2\" halign=\"left\">Boot1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>Class</th>\n",
       "      <th>PredClass</th>\n",
       "      <th>Class</th>\n",
       "      <th>PredClass</th>\n",
       "      <th>Class</th>\n",
       "      <th>PredClass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\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>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>2</th>\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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\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>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>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85438</th>\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>85439</th>\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>85440</th>\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>85441</th>\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>85442</th>\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",
       "  </tbody>\n",
       "</table>\n",
       "<p>85443 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Orig           Boot0           Boot1          \n",
       "      Class PredClass Class PredClass Class PredClass\n",
       "0         0         0     0         0     0         0\n",
       "1         0         0     0         0     0         0\n",
       "2         0         0     0         1     0         0\n",
       "3         0         0     0         0     0         0\n",
       "4         0         0     0         0     0         0\n",
       "...     ...       ...   ...       ...   ...       ...\n",
       "85438     0         0     0         0     0         0\n",
       "85439     0         1     0         0     0         0\n",
       "85440     0         0     0         0     0         0\n",
       "85441     0         0     0         0     0         0\n",
       "85442     0         0     0         0     0         0\n",
       "\n",
       "[85443 rows x 6 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# reset random row nums from #train_test_split() #before #concat!\n",
    "ytest=ytest.reset_index(drop=True) #else #concat #ignore_index=True\n",
    "hardpredtst_tuned_thresh = pd.Series\\\n",
    "  (hardpredtst_tuned_thresh, name='PredClass') #was #Numpy #array\n",
    "dforig = pd.concat([ytest, hardpredtst_tuned_thresh], axis=1)\n",
    "\n",
    "np.random.seed(13)  # to get same #sample #datasets #every #time\n",
    "df0, df1 = make_boot_df(dforig), make_boot_df(dforig)\n",
    "pd.concat([dforig,df0,df1], axis=1, keys=['Orig','Boot0','Boot1'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making each boot sample dataset and calculating metrics for it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# gist raw_metric_samples.py https://gist.github.com/DavidRosen/4c80d1e295c39c089a62630fef878e26"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████████████████| 10/10 [00:01<00:00,  5.78it/s]\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style  type=\"text/css\" >\n",
       "</style><table id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59\" ><thead>    <tr>        <th class=\"blank level0\" ></th>        <th class=\"col_heading level0 col0\" >0</th>        <th class=\"col_heading level0 col1\" >1</th>        <th class=\"col_heading level0 col2\" >2</th>        <th class=\"col_heading level0 col3\" >3</th>        <th class=\"col_heading level0 col4\" >4</th>        <th class=\"col_heading level0 col5\" >5</th>        <th class=\"col_heading level0 col6\" >6</th>        <th class=\"col_heading level0 col7\" >7</th>        <th class=\"col_heading level0 col8\" >8</th>        <th class=\"col_heading level0 col9\" >9</th>    </tr>    <tr>        <th class=\"index_name level0\" >Metric (class 1 +ve)</th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>    </tr></thead><tbody>\n",
       "                <tr>\n",
       "                        <th id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59level0_row0\" class=\"row_heading level0 row0\" >Recall</th>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col0\" class=\"data row0 col0\" >93.46%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col1\" class=\"data row0 col1\" >89.81%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col2\" class=\"data row0 col2\" >96.62%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col3\" class=\"data row0 col3\" >89.38%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col4\" class=\"data row0 col4\" >89.57%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col5\" class=\"data row0 col5\" >89.68%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col6\" class=\"data row0 col6\" >82.82%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col7\" class=\"data row0 col7\" >89.74%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col8\" class=\"data row0 col8\" >88.07%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row0_col9\" class=\"data row0 col9\" >85.71%</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59level0_row1\" class=\"row_heading level0 row1\" >Specificity (Recall of −ve)</th>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col0\" class=\"data row1 col0\" >94.29%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col1\" class=\"data row1 col1\" >94.14%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col2\" class=\"data row1 col2\" >94.29%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col3\" class=\"data row1 col3\" >94.35%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col4\" class=\"data row1 col4\" >94.27%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col5\" class=\"data row1 col5\" >94.09%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col6\" class=\"data row1 col6\" >94.28%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col7\" class=\"data row1 col7\" >94.21%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col8\" class=\"data row1 col8\" >94.14%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row1_col9\" class=\"data row1 col9\" >94.22%</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59level0_row2\" class=\"row_heading level0 row2\" >Balanced Accuracy</th>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col0\" class=\"data row2 col0\" >93.88%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col1\" class=\"data row2 col1\" >91.98%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col2\" class=\"data row2 col2\" >95.45%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col3\" class=\"data row2 col3\" >91.86%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col4\" class=\"data row2 col4\" >91.92%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col5\" class=\"data row2 col5\" >91.88%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col6\" class=\"data row2 col6\" >88.55%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col7\" class=\"data row2 col7\" >91.98%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col8\" class=\"data row2 col8\" >91.11%</td>\n",
       "                        <td id=\"T_d4a35fb6_10c1_11ec_a58e_e470b8cd6e59row2_col9\" class=\"data row2 col9\" >89.97%</td>\n",
       "            </tr>\n",
       "    </tbody></table>"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x2a175987cc8>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "met=[ metrics.recall_score, specificity_score, \n",
    "      metrics.balanced_accuracy_score\n",
    "    ]\n",
    "np.random.seed(13)\n",
    "raw_metric_samples(met, ytest, hardpredtst_tuned_thresh, \n",
    "         nboots=10).style.format('{:.2%}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Histogram of bootstrap sample datasets showing confidence interval for Recall"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 500/500 [00:18<00:00, 26.98it/s]\n"
     ]
    }
   ],
   "source": [
    "metr = metrics.recall_score\n",
    "point = metr(ytest, hardpredtst_tuned_thresh)\n",
    "np.random.seed(13)\n",
    "data = raw_metric_samples(metr, ytest, hardpredtst_tuned_thresh, \n",
    "               nboots=500).transpose()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1500x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "(lower, upper) = data.quantile([0.025, 0.975]).iloc[:,0]\n",
    "\n",
    "import matplotlib, matplotlib.ticker as mtick\n",
    "import seaborn as sns; sns.set_style('whitegrid')  #optional\n",
    "matplotlib.rcParams[\"figure.dpi\"] = 300\n",
    "ax = data.hist(bins=50, figsize=(5, 2), alpha=0.4)[0][0]\n",
    "ax.xaxis.set_major_formatter( mtick.PercentFormatter(1.0))\n",
    "for x in lower, point, upper:\n",
    "    ax.plot([x, x],[0, 30], lw=2.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## How to calculate all the confidence intervals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# gist ci.py \n",
    "# https://gist.github.com/DavidRosen/c85a2d075f64e0c9fd02e5bfbc968eb0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Calculate the results!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████| 500/500 [01:12<00:00,  6.94it/s]\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<style  type=\"text/css\" >\n",
       "</style><table id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59\" ><thead>    <tr>        <th class=\"index_name level0\" >500 Boot Samples</th>        <th class=\"col_heading level0 col0\" >Point Estim</th>        <th class=\"col_heading level0 col1\" >2.5%ile</th>        <th class=\"col_heading level0 col2\" >97.5%ile</th>    </tr>    <tr>        <th class=\"index_name level0\" >Metric (class 1 +ve)</th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>        <th class=\"blank\" ></th>    </tr></thead><tbody>\n",
       "                <tr>\n",
       "                        <th id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59level0_row0\" class=\"row_heading level0 row0\" >Recall</th>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row0_col0\" class=\"data row0 col0\" >90.54%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row0_col1\" class=\"data row0 col1\" >85.00%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row0_col2\" class=\"data row0 col2\" >95.25%</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59level0_row1\" class=\"row_heading level0 row1\" >Specificity (Recall of −ve)</th>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row1_col0\" class=\"data row1 col0\" >94.25%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row1_col1\" class=\"data row1 col1\" >94.11%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row1_col2\" class=\"data row1 col2\" >94.40%</td>\n",
       "            </tr>\n",
       "            <tr>\n",
       "                        <th id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59level0_row2\" class=\"row_heading level0 row2\" >Balanced Accuracy</th>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row2_col0\" class=\"data row2 col0\" >92.39%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row2_col1\" class=\"data row2 col1\" >89.61%</td>\n",
       "                        <td id=\"T_0b181bf4_10c2_11ec_afb5_e470b8cd6e59row2_col2\" class=\"data row2 col2\" >94.73%</td>\n",
       "            </tr>\n",
       "    </tbody></table>"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x2a1769cc508>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.random.seed(13)\n",
    "ci(met,ytest, hardpredtst_tuned_thresh).style.format('{:.2%}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2021-09-08 12:30:13.767097\n"
     ]
    }
   ],
   "source": [
    "print(datetime.datetime.now())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}