conformal_classifier_MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conformal inference: classification"
   ]
  },
  {
   "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 sklearn.metrics import accuracy_score, classification_report, confusion_matrix\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn import svm\n",
    "from sklearn.datasets import fetch_openml\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler, LabelEncoder\n",
    "from sklearn.svm import SVC\n",
    "from utils import print_image_by_id, plot_scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MNIST dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Fetch MNIST data without using pandas\n",
    "# mnist = fetch_openml('mnist_784', data_home='/Users/mn9lu5b/OneDrive - Allianz/AAC Use Cases', as_frame=False)\n",
    "\n",
    "# X, y = mnist.data, mnist.target"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Zalando's Fashion-MNIST dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/mn9lu5b/Library/CloudStorage/OneDrive-Allianz/Github/Private AZ Account/uncertainty_quantification/env_uq/lib/python3.11/site-packages/sklearn/datasets/_openml.py:1022: FutureWarning: The default value of `parser` will change from `'liac-arff'` to `'auto'` in 1.4. You can set `parser='auto'` to silence this warning. Therefore, an `ImportError` will be raised from 1.4 if the dataset is dense and pandas is not installed. Note that the pandas parser may return different data types. See the Notes Section in fetch_openml's API doc for details.\n",
      "  warn(\n"
     ]
    }
   ],
   "source": [
    "fashion_mnist = fetch_openml(name='Fashion-MNIST', data_home='/Users/mn9lu5b/OneDrive - Allianz/AAC Use Cases', version=1, as_frame=False)\n",
    "\n",
    "X, y = fashion_mnist.data, fashion_mnist.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(70000, 784) (70000,)\n"
     ]
    }
   ],
   "source": [
    "print(X.shape, y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#print(mnist.DESCR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Labels are strings, we need to convert them to int\n",
    "le = LabelEncoder()\n",
    "\n",
    "y = le.fit_transform(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "View the data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAK4AAADECAYAAAAGYxrSAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAT/klEQVR4nO2de1BV1fvGnwPKOVyUW4iIigqIiHc0hfzCCCogaqOmmeNE3nImHNMhc8wpJ0wHC03HvEQo/AGNqTRU1nhHJTUVL2ljDDRcTFQQEDQBscP6/dGPMx72u/AcwXTl+5lxRh7evffai4fFeddVJ4QQYBjFsHnWBWCYJ4GNyygJG5dREjYuoyRsXEZJ2LiMkrBxGSVh4zJKwsZllISN+x9Ep9Nh0aJF7Xa/kpIS6HQ6pKent9s928oTGTc9PR06nQ55eXntXZ7njh07diAwMBAGgwH+/v7YvHlzuz/j2LFj0Ol02Lt3b7vf+3nh0KFDGD16NBwcHODq6orXXnsNJSUlT3w/bnFb4csvv8T8+fMRFBSEzZs3IyQkBIsXL8a6deueddGUYt++fYiOjsaDBw+QlJSEhIQEHD9+HKNHj8bt27ef6J4d2rmM/xnq6+uxcuVKxMbGmlrCBQsWoKmpCatXr8bbb78NV1fXZ1xKNVi+fDn69OmDkydPws7ODgAwadIkDBs2DElJSVi/fr3V92y3Fvett96Ck5MTrl27hokTJ8LJyQne3t7YsmULAODKlSuIiIiAo6MjfHx88PXXX5tdX11djffeew8DBw6Ek5MTOnfujJiYGPz666+aZ5WWlmLy5MlwdHREly5dsHTpUhw4cAA6nQ7Hjh0ziz1z5gyio6Ph7OwMBwcHhIeH4+TJk499n5ycHFRVVeGdd94x0+Pj43H//n38+OOPVtZQ20lOTkZoaCjc3d1hb2+P4ODgVj9eZGZmIiAgAAaDAcHBwThx4oQmpqysDHPnzoWnpyf0ej2CgoKwc+fOx5bl4cOHyM/Px82bN1uNq66uxtWrVzFlyhSTaQFg8ODBCAwMxK5dux77LIp2/ahgNBoRExODHj164NNPP0WvXr2waNEipKenIzo6GsOHD8e6devQqVMnvPnmmyguLjZdW1RUhOzsbEycOBEbNmzAsmXLcOXKFYSHh+PGjRumuPv37yMiIgKHDx/G4sWLsXLlSpw6dQrLly/XlOfo0aMICwvD3bt3sWrVKqxduxY1NTWIiIjA2bNnW32XixcvAgCGDx9upgcHB8PGxsb0/X+TTZs2YejQoUhMTMTatWvRoUMHTJ8+nfwlOn78OJYsWYLZs2cjMTERVVVViI6Oxm+//WaKKS8vx6hRo3D48GEsWrQImzZtgp+fH+bNm4eNGze2WpaysjIEBgZixYoVrcY9ePAAAGBvb6/5noODA27cuIFbt25Z8PYtEE9AWlqaACDOnTtn0uLi4gQAsXbtWpN2584dYW9vL3Q6ndi1a5dJz8/PFwDEqlWrTFpDQ4MwGo1mzykuLhZ6vV4kJiaatPXr1wsAIjs726TV19eLfv36CQAiJydHCCFEU1OT8Pf3F1FRUaKpqckUW1dXJ3r37i3GjRvX6jvGx8cLW1tb8nseHh5i5syZrV5vDTk5OQKA2LNnT6txdXV1Zl83NjaKAQMGiIiICDMdgAAg8vLyTFppaakwGAxiypQpJm3evHnCy8tLVFZWml0/c+ZM4ezsbHpecXGxACDS0tJMMc1aXFxcq2U2Go3CxcVFREZGmumVlZXC0dFRU05LaffkbP78+ab/u7i4ICAgAI6OjpgxY4ZJDwgIgIuLC4qKikyaXq+Hjc0/xTEajaiqqoKTkxMCAgJw4cIFU9z+/fvh7e2NyZMnmzSDwYAFCxaYlePSpUsoLCzErFmzUFVVhcrKSlRWVuL+/fuIjIzEiRMn0NTUJH2P+vp6sz9tj2IwGFBfX29hjbQfj7Zad+7cQW1tLf73v/+Z1U8zISEhCA4ONn3ds2dPvPrqqzhw4ACMRiOEEMjKysKkSZMghDDVT2VlJaKiolBbW0vet5levXpBCPHYLjIbGxssXLgQR44cwYoVK1BYWIjz589jxowZaGxsBIAnqst2Tc4MBgM8PDzMNGdnZ3Tv3h06nU6j37lzx/R1U1MTNm3ahK1bt6K4uBhGo9H0PXd3d9P/S0tL4evrq7mfn5+f2deFhYUAgLi4OGl5a2trpQmWvb29qWJb0tDQQP7pa6axsRHV1dVmmoeHB2xtbaXXWMK+ffvwySef4NKlS6Y/wQA0dQEA/v7+Gq1v376oq6vD7du3YWNjg5qaGqSkpCAlJYV8XkVFRZvK20xiYiIqKyvx6aefIikpCQAwfvx4zJs3D9u3b4eTk5PV92xX48p+MDJdPLJqaO3atfjwww8xd+5crF69Gm5ubrCxscGSJUtabRllNF/z2WefYciQIWRMaxXm5eUFo9GIiooKdOnSxaQ3NjaiqqoK3bp1k1576tQpjBkzxkwrLi5Gr169LH+BFuTm5mLy5MkICwvD1q1b4eXlhY4dOyItLU2T6FpCc/3Mnj1b+ss9aNCgJy7vo9jZ2SE1NRVr1qxBQUEBPD090bdvX8yaNQs2NjaaRscSnpvusL1792LMmDHYsWOHmV5TU4OXXnrJ9LWPjw+uXr0KIYRZS/PHH3+YXefr6wsA6Ny5M8aOHWt1eZrNnpeXhwkTJpj0vLw8NDU1SX8ZgH8y5kOHDplpXbt2tboMj5KVlQWDwYADBw5Ar9eb9LS0NDK++S/OoxQUFMDBwcH0V7FTp04wGo1PVD9PgqenJzw9PQH883Hw2LFjGDly5BO1uM/NAIStra1ZCwwAe/bsQVlZmZkWFRWFsrIyfP/99yatoaEBX331lVlccHAwfH19kZycjL/++kvzvMd1fEdERMDNzQ3btm0z07dt2wYHBwfExsZKr3V1dcXYsWPN/hkMhlaf9zhsbW2h0+nMPkKVlJQgOzubjD99+rTZZ9Q///wT3333HcaPHw9bW1vY2tpi2rRpyMrKMutpaOZx9WNpd5iM5ORk3Lx5EwkJCU90/XPT4k6cOBGJiYmYM2cOQkNDceXKFWRmZqJPnz5mcQsXLsQXX3yBN954A++++y68vLyQmZlpMkZzK2xjY4PU1FTExMQgKCgIc+bMgbe3N8rKypCTk4POnTvjhx9+kJbH3t4eq1evRnx8PKZPn46oqCjk5uYiIyMDa9asgZubW7vXQVZWFvLz8zV6XFwcYmNjsWHDBkRHR2PWrFmoqKjAli1b4Ofnh8uXL2uuGTBgAKKiorB48WLo9Xps3boVAPDxxx+bYpKSkpCTk4ORI0diwYIF6N+/P6qrq3HhwgUcPnxY8zn9UZq7w+Li4h6boGVkZCArKwthYWFwcnLC4cOHsXv3bsyfPx/Tpk2zsHZaYHU/hJB3hzk6Ompiw8PDRVBQkEb38fERsbGxpq8bGhpEQkKC8PLyEvb29uKVV14Rp0+fFuHh4SI8PNzs2qKiIhEbGyvs7e2Fh4eHSEhIEFlZWQKA+OWXX8xiL168KKZOnSrc3d2FXq8XPj4+YsaMGeLIkSMWvWtKSooICAgQdnZ2wtfXV3z++edm3WvtQXN3mOxfbm6uEEKIHTt2CH9/f6HX60W/fv1EWlqaWLVqlWj5YwQg4uPjRUZGhil+6NChpq7CRykvLxfx8fGiR48eomPHjqJr164iMjJSpKSkmGLa0h0mhBBnzpwRYWFhwtXVVRgMBjF48GCxffv2NtWj7v9fVHk2btyIpUuX4vr16/D29n7WxWGeMkoat76+3qw7qqGhAUOHDoXRaERBQcEzLBnzb/HcfMa1hqlTp6Jnz54YMmQIamtrkZGRgfz8fGRmZj7rojH/EkoaNyoqCqmpqcjMzITRaET//v2xa9cuvP7668+6aMy/hJIfFRjmuenHZRhrYOMySsLGZZSEjcsoCRuXURI2LqMkbFxGSdi4jJKwcRklYeMySsLGZZSEjcsoCRuXURI2LqMkbFxGSdi4jJKwcRklYeMySsLGZZREycWSzxJqiR61W6KMe/fukfrPP/+s0WJiYtpULgBmWzY106HD0/mxW7N80Zo6o+AWl1ESNi6jJGxcRknYuIySsHEZJeFeBSuhtvWXHRXQcpd0AEhNTSVjqTMlHB0dyVhqk+iXX36ZjLWmB4HqFZAdY0DFWvMsqrcDkNdlS7jFZZSEjcsoCRuXURI2LqMknJxZCZVUyBKKo0eParSWx0g106NHD4326CF8j1JXV6fRDh48SMa2PHETgOnIppZQw7DWHCpInW4EwHRi6KM4ODhYfF/ynm26mmGeEWxcRknYuIySsHEZJWHjMkrCvQpWYmdnZ3HsuXPnNFpJSQkZSw2tyoZbx48fr9EuXrxIxr7//vsabfjw4WTswIEDNVpgYCAZe/bsWY1GvS8AhIaGarSQkBAy1tnZmdRbwi0uoyRsXEZJ2LiMkrBxGSXhkyUlyKqFGhaVDeNSiVFNTQ0Z27FjR41GDZXKGDFiBKn7+flpNFmCSb3zrVu3yFhq7q1sTvCePXs0Wnx8PBkbERFB6i3hFpdREjYuoyRsXEZJ2LiMkrBxGSV5oXoV2uNVqV6FUaNGkbGy4V0KqmyySdx6vd7i+1IrgmX7dg0bNkyj+fv7k7FU2fbv30/GFhUVabQbN26QsZbCLS6jJGxcRknYuIySsHEZJXmh5uO2dTNhGa6urqR+8+ZNjUZttQTQK3ofPnxIxlKraakkDADq6+s1mqweqM2lT506RcZSyWR5eTkZGx0dTeptgVtcRknYuIySsHEZJWHjMkrCxmWU5IXqVXhaUHt5AfQ+Y7KVu1RvQ9euXclYd3d3jSYbXqYmo1tztBTVKyG7r2yI+vr166TeFrjFZZSEjcsoCRuXURI2LqMkL1RyJktKrDlJhxpulc0tpebNylbYNjY2WnQ9QJ/GU1tbS8ZSiZwsmaTK4OTkRMbevXtXo1FbOAHA/fv3NVpeXh4ZK9seqiXc4jJKwsZllISNyygJG5dREjYuoyQvVK+CbAK1NUdAffPNNxqNmjAOAB4eHhpNNoRKPY/KxgHg2rVrGo3aewygJ6jLztylJq7LyltZWanRZPuBXbp0SaP9/fffZKylcIvLKAkbl1ESNi6jJGxcRkleqC2YZAmBLFmhOHPmjEabMGECGUvNsaUSQYBOzmRn41Iret3c3MhY6p1lq4epZFC2gtnScgHAsmXLNNrs2bMtvi8Ft7iMkrBxGSVh4zJKwsZllISNyyjJUxnypToqZNk0NYlb1tHR1iOVrOk9kBETE6PRZJOtqV4FarK2DGrIGKB7ChoaGshYa84epupHVr/Uz/Py5ctkrKXn81oDt7iMkrBxGSVh4zJKwsZllKRN2Yo1w5ftkRi1lRMnTpB6VlaWRqM2OQYABwcHjUatpAXoubCyOcFU/VDPAuh6p54F0EmbrAzU6mEZVJIpu/7bb7/VaJMmTbL4WRTc4jJKwsZllISNyygJG5dREjYuoyTPfCJ5dXW1RpPtxVVQUGBxLJXJUtcD9B5dsg2YqSFU2UrYbt26aTTZkC81uZtaSQvQ5ZXtBxYaGqrR7t27R8bm5uZqNNmQLzWMK3s3aoPq33//nYy1FG5xGSVh4zJKwsZllISNyyhJm5Kz06dPk/pHH32k0W7fvk3G1tTUaDRZQkAlTC4uLmQsNewsS2CoZEdWLdQc28DAQDKW2q5pxIgRZCy1UTJVN4D8hB2K3r17azTZ6mFqXrFsGJeqS9mWUdSm07KE1lK4xWWUhI3LKAkbl1ESNi6jJGxcRkks7lWgJi+HhISQsdQwrGwiOdWDYM2EZtl+YFT2bw2y45eqqqo0Wnp6Ohl78OBBjbZt2zYy1svLS6PJ9uKiegp8fX3J2MLCQo1GvQNAr6KW1S/VCyLbk4zq4SktLSVjLYVbXEZJ2LiMkrBxGSVh4zJKYnFytnPnTo22YsUKMrZPnz4aTTYcSM0Nla1YpZAlD1Ry1b17dzLW29tbo8mGqKlh51u3bpGx2dnZGk22VVJxcbFGk9XZ+fPnLdIAOqmWnRFMvZs1W0bJrETdg9ogGwB69Ohh0bO4xWWUhI3LKAkbl1ESNi6jJGxcRkks3tCrS5cuGk2WpVM9BbJMtmfPnhZdD9BDitTQI0Afn+Tj40PGUs+TDbdSuuzc3ylTpmi0gQMHkrHU5HDZ0CxVl7IJ9dQwrqy81ApmWa8CNVQv61WgdNmKa+5VYP7TsHEZJWHjMkrCxmWUxOLkjErEZKtxqQ/YsuFLamhVlmhQp9BYczKNbCjZmlNsqBWysg2uqQ2fr169SsZSK2ypxBWgz9eVlZeqH9ncaCqRk8VSq3RlQ9/Udk2XLl0iYyMjI0m9JdziMkrCxmWUhI3LKAkbl1ESNi6jJBb3KgwZMkSjUUOaAJCWlqbRqE2OAXp1qmy4lcroZUOSVNYrW4VK9SrIykDFyo5foo57olbzAnQPjWxoliqDrCfGmuF36h6y+1LDw1SvBEBPkvf09CRjLYVbXEZJ2LiMkrBxGSVh4zJK8lRO3fnpp580WnJyMhlbUVGh0WTDuFSiYM0m0LIhX2rIVrZ6mKouWXJG3UOWTFK6NWWQQcVSc6tlyBJaqt5lQ76DBg3SaLt377a4DOTz23Q1wzwj2LiMkrBxGSVh4zJKwsZllMTiXgUqS5dl9NZw9OhRjfbBBx+QseXl5RpNtgEz9VqyCd9U5iybQG1Nlk71NshWRlN1SU0uB+TvYSnUcC1AD1HLzjQeN26cRpMdm0WdJ9xWuMVllISNyygJG5dREjYuoyRPZcj33yQ/P5/UqdXD1OpYALh+/bpGk23XRCU2shNvmKcHt7iMkrBxGSVh4zJKwsZllISNyyiJ8r0KzIsJt7iMkrBxGSVh4zJKwsZllISNyygJG5dREjYuoyRsXEZJ2LiMkrBxGSVh4zJKwsZllISNyygJG5dREjYuoyRsXEZJ2LiMkrBxGSVh4zJKwsZllOT/AIAoY3F4R0eMAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "images = X.copy()\n",
    "\n",
    "# Plotting each of the first few images separately\n",
    "for i in range(3):\n",
    "    print_image_by_id(id=i, images=images, labels=y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Partition the data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_rest1, y_train, y_rest1 = train_test_split(X, y, train_size=0.8, random_state=42, stratify=y)\n",
    "X_test, X_rest2, y_test, y_rest2 = train_test_split(X_rest1, y_rest1, train_size=0.8, random_state=42, stratify=y_rest1)\n",
    "X_calib, X_new, y_calib, y_new = train_test_split(X_rest2, y_rest2, train_size=0.8, random_state=42, stratify=y_rest2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(56000, 784)\n",
      "(11200, 784)\n",
      "(2240, 784)\n",
      "(560, 784)\n"
     ]
    }
   ],
   "source": [
    "scaler = StandardScaler()\n",
    "\n",
    "X_train = scaler.fit_transform(X_train)\n",
    "X_test = scaler.transform(X_test)\n",
    "X_calib = scaler.transform(X_calib)\n",
    "X_new = scaler.transform(X_new)\n",
    "\n",
    "print(X_train.shape)\n",
    "print(X_test.shape)\n",
    "print(X_calib.shape)\n",
    "print(X_new.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train some base model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-4 {color: black;}#sk-container-id-4 pre{padding: 0;}#sk-container-id-4 div.sk-toggleable {background-color: white;}#sk-container-id-4 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-4 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-4 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-4 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-4 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-4 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-4 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-4 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-4 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-4 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-4 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-4 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-4 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-4 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-4 div.sk-item {position: relative;z-index: 1;}#sk-container-id-4 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-4 div.sk-item::before, #sk-container-id-4 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-4 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-4 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-4 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-4 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-4 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-4 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-4 div.sk-label-container {text-align: center;}#sk-container-id-4 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-4 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-4\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(random_state=42)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" checked><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(random_state=42)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(random_state=42)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = RandomForestClassifier(random_state=42)\n",
    "#model = GaussianNB()\n",
    "#model = SVC(gamma='auto', probability=True)\n",
    "\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Naive approach"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2240, 10)\n",
      "Accuracy: 88.3036%\n",
      "(2240,)\n",
      "Coverage: 0.9987\n"
     ]
    }
   ],
   "source": [
    "y_probs = model.predict_proba(X_calib)\n",
    "y_pred = model.predict(X_calib)\n",
    "\n",
    "print(y_probs.shape)\n",
    "\n",
    "alpha = 0.05\n",
    "print(f\"Accuracy: {(y_pred == y_calib).mean() * 100:.4f}%\")\n",
    "\n",
    "high_prob = np.amax(y_probs, axis=1)\n",
    "print(high_prob.shape)\n",
    "\n",
    "high_prob_indices = np.where(high_prob >= (1 - alpha))\n",
    "\n",
    "# Let's count how often we are correct\n",
    "matches = (y_pred == y_calib)\n",
    "\n",
    "# Calculate marginal/average coverage\n",
    "coverage = np.mean(matches[high_prob_indices])\n",
    "print(f\"Coverage: {coverage:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the distribution of the probabilities\n",
    "plt.hist(high_prob, bins=30);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.84      0.90      0.87       224\n",
      "           1       0.99      0.96      0.98       224\n",
      "           2       0.76      0.78      0.77       224\n",
      "           3       0.87      0.92      0.89       224\n",
      "           4       0.75      0.84      0.79       224\n",
      "           5       0.97      0.97      0.97       224\n",
      "           6       0.77      0.57      0.65       224\n",
      "           7       0.96      0.95      0.96       224\n",
      "           8       0.95      0.97      0.96       224\n",
      "           9       0.97      0.98      0.97       224\n",
      "\n",
      "    accuracy                           0.88      2240\n",
      "   macro avg       0.88      0.88      0.88      2240\n",
      "weighted avg       0.88      0.88      0.88      2240\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#print(confusion_matrix(y_test, y_pred))\n",
    "print(classification_report(y_calib, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conformal approach"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "n = len(y_calib)\n",
    "\n",
    "y_score = model.predict_proba(X_calib)\n",
    "\n",
    "# Select the probabilities for the true class\n",
    "prob_true_class = y_score[np.arange(n), y_calib]\n",
    "\n",
    "# Turn into conformal uncertainty scores (larger means more uncertain)\n",
    "uncertainty_scores = 1 - prob_true_class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Quantile: 0.78\n"
     ]
    }
   ],
   "source": [
    "# Set the confidence level\n",
    "alpha = 0.05\n",
    "q_level = np.ceil((n+1) * (1-alpha))/n    # apply finite sample correction for small samples\n",
    "\n",
    "qhat = np.quantile(uncertainty_scores, q_level, method='higher') \n",
    "print('Quantile:', qhat)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualize the uncertainty scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAc8AAAF1CAYAAACOKWE4AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAxCUlEQVR4nO3deXQUZd728StbhwTojkGSkDGsghDBBVBoEUclQ4QMouCCIgRkxMHACFFURhAFHoKgiDosDw4m+CgvioqjrCIKKgRRBEVAlEUDQzqgSAKB7PX+kUkzTYhQnaWL5Ps5p86prrq761cVyJW7lrv9DMMwBAAAzpu/rwsAAOBCQ3gCAGAS4QkAgEmEJwAAJhGeAACYRHgCAGAS4QkAgEmEJwAAJhGeAACYRHgCAGBSoK8L+Pe//63HH39cK1eu1MmTJ3XppZcqNTVVnTt3liQZhqGJEyfqlVde0bFjx9StWzfNnTtXrVu3dn/G0aNHNWrUKH3wwQfy9/dX//799eKLL6pBgwbnVUNJSYkOHTqkhg0bys/Pr1r2EwBgbYZh6Pjx44qOjpa//zn6loYPHT161GjWrJkxZMgQ44svvjD27dtnrF692tizZ4+7zbRp0wyHw2G89957xjfffGPceuutRosWLYxTp06529xyyy3GlVdeaWzatMn47LPPjEsvvdS45557zruOAwcOGJKYmJiYmJiMAwcOnDM3/AzDdwPDP/HEE9qwYYM+++yzs643DEPR0dF65JFH9Oijj0qSsrOzFRkZqbS0NA0YMEC7du1SbGysvvzyS3dvddWqVerdu7cOHjyo6Ojoc9aRnZ2tsLAwHThwQHa7vep2EABwwcjJyVFMTIyOHTsmh8Pxu219etr2/fffV3x8vO68806tX79ef/jDH/TQQw/pgQcekCTt379fLpdLcXFx7vc4HA516dJF6enpGjBggNLT0xUWFuYOTkmKi4uTv7+/vvjiC91+++3ltpufn6/8/Hz36+PHj0uS7HY74QnAGk6elK65pnT+yy+l0FDf1lOHnM/lO5/eMLRv3z739cvVq1drxIgR+tvf/qaFCxdKklwulyQpMjLS432RkZHudS6XSxERER7rAwMDFR4e7m5zppSUFDkcDvcUExNT1bsGAJVjGNLOnaUT3xxpOT4Nz5KSEnXs2FFTp07V1VdfreHDh+uBBx7QvHnzqnW748aNU3Z2tns6cOBAtW4PAFC7+DQ8mzRpotjYWI9l7dq1U0ZGhiQpKipKkpSVleXRJisry70uKipKhw8f9lhfVFSko0ePutucKTg42H2KllO1AACzfBqe3bp10+7duz2W/fDDD2rWrJkkqUWLFoqKitLatWvd63NycvTFF1/I6XRKkpxOp44dO6YtW7a423z88ccqKSlRly5damAvAAB1jU9vGBozZoyuu+46TZ06VXfddZc2b96s+fPna/78+ZJKL9qOHj1aU6ZMUevWrdWiRQtNmDBB0dHRuu222ySV9lRvueUW9+newsJCjRw5UgMGDDivO20BADDNxGOZ1eKDDz4w2rdvbwQHBxtt27Y15s+f77G+pKTEmDBhghEZGWkEBwcbPXr0MHbv3u3R5tdffzXuueceo0GDBobdbjeGDh1qHD9+/LxryM7ONiQZ2dnZVbJPAFBpJ04YRumtQqXzqHZmssCnz3laRU5OjhwOh7Kzs7n+CcAaTp6Uyu4J2bmTR1VqgJks8PnwfACAswgNlX76yddVoAIMDA8AgEmEJwAAJhGeAGBFp06VDs93zTWl87AUrnlWoeZPLD9nm5+mJdRAJQAueCUl0ldfnZ6HpdDzBADAJMITAACTCE8AAEwiPAEAMInwBADAJO62BQCruvhiX1eAChCeAGBF9etLR474ugpUgNO2AACYRHgCAGAS4QkAVnTqlHTjjaUTw/NZDtc8AcCKSkqk9etPz8NS6HkCAGAS4QkAgEmEJwAAJhGeAACYRHgCAGASd9sCgFWFhvq6AlSA8AQAK6pfX8rN9XUVqACnbQEAMInwBADAJMITAKwoL09KSCid8vJ8XQ3OwDVPALCi4mJpxYrT87AUep4AAJhEeAIAYBLhCQCASYQnAAAmEZ4AAJhEeAIAYBKPqgCAFdWvLxmGr6tABeh5AgBgEuEJAIBJhCcAWFFennTnnaUTw/NZDuEJAFZUXCy9/XbpxPB8lkN4AgBgEuEJAIBJPg3Pp59+Wn5+fh5T27Zt3evz8vKUlJSkRo0aqUGDBurfv7+ysrI8PiMjI0MJCQkKDQ1VRESExo4dq6KiopreFQBAHeLz5zwvv/xyffTRR+7XgYGnSxozZoyWL1+uJUuWyOFwaOTIkerXr582bNggSSouLlZCQoKioqK0ceNGZWZmavDgwQoKCtLUqVNrfF8AAHWDz8MzMDBQUVFR5ZZnZ2drwYIFWrRokW6++WZJUmpqqtq1a6dNmzapa9eu+vDDD7Vz50599NFHioyM1FVXXaXJkyfr8ccf19NPPy2bzVbTuwMAqAN8fs3zxx9/VHR0tFq2bKmBAwcqIyNDkrRlyxYVFhYqLi7O3bZt27Zq2rSp0tPTJUnp6enq0KGDIiMj3W3i4+OVk5OjHTt2VLjN/Px85eTkeEwAAJwvn4Znly5dlJaWplWrVmnu3Lnav3+/unfvruPHj8vlcslmsyksLMzjPZGRkXK5XJIkl8vlEZxl68vWVSQlJUUOh8M9xcTEVO2OAUBlhYZKJ06UTqGhvq4GZ/DpadtevXq556+44gp16dJFzZo101tvvaWQkJBq2+64ceOUnJzsfp2Tk0OAArAWP7/S8W1hST4/bfvfwsLC1KZNG+3Zs0dRUVEqKCjQsWPHPNpkZWW5r5FGRUWVu/u27PXZrqOWCQ4Olt1u95gAADhflgrPEydOaO/evWrSpIk6deqkoKAgrV271r1+9+7dysjIkNPplCQ5nU5t375dhw8fdrdZs2aN7Ha7YmNja7x+AKgy+fnSkCGlU36+r6vBGXx62vbRRx9Vnz591KxZMx06dEgTJ05UQECA7rnnHjkcDg0bNkzJyckKDw+X3W7XqFGj5HQ61bVrV0lSz549FRsbq0GDBmn69OlyuVwaP368kpKSFBwc7MtdA4DKKSqSFi4snZ89W+J3mqX4NDwPHjyoe+65R7/++qsaN26s66+/Xps2bVLjxo0lSS+88IL8/f3Vv39/5efnKz4+XnPmzHG/PyAgQMuWLdOIESPkdDpVv359JSYmatKkSb7aJQBAHeBnGHzbak5OjhwOh7Kzsyt1/bP5E8vP2eanaQlefz6AOiQ3V2rQoHT+xAluHqoBZrLAUtc8AQC4EBCeAACYRHgCAGAS4QkAgEk+HxgeAHAWoaFS2TPsDM9nOYQnAFiRn5/0n8f2YD2ctgUAwCTCEwCsKD9fSkoqnRiez3IITwCwoqIiac6c0qmoyNfV4AyEJwAAJhGeAACYRHgCAGAS4QkAgEmEJwAAJhGeAACYxAhDAGBFISHS/v2n52EphCcAWJG/v9S8ua+rQAU4bQsAgEmEJwBYUUGBNHZs6VRQ4OtqcAbCEwCsqLBQeu650qmw0NfV4AyEJwAAJhGeAACYRHgCAGAS4QkAgEmEJwAAJhGeAACYxAhDAGBFISHSd9+dnoelEJ4AYEX+/tLll/u6ClSA07YAAJhEzxMArKigQJo6tXT+73+XbDbf1gMPhCcAWFFhofTMM6XzY8cSnhbDaVsAAEwiPAEAMInwBADAJMITAACTCE8AAEwiPAEAMIlHVQDAiurVkzZvPj0PSyE8AcCKAgKka67xdRWoAKdtAQAwyTLhOW3aNPn5+Wn06NHuZXl5eUpKSlKjRo3UoEED9e/fX1lZWR7vy8jIUEJCgkJDQxUREaGxY8eqqKiohqsHgCpWUCDNmFE6FRT4uhqcwRLh+eWXX+p///d/dcUVV3gsHzNmjD744AMtWbJE69ev16FDh9SvXz/3+uLiYiUkJKigoEAbN27UwoULlZaWpqeeeqqmdwEAqlZhofTYY6VTYaGvq8EZfB6eJ06c0MCBA/XKK6/ooosuci/Pzs7WggULNHPmTN18883q1KmTUlNTtXHjRm3atEmS9OGHH2rnzp16/fXXddVVV6lXr16aPHmyZs+erYLf+UstPz9fOTk5HhMAAOfL5+GZlJSkhIQExcXFeSzfsmWLCgsLPZa3bdtWTZs2VXp6uiQpPT1dHTp0UGRkpLtNfHy8cnJytGPHjgq3mZKSIofD4Z5iYmKqeK8AALWZT8Nz8eLF+vrrr5WSklJuncvlks1mU1hYmMfyyMhIuVwud5v/Ds6y9WXrKjJu3DhlZ2e7pwMHDlRyTwAAdYnPHlU5cOCAHn74Ya1Zs0b1avgZpuDgYAUHB9foNgEAtYfPep5btmzR4cOH1bFjRwUGBiowMFDr16/XSy+9pMDAQEVGRqqgoEDHjh3zeF9WVpaioqIkSVFRUeXuvi17XdYGAICq5rPw7NGjh7Zv365t27a5p86dO2vgwIHu+aCgIK1du9b9nt27dysjI0NOp1OS5HQ6tX37dh0+fNjdZs2aNbLb7YqNja3xfQIA1A0+O23bsGFDtW/f3mNZ/fr11ahRI/fyYcOGKTk5WeHh4bLb7Ro1apScTqe6du0qSerZs6diY2M1aNAgTZ8+XS6XS+PHj1dSUhKnZQFc2OrVkz755PQ8LMXSw/O98MIL8vf3V//+/ZWfn6/4+HjNmTPHvT4gIEDLli3TiBEj5HQ6Vb9+fSUmJmrSpEk+rBoAqkBAgHTjjb6uAhXwMwzD8HURvpaTkyOHw6Hs7GzZ7XavP6f5E8vP2eanaQlefz4AoPqYyQJL9zwBoM4qLJTmzy+dHz5cCgrybT3wQHgCgBUVFEgjR5bODxlCeFqMz0cYAgDgQkN4AgBgEuEJAIBJhCcAACYRngAAmER4AgBgEo+qAIAVBQdLy5adnoeleBWe+/btU8uWLau6FgBAmcBAKYERyazKq9O2l156qW666Sa9/vrrysvLq+qaAACwNK/C8+uvv9YVV1yh5ORkRUVF6cEHH9TmzZurujYAqLsKC6W0tNKpsNDX1eAMXoXnVVddpRdffFGHDh3Sq6++qszMTF1//fVq3769Zs6cqSNHjlR1nQBQtxQUSEOHlk4FBb6uBmeo1N22gYGB6tevn5YsWaJnn31We/bs0aOPPqqYmBgNHjxYmZmZVVUnAACWUanw/Oqrr/TQQw+pSZMmmjlzph599FHt3btXa9as0aFDh9S3b9+qqhMAAMvw6m7bmTNnKjU1Vbt371bv3r312muvqXfv3vL3L83iFi1aKC0tTc2bN6/KWgEAsASvwnPu3Lm6//77NWTIEDVp0uSsbSIiIrRgwYJKFQcAgBV5FZ4//vjjOdvYbDYlJiZ68/EAAFiaV9c8U1NTtWTJknLLlyxZooULF1a6KAAArMyr8ExJSdHFF19cbnlERISmTp1a6aIAoM4LDpbeeqt0Yng+y/HqtG1GRoZatGhRbnmzZs2UkZFR6aIAoM4LDJTuvNPXVaACXvU8IyIi9O2335Zb/s0336hRo0aVLgoAACvzqud5zz336G9/+5saNmyoG264QZK0fv16PfzwwxowYECVFggAdVJRkbR0aen87beX9kRhGV79NCZPnqyffvpJPXr0UOB/fqAlJSUaPHgw1zwBoCrk50t33VU6f+IE4WkxXv00bDab3nzzTU2ePFnffPONQkJC1KFDBzVr1qyq6wMAwHIq9adMmzZt1KZNm6qqBQCAC4JX4VlcXKy0tDStXbtWhw8fVklJicf6jz/+uEqKAwDAirwKz4cfflhpaWlKSEhQ+/bt5efnV9V1AQBgWV6F5+LFi/XWW2+pd+/eVV0PAACW59VznjabTZdeemlV1wIAwAXBq/B85JFH9OKLL8owjKquBwAgSTablJpaOtlsvq4GZ/DqtO3nn3+uTz75RCtXrtTll1+uoKAgj/XvvvtulRQHAHVWUJA0ZIivq0AFvArPsLAw3X777VVdCwAAFwSvwjM1NbWq6wAA/LeiImn16tL5+HhGGLIYr38aRUVFWrdunfbu3at7771XDRs21KFDh2S329WgQYOqrBEA6p78fOnPfy6dZ3g+y/Hqp/Hzzz/rlltuUUZGhvLz8/WnP/1JDRs21LPPPqv8/HzNmzevqusEAMAyvLrb9uGHH1bnzp3122+/KSQkxL389ttv19q1a6usOAAArMirnudnn32mjRs3ynbG7dPNmzfXv//97yopDAAAq/Kq51lSUqLi4uJyyw8ePKiGDRtWuigAAKzMq/Ds2bOnZs2a5X7t5+enEydOaOLEiQzZBwCo9bwKz+eff14bNmxQbGys8vLydO+997pP2T777LPn/Tlz587VFVdcIbvdLrvdLqfTqZUrV7rX5+XlKSkpSY0aNVKDBg3Uv39/ZWVleXxGRkaGEhISFBoaqoiICI0dO1ZFRUXe7BYAAOfFq2uel1xyib755hstXrxY3377rU6cOKFhw4Zp4MCBHjcQnc/nTJs2Ta1bt5ZhGFq4cKH69u2rrVu36vLLL9eYMWO0fPlyLVmyRA6HQyNHjlS/fv20YcMGSaVfjZaQkKCoqCht3LhRmZmZGjx4sIKCgjR16lRvdg0ArMFmk/7xj9PzsBQ/w2ID1IaHh2vGjBm644471LhxYy1atEh33HGHJOn7779Xu3btlJ6erq5du2rlypX685//rEOHDikyMlKSNG/ePD3++OM6cuRIuRuayuTn5ys/P9/9OicnRzExMcrOzpbdbve69uZPLD9nm5+mJXj9+QCA6pOTkyOHw3FeWeBVz/O111773fWDBw82/ZnFxcVasmSJcnNz5XQ6tWXLFhUWFiouLs7dpm3btmratKk7PNPT09WhQwd3cEpSfHy8RowYoR07dujqq68+67ZSUlL0zDPPmK4RAACpEl+G/d8KCwt18uRJ2Ww2hYaGmgrP7du3y+l0Ki8vTw0aNNDSpUsVGxurbdu2yWazKSwszKN9ZGSkXC6XJMnlcnkEZ9n6snUVGTdunJKTk92vy3qeAGAZxcXSZ5+VznfvLgUE+LYeePAqPH/77bdyy3788UeNGDFCY8eONfVZl112mbZt26bs7Gy9/fbbSkxM1Pr1670p67wFBwcrODi4WrcBAJWSlyfddFPp/IkTUv36vq0HHry62/ZsWrdurWnTppXrlZ5L2Rdrd+rUSSkpKbryyiv14osvKioqSgUFBTp27JhH+6ysLEVFRUmSoqKiyt19W/a6rA0AAFWtysJTkgIDA3Xo0KFKfUZJSYny8/PVqVMnBQUFeQz3t3v3bmVkZMjpdEqSnE6ntm/frsOHD7vbrFmzRna7XbGxsZWqAwCAinh12vb999/3eG0YhjIzM/WPf/xD3bp1O+/PGTdunHr16qWmTZvq+PHjWrRokdatW6fVq1fL4XBo2LBhSk5OVnh4uOx2u0aNGiWn06muXbtKKh2sITY2VoMGDdL06dPlcrk0fvx4JSUlcVoWAFBtvArP2267zeO1n5+fGjdurJtvvlnPP//8eX/O4cOHNXjwYGVmZsrhcOiKK67Q6tWr9ac//UmS9MILL8jf31/9+/dXfn6+4uPjNWfOHPf7AwICtGzZMo0YMUJOp1P169dXYmKiJk2a5M1uAQBwXiz3nKcvmHm25/fwnCeAKpObK5V9NzI3DNUIM1lQpdc8AQCoC7w6bfvfz0iey8yZM73ZBADUbUFB0vTpp+dhKV6F59atW7V161YVFhbqsssukyT98MMPCggIUMeOHd3t/Pz8qqZKAKhrbDbJ5HPzqDlehWefPn3UsGFDLVy4UBdddJGk0oEThg4dqu7du+uRRx6p0iIBALASr7+SLCUlxR2cknTRRRdpypQppu62BQBUoLhY+vLL0qm42NfV4Axe9TxzcnJ05MiRcsuPHDmi48ePV7ooAKjz8vKka68tneduW8vxqud5++23a+jQoXr33Xd18OBBHTx4UO+8846GDRumfv36VXWNAABYilc9z3nz5unRRx/Vvffeq8LCwtIPCgzUsGHDNGPGjCotEAAAq/EqPENDQzVnzhzNmDFDe/fulSS1atVK9TmtAACoAyo1SEJmZqYyMzPVunVr1a9fXwxWBACoC7wKz19//VU9evRQmzZt1Lt3b2VmZkqShg0bxmMqAIBaz6vwHDNmjIKCgpSRkaHQ0FD38rvvvlurVq2qsuIAALAir655fvjhh1q9erUuueQSj+WtW7fWzz//XCWFAUCdFhQkTZx4eh6W4lV45ubmevQ4yxw9epTv0QSAqmCzSU8/7esqUAGvTtt2795dr732mvu1n5+fSkpKNH36dN10001VVhwAAFbkVc9z+vTp6tGjh7766isVFBToscce044dO3T06FFt2LChqmsEgLqnpETatat0vl07yZ9vkLQSr34a7du31w8//KDrr79effv2VW5urvr166etW7eqVatWVV0jANQ9p05J7duXTqdO+boanMF0z7OwsFC33HKL5s2bpyeffLI6agIAwNJM9zyDgoL07bffVkctAABcELw6bXvfffdpwYIFVV0LAAAXBK9uGCoqKtKrr76qjz76SJ06dSo3pu3MmTOrpDgAAKzIVHju27dPzZs313fffaeOHTtKkn744QePNn5+flVXHQAAFmQqPFu3bq3MzEx98sknkkqH43vppZcUGRlZLcUBAGBFpsLzzG9NWblypXJzc6u0IACASofke/TR0/OwFK+ueZbhK8gAoJrYbNKMGb6uAhUwdbetn59fuWuaXOMEANQ1pk/bDhkyxD34e15env7617+Wu9v23XffrboKAaAuKimRMjJK55s2ZXg+izEVnomJiR6v77vvviotBgDwH6dOSS1alM6fOCGd0UmBb5kKz9TU1OqqAwCACwbnAQAAMInwBADAJMITAACTCE8AAEwiPAEAMKlSIwwBAKpJYKD00EOn52Ep/EQAwIqCg6XZs31dBSrAaVsAAEyi5wkAVmQY0i+/lM5ffLHEOOKWQngCgBWdPClFRJTOMzyf5fj0tG1KSoquueYaNWzYUBEREbrtttu0e/dujzZ5eXlKSkpSo0aN1KBBA/Xv319ZWVkebTIyMpSQkKDQ0FBFRERo7NixKioqqsldAQDUIT4Nz/Xr1yspKUmbNm3SmjVrVFhYqJ49e3p8wfaYMWP0wQcfaMmSJVq/fr0OHTqkfv36udcXFxcrISFBBQUF2rhxoxYuXKi0tDQ99dRTvtglAEAd4GdY6Butjxw5ooiICK1fv1433HCDsrOz1bhxYy1atEh33HGHJOn7779Xu3btlJ6erq5du2rlypX685//rEOHDikyMlKSNG/ePD3++OM6cuSIbDbbObebk5Mjh8Oh7Oxs2e12r+tv/sTyc7b5aVqC158PoA7JzZUaNCid57RtjTCTBZa62zY7O1uSFB4eLknasmWLCgsLFRcX527Ttm1bNW3aVOnp6ZKk9PR0dejQwR2ckhQfH6+cnBzt2LHjrNvJz89XTk6OxwQAwPmyTHiWlJRo9OjR6tatm9q3by9JcrlcstlsCgsL82gbGRkpl8vlbvPfwVm2vmzd2aSkpMjhcLinmJiYKt4bAEBtZpnwTEpK0nfffafFixdX+7bGjRun7Oxs93TgwIFq3yYAoPawxKMqI0eO1LJly/Tpp5/qkksucS+PiopSQUGBjh075tH7zMrKUlRUlLvN5s2bPT6v7G7csjZnCg4OVnBwcBXvBQBUocBAKTHx9Dwsxac9T8MwNHLkSC1dulQff/yxWrRo4bG+U6dOCgoK0tq1a93Ldu/erYyMDDmdTkmS0+nU9u3bdfjwYXebNWvWyG63KzY2tmZ2BACqWnCwlJZWOvHHvuX49M+ZpKQkLVq0SP/617/UsGFD9zVKh8OhkJAQORwODRs2TMnJyQoPD5fdbteoUaPkdDrVtWtXSVLPnj0VGxurQYMGafr06XK5XBo/frySkpLoXQIAqoVPw3Pu3LmSpBtvvNFjeWpqqoYMGSJJeuGFF+Tv76/+/fsrPz9f8fHxmjNnjrttQECAli1bphEjRsjpdKp+/fpKTEzUpEmTamo3AKDqGUbpKEOSFBrK8HwWY6nnPH2F5zwBWA7Peda4C/Y5TwAALgSEJwAAJhGeAACYRHgCAGAS4QkAgEmEJwAAJjHmEwBYUUCA9J+vYlRAgG9rQTmEJwBYUb160pIlvq4CFeC0LQAAJtHztCBGKgIAa6PnCQBWlJtbOp6tn1/pPCyF8AQAwCTCEwAAkwhPAABMIjwBADCJ8AQAwCTCEwAAk3jOs4adzzOcAKCAAKl379PzsBTCEwCsqF49aTl/bFsVp20BADCJ8AQAwCTCEwCsKDdXql+/dGJ4PsvhmicAWNXJk76uABWg5wkAgEmEJwAAJhGeAACYRHgCAGAS4QkAgEncbQsAVuTvL/3xj6fnYSmEJwBYUUiItG6dr6tABfhzBgAAkwhPAABMIjwBwIpyc6XGjUsnhuezHK55AoBV/fKLrytABeh5AgBgEuEJAIBJhCcAACYRngAAmER4AgBgEnfbAoAV+ftLnTufnoel+PQn8umnn6pPnz6Kjo6Wn5+f3nvvPY/1hmHoqaeeUpMmTRQSEqK4uDj9+OOPHm2OHj2qgQMHym63KywsTMOGDdOJEydqcC8AoBqEhEhfflk6hYT4uhqcwac9z9zcXF155ZW6//771a9fv3Lrp0+frpdeekkLFy5UixYtNGHCBMXHx2vnzp2qV6+eJGngwIHKzMzUmjVrVFhYqKFDh2r48OFatGhRTe+OpTR/Yvk52/w0LaEGKgGA2sen4dmrVy/16tXrrOsMw9CsWbM0fvx49e3bV5L02muvKTIyUu+9954GDBigXbt2adWqVfryyy/V+T+nN15++WX17t1bzz33nKKjo2tsXwAAdYdlT6Tv379fLpdLcXFx7mUOh0NdunRRenq6JCk9PV1hYWHu4JSkuLg4+fv764svvqjws/Pz85WTk+MxAYClnDwpNW9eOp086etqcAbLhqfL5ZIkRUZGeiyPjIx0r3O5XIqIiPBYHxgYqPDwcHebs0lJSZHD4XBPMTExVVw9AFSSYUg//1w6GYavq8EZLBue1WncuHHKzs52TwcOHPB1SQCAC4hlwzMqKkqSlJWV5bE8KyvLvS4qKkqHDx/2WF9UVKSjR4+625xNcHCw7Ha7xwQAwPmybHi2aNFCUVFRWrt2rXtZTk6OvvjiCzmdTkmS0+nUsWPHtGXLFnebjz/+WCUlJerSpUuN1wwAqBt8erftiRMntGfPHvfr/fv3a9u2bQoPD1fTpk01evRoTZkyRa1bt3Y/qhIdHa3bbrtNktSuXTvdcssteuCBBzRv3jwVFhZq5MiRGjBgAHfaAgCqjU/D86uvvtJNN93kfp2cnCxJSkxMVFpamh577DHl5uZq+PDhOnbsmK6//nqtWrXK/YynJL3xxhsaOXKkevToIX9/f/Xv318vvfRSje9LbcXzogBQnk/D88Ybb5TxO3eR+fn5adKkSZo0aVKFbcLDw+v8gAgAaiE/Pyk29vQ8LIWxbQHAikJDpR07fF0FKmDZG4YAALAqwhMAAJM4bXuBOp8beQBcwE6elK65pnT+yy9LT+PCMghPALAiw5B27jw9D0vhtC0AACYRngAAmMRpW9QIBlsAUJsQnnUYNx0BgHc4bQsAgEn0PAHAivz8pGbNTs/DUghPALCi0FDpp598XQUqwGlbAABMIjwBADCJ07aAl851tzKP3qBSTp2SbrihdP7TT6WQEN/WAw+EJwBYUUmJ9NVXp+dhKZy2BQDAJMITAACTOG2LCwZD/PkePwOgFOGJSrvQhvkjAABUFuEJ1AL8QQDULMITAKzq4ot9XQEqQHgCPkSPERWqX186csTXVaAC3G0LAIBJ9DwBi7vQbsiqKfTa4UuEJ4AaR/Cdh1OnpF69SudXrmR4PoshPFGrVFUvjd4efK6kRFq//vQ8LIVrngAAmETPE5ZR23p7VtsfvgXG+v77ZxRSkKdd/5lvN2GVTtnqSeLnZBX0PAEAMImeJ4AqZbUeN1AdCE8AllRTIVwb7/ytjftkNYQnAFSB6rimfDIo2NtyUM0ITwCSON1a3cwe31O2eopNfqeaqkFlEZ4AgLPiDu2KEZ4Aai1606guhCcAnIMvQji4qEBzl06VJI24/e/KD7TVeA2oGOEJABbkX1Kim/d95Z6HtRCeAHABYfxmayA8AQDVprY+c1prwnP27NmaMWOGXC6XrrzySr388su69tprfV0WANRadbn3WivC880331RycrLmzZunLl26aNasWYqPj9fu3bsVERHh6/IAAL+jqkK4JnuwtWJg+JkzZ+qBBx7Q0KFDFRsbq3nz5ik0NFSvvvqqr0sDANRCF3zPs6CgQFu2bNG4cePcy/z9/RUXF6f09PSzvic/P1/5+fnu19nZ2ZKknJycStVSkn+yUu8HgDLFBXkq+41UnH9SJQZ33J5LZX+Hl73fMIxztr3gw/OXX35RcXGxIiMjPZZHRkbq+++/P+t7UlJS9Mwzz5RbHhMTUy01AoA3HGUzcwb7sowLhmNW1XzO8ePH5XA4frfNBR+e3hg3bpySk5Pdr0tKSnT06FE1atRIfn5+Xn1mTk6OYmJidODAAdnt9qoqtVbg2Pw+jk/FODYV49hUzNtjYxiGjh8/rujo6HO2veDD8+KLL1ZAQICysrI8lmdlZSkqKuqs7wkODlZwsOe3FYSFhVVJPXa7nX/IFeDY/D6OT8U4NhXj2FTMm2Nzrh5nmQv+hiGbzaZOnTpp7dq17mUlJSVau3atnE6nDysDANRWF3zPU5KSk5OVmJiozp0769prr9WsWbOUm5uroUOH+ro0AEAtVCvC8+6779aRI0f01FNPyeVy6aqrrtKqVavK3URUnYKDgzVx4sRyp4PBsTkXjk/FODYV49hUrCaOjZ9xPvfkAgAAtwv+micAADWN8AQAwCTCEwAAkwhPAABMIjxNmD17tpo3b6569eqpS5cu2rx58++2X7Jkidq2bat69eqpQ4cOWrFiRQ1VWvPMHJtXXnlF3bt310UXXaSLLrpIcXFx5zyWFzqz/3bKLF68WH5+frrtttuqt0AfMntsjh07pqSkJDVp0kTBwcFq06ZNrf2/ZfbYzJo1S5dddplCQkIUExOjMWPGKC8vr4aqrTmffvqp+vTpo+joaPn5+em9994753vWrVunjh07Kjg4WJdeeqnS0tIqV4SB87J48WLDZrMZr776qrFjxw7jgQceMMLCwoysrKyztt+wYYMREBBgTJ8+3di5c6cxfvx4IygoyNi+fXsNV179zB6be++915g9e7axdetWY9euXcaQIUMMh8NhHDx4sIYrrxlmj0+Z/fv3G3/4wx+M7t27G3379q2ZYmuY2WOTn59vdO7c2ejdu7fx+eefG/v37zfWrVtnbNu2rYYrr35mj80bb7xhBAcHG2+88Yaxf/9+Y/Xq1UaTJk2MMWPG1HDl1W/FihXGk08+abz77ruGJGPp0qW/237fvn1GaGiokZycbOzcudN4+eWXjYCAAGPVqlVe10B4nqdrr73WSEpKcr8uLi42oqOjjZSUlLO2v+uuu4yEhASPZV26dDEefPDBaq3TF8wemzMVFRUZDRs2NBYuXFhdJfqUN8enqKjIuO6664x//vOfRmJiYq0NT7PHZu7cuUbLli2NgoKCmirRZ8wem6SkJOPmm2/2WJacnGx069atWuv0tfMJz8cee8y4/PLLPZbdfffdRnx8vNfb5bTteSj72rO4uDj3snN97Vl6erpHe0mKj4+vsP2Fyptjc6aTJ0+qsLBQ4eHh1VWmz3h7fCZNmqSIiAgNGzasJsr0CW+Ozfvvvy+n06mkpCRFRkaqffv2mjp1qoqLi2uq7BrhzbG57rrrtGXLFvep3X379mnFihXq3bt3jdRsZdXx+7hWjDBU3bz52jOXy3XW9i6Xq9rq9AVvjs2ZHn/8cUVHR5f7x10beHN8Pv/8cy1YsEDbtm2rgQp9x5tjs2/fPn388ccaOHCgVqxYoT179uihhx5SYWGhJk6cWBNl1whvjs29996rX375Rddff70Mw1BRUZH++te/6u9//3tNlGxpFf0+zsnJ0alTpxQSEmL6M+l5wqemTZumxYsXa+nSpapXr56vy/G548ePa9CgQXrllVd08cUX+7ocyykpKVFERITmz5+vTp066e6779aTTz6pefPm+bo0n1u3bp2mTp2qOXPm6Ouvv9a7776r5cuXa/Lkyb4urVai53kevPnas6ioKFPtL1TeHJsyzz33nKZNm6aPPvpIV1xxRXWW6TNmj8/evXv1008/qU+fPu5lJSUlkqTAwEDt3r1brVq1qt6ia4g3/3aaNGmioKAgBQQEuJe1a9dOLpdLBQUFstls1VpzTfHm2EyYMEGDBg3SX/7yF0lShw4dlJubq+HDh+vJJ5+Uv3/d7StV9PvYbrd71euU6HmeF2++9szpdHq0l6Q1a9bUuq9J8/Yr4aZPn67Jkydr1apV6ty5c02U6hNmj0/btm21fft2bdu2zT3deuutuummm7Rt2zbFxMTUZPnVypt/O926ddOePXvcf1BI0g8//KAmTZrUmuCUvDs2J0+eLBeQZX9kGHV8CPNq+X3s9a1GdczixYuN4OBgIy0tzdi5c6cxfPhwIywszHC5XIZhGMagQYOMJ554wt1+w4YNRmBgoPHcc88Zu3btMiZOnFirH1Uxc2ymTZtm2Gw24+233zYyMzPd0/Hjx321C9XK7PE5U22+29bsscnIyDAaNmxojBw50ti9e7exbNkyIyIiwpgyZYqvdqHamD02EydONBo2bGj8v//3/4x9+/YZH374odGqVSvjrrvu8tUuVJvjx48bW7duNbZu3WpIMmbOnGls3brV+Pnnnw3DMIwnnnjCGDRokLt92aMqY8eONXbt2mXMnj2bR1Vq0ssvv2w0bdrUsNlsxrXXXmts2rTJve6Pf/yjkZiY6NH+rbfeMtq0aWPYbDbj8ssvN5YvX17DFdccM8emWbNmhqRy08SJE2u+8Bpi9t/Of6vN4WkY5o/Nxo0bjS5duhjBwcFGy5Ytjf/5n/8xioqKarjqmmHm2BQWFhpPP/200apVK6NevXpGTEyM8dBDDxm//fZbzRdezT755JOz/g4pOx6JiYnGH//4x3LvueqqqwybzWa0bNnSSE1NrVQNfCUZAAAmcc0TAACTCE8AAEwiPAEAMInwBADAJMITAACTCE8AAEwiPAEAMInwBADAJMITgIcFCxaoZ8+e1bqNAQMG6Pnnn6/WbQDViRGGAAv69NNPNWPGDG3ZskWZmZlaunSpbrvttmrfbl5enlq2bKklS5aoW7du1bad7777TjfccIP2798vh8NRbdsBqgs9T8CCcnNzdeWVV2r27Nk1ut23335bdru9WoNTktq3b69WrVrp9ddfr9btANWF8AQsqFevXpoyZYpuv/32Kv3c3377TQMHDlTjxo0VEhKi1q1bKzU11b1+8eLFHt8l+umnnyooKEgul8vjc0aPHq3u3bufdRvr1q2TzWbTZ5995l42ffp0RUREeHynYp8+fbR48eKq2jWgRhGeQB0yYcIE7dy5UytXrtSuXbs0d+5cXXzxxe71n3/+ucf3q95www1q2bKl/u///s+9rLCwUG+88Ybuv//+s27jxhtv1OjRozVo0CBlZ2dr69atmjBhgv75z38qMjLS3e7aa6/V5s2blZ+fXw17ClQvwhOoQzIyMnT11Verc+fOat68ueLi4tw9zWPHjik7O1vR0dEe7xk2bJhH7/SDDz5QXl6e7rrrrgq3M2XKFF100UUaPny47rvvPiUmJurWW2/1aBMdHa2CgoJyvVrgQkB4ArVARkaGGjRo4J6mTp161nYjRozQ4sWLddVVV+mxxx7Txo0b3etOnTolSapXr57He4YMGaI9e/Zo06ZNkqS0tDTdddddql+/foX12Gw2vfHGG3rnnXeUl5enF154oVybkJAQSdLJkyfN7SxgAYG+LgBA5UVHR2vbtm3u1+Hh4Wdt16tXL/38889asWKF1qxZox49eigpKUnPPfecGjVqJD8/P/32228e74mIiFCfPn2UmpqqFi1aaOXKlVq3bt05ayoL5qNHj+ro0aPlwvbo0aOSpMaNG5vYU8Aa6HkCtUBgYKAuvfRS91RReEqlYZWYmKjXX39ds2bN0vz58yWV9hZjY2O1c+fOcu/5y1/+ojfffFPz589Xq1atznk37t69ezVmzBi98sor6tKlixITE1VSUuLR5rvvvtMll1zicc0VuFAQnoAFnThxQtu2bXP3Jvfv369t27YpIyOjUp/71FNP6V//+pf27NmjHTt2aNmyZWrXrp17fXx8vD7//PNy74uPj5fdbteUKVM0dOjQ391GcXGx7rvvPsXHx2vo0KFKTU3Vt99+W25QhM8++6zaB2MAqo0BwHI++eQTQ1K5KTExsVKfO3nyZKNdu3ZGSEiIER4ebvTt29fYt2+fe/2OHTuMkJAQ49ixY+XeO2HCBCMgIMA4dOhQuXWSjNTUVMMwDOOZZ54xmjRpYvzyyy/u9e+8845hs9mMbdu2GYZhGKdOnTIcDoeRnp5eqf0BfIURhgB4uPPOO9WxY0eNGzfOY/mwYcN05MgRvf/++x7L9+/frzZt2mjnzp1q3br1eW1j7ty5Wrp0qT788MMqqxuoSZy2BeBhxowZatCggft1dna2Pv/8cy1atEijRo0q137FihUaPnz4eQenJAUFBenll1+uknoBX6DnCeB33Xjjjdq8ebMefPDBsz5yAtRFhCcAACZx2hYAAJMITwAATCI8AQAwifAEAMAkwhMAAJMITwAATCI8AQAwifAEAMCk/w/G5nUnonNPCwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 500x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5, 4))\n",
    "plt.hist(uncertainty_scores, bins=40)\n",
    "plt.axvline(x=qhat, color='r', linestyle='--')\n",
    "plt.xlabel('1 - s(y,x)')\n",
    "plt.ylabel('Frequency')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Calculate prediction set per example:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[False, False, False, ..., False,  True, False],\n",
       "       [False, False,  True, ..., False, False, False],\n",
       "       [False, False, False, ..., False, False, False],\n",
       "       ...,\n",
       "       [False, False, False, ..., False, False, False],\n",
       "       [False, False, False, ..., False, False, False],\n",
       "       [False, False, False, ..., False,  True, False]])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction_sets = (1 - model.predict_proba(X_calib) <= qhat)\n",
    "prediction_sets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check sizes of prediction sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    1639\n",
       "2     550\n",
       "3      50\n",
       "0       1\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(prediction_sets.sum(axis=1)).value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "View predictions alongside prediction sets:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 8 -> Prediction: 8 -> Pred. set: ['8']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 2 -> Prediction: 2 -> Pred. set: ['2' '4']\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAK4AAADECAYAAAAGYxrSAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAASgklEQVR4nO2de0zV5R/H38cL5+BB8C5IcFBQFBIxm5s6O4UaMtCVm7BaQ+0yV7rykpW1ommzyEuXWf7hBZfg3NRisLUsDbMZli1LWpEYqIlponJE4Kgdnt8fyhnH7+eB78P3oD6/Pq+NjfPh833O833Oe1/O5/N5LjYhhADDaEa3O90BhukMLFxGS1i4jJawcBktYeEyWsLCZbSEhctoCQuX0RIWLqMlLNz/Q+Li4pCVlRXUNm02G958882gtmmFTgl369atsNls+PHHH4Pdn7uKDRs2YPbs2YiNjYXNZsPcuXO75H1OnDgBm82GNWvWdEn7d5rDhw9j4cKFSE5OhtPpRGxsLLKzs3Hs2LFOt9kjiP37vyM/Px8NDQ0YP348/v777zvdHW3Jz8/HwYMHMXv2bKSkpODs2bNYv3497rvvPhw6dAj33nuvcpss3Hb45ptv/E/bsLCwO90dbVmyZAm2b9+OkJAQvy0nJwejR4/GO++8g8LCQuU2g/Ydd+7cuQgLC8OpU6eQlZWFsLAwREdH46OPPgIAVFRUIC0tDU6nEy6XC9u3bw+4/uLFi3jxxRcxevRohIWFITw8HBkZGfjll18M73Xy5EnMnDkTTqcTgwYNwuLFi7Fnzx7YbDbs378/wPf777/H9OnTERERgV69esHtduPgwYOm7snlcsFms3VuQLqAgoICpKWlYdCgQbDb7UhKSsKGDRuk/l9++SVSU1PhcDiQlJSETz/91OBTX1+PRYsWISYmBna7HQkJCcjPz0dLS0uH/amsrMSpU6c69Js4cWKAaAFg+PDhSE5Oxu+//97h9RRBDc58Ph8yMjIQExODd999F3FxcVi4cCG2bt2K6dOn4/7770d+fj569+6N3Nxc1NTU+K+trq5GcXExsrKysG7dOixbtgwVFRVwu904c+aM36+xsRFpaWnYu3cvnn/+ebz22mv47rvv8PLLLxv68/XXX+OBBx7A5cuXkZeXh1WrVqG+vh5paWn44Ycfgnnrt4UNGzbA5XLh1Vdfxdq1axETE4PnnnvO/3BoS1VVFXJycpCRkYG3334bPXr0wOzZs/HVV1/5fZqamuB2u1FYWIjc3Fx8+OGHmDRpEpYvX44lS5Z02J9Ro0YhNze3U/cihMC5c+cwYMCATl0P0QkKCgoEAHH48GG/bc6cOQKAWLVqld926dIlERoaKmw2m9ixY4ffXllZKQCIvLw8v83r9QqfzxfwPjU1NcJut4sVK1b4bWvXrhUARHFxsd/W3NwsRo4cKQCIsrIyIYQQLS0tYvjw4SI9PV20tLT4fZuamsTQoUPFtGnTlO7Z6XSKOXPmKF1jlpqaGgFArF69ul2/pqYmgy09PV0MGzYswOZyuQQAsXv3br/N4/GIqKgoMXbsWL9t5cqVwul0imPHjgVc/8orr4ju3buLU6dO+W23fl6tNrfb3dHtkWzbtk0AEJs3b+7U9UFPhz399NP+3/v06YPExEQ4nU5kZ2f77YmJiejTpw+qq6v9Nrvdjm7dbnTH5/PhwoULCAsLQ2JiIn766Se/3xdffIHo6GjMnDnTb3M4HHjmmWcC+vHzzz+jqqoKjz/+OC5cuIC6ujrU1dWhsbERU6ZMwYEDB0z9O7ybCA0N9f/u8XhQV1cHt9uN6upqeDyeAN8hQ4bg0Ucf9b8ODw9Hbm4ujhw5grNnzwIAdu7cicmTJ6Nv377+8amrq8PUqVPh8/lw4MCBdvsjhDB8NTNDZWUlFixYgAkTJmDOnDnK1wNBDs4cDgcGDhwYYIuIiMA999xj+K4YERGBS5cu+V+3tLTggw8+wMcff4yamhr4fD7/3/r37+///eTJk4iPjze0l5CQEPC6qqoKANodGI/Hg759+5q8O/P4fD6cP38+wNavXz/D9zxVDh48iLy8PJSXl6OpqSngbx6PBxEREf7XCQkJhjEaMWIEgBvpt8jISFRVVeHo0aOGz6yVf/75x1J/Kc6ePYvMzExERERg165d6N69e6faCapwZZ2Q2UWbVUOrVq3C66+/jieffBIrV65Ev3790K1bNyxatKhTT8bWa1avXo3U1FTSp6syBX/99ReGDh0aYCsrK8ODDz7Y6Tb//PNPTJkyBSNHjsS6desQExODkJAQfP7553jvvfc6PUbTpk3DSy+9RP69VejBwuPxICMjA/X19fj2228xZMiQTrd116TDdu3ahYceegibN28OsNfX1wd8gXe5XPjtt98ghAh4ohw/fjzguvj4eAA3/kVOnTq1C3tuJDIyMiAIAoAxY8ZYarO0tBRXr15FSUkJYmNj/faysjLS//jx44Yxak34x8XFAbgxRleuXLkt4+P1ejFjxgwcO3YMe/fuRVJSkqX27pqSb/fu3QOewMCN72C1tbUBtvT0dNTW1qKkpMRv83q92LhxY4DfuHHjEB8fjzVr1uDKlSuG97v1X3kwcTgcmDp1asCP1a8krf+12o6Rx+NBQUEB6X/mzBl89tln/teXL1/GJ598gtTUVERGRgIAsrOzUV5ejj179hiur6+vx7///ttun8ymw3w+H3JyclBeXo6dO3diwoQJHV7TEXfNEzcrKwsrVqzAvHnzMHHiRFRUVKCoqAjDhg0L8Js/fz7Wr1+Pxx57DC+88AKioqJQVFQEh8MBAP4nTLdu3bBp0yZkZGQgOTkZ8+bNQ3R0NGpra1FWVobw8HCUlpa226fS0lJ/Hvn69es4evQo3nrrLQDAzJkzkZKSEtQx2LdvH7xer8H+yCOP4OGHH0ZISAhmzJiB+fPn48qVK9i4cSMGDRpEVvVGjBiBp556CocPH8bgwYOxZcsWnDt3LkDoy5YtQ0lJCbKysjB37lyMGzcOjY2NqKiowK5du3DixIl201WjRo2C2+3uMEBbunQpSkpKMGPGDFy8eNFQcHjiiSc6GBmCzqQiZOkwp9Np8HW73SI5Odlgd7lcIjMz0//a6/WKpUuXiqioKBEaGiomTZokysvLhdvtNqRcqqurRWZmpggNDRUDBw4US5cuFbt37xYAxKFDhwJ8jxw5ImbNmiX69+8v7Ha7cLlcIjs7W+zbt6/D+2xN8VE/BQUFHV5vltZ0mOxn27ZtQgghSkpKREpKinA4HCIuLk7k5+eLLVu2CACipqbG317r2O7Zs0ekpKQIu90uRo4cKXbu3Gl474aGBrF8+XKRkJAgQkJCxIABA8TEiRPFmjVrxLVr1/x+sJAOc7vd7d5fZ7Dd7ID2vP/++1i8eDFOnz6N6OjoO90dpovRUrjNzc0BOU2v14uxY8fC5/NZmnHE6MNd8x1XhVmzZiE2NhapqanweDwoLCxEZWUlioqK7nTXmNuElsJNT0/Hpk2bUFRUBJ/Ph6SkJOzYsQM5OTl3umvMbULLrwoMc9fkcRlGBRYuoyUsXEZLTAdnv/76q+lGW6cnmoFaYSBbdUDZZRN4evQw3tq2bdtI3549expsrWXRW6FCgmvXrpG+VH+p9wJurAC5lcuXL5O+1IQaquIGAM8++6zB1nbmXVsaGxsNtrZpx7ZQ4yALlyi7bFKQ2fVn/MRltISFy2gJC5fREhYuoyW3rXIWjICLCvpkwQO17OSPP/4gfdtOzG6vX8CNtXG3IgtKqACEChplyPpw67IdANJl3q3TMNuyYsUKy32g7k1lHFQCeAp+4jJawsJltISFy2gJC5fREhYuoyV3PKtAZRCoyB0AuaGGrITaujy9LYmJiab7JmtXpdRJ0dHK2bb07t2btFPjEBUVRfpOnjzZYGtdWHor1Gdx9epV0pcqG6vs7WB1FyF+4jJawsJltISFy2gJC5fREkvBmaxspzLHliqBygIjmZ3ijTfeMNja7mbYFqpsLJtjq9IHKmhT2eG8ubmZtFNBlGzHGWr7KVnwSyELPGXjY7YNqzu98xOX0RIWLqMlLFxGS1i4jJawcBktMZ1VoDIIssiQ8lVZjSubbE2VOvPy8khfqrSq0q6svyqToim7SqlT1l+VEvOtO7UDwPr160nfBQsWGGyy/qrch9UyOQU/cRktYeEyWsLCZbSEhctoiengzOpWSbIAhgqCZEGJymFu1FxWWR+uX79usMm2NFJZlUyVh1W2KZKNLzUXVuZLnfZDrRIG6P7KtmtSWeXLwRnD3ISFy2gJC5fREhYuoyUsXEZLLGUVVEqdKr6yKP3EiRMGmyzqpdqQ9SE8PNxg69+/P+lLtSGL0nv16mWwnT59mvS99bR1gM52AEBDQ4PBRk0YB+jSt2zl7qVLlww22Upjql2V8jBPJGf+k7BwGS1h4TJawsJltMTSKt9glHxVArm6ujqDTVYeVmmXCjSoQAUABg8ebLDJVrxSWx3JgknZCTsU1PiGhYWRvlQgJutvbW2twSY7BYcKfmXBmeyercBPXEZLWLiMlrBwGS1h4TJawsJltKRLJpJbLfnK2j1z5ozBRq3QBeioVzY5nCrNyvYIO3/+vMEmi9I9Ho/BJptATZV3ZXudUSVmKuMC0PchuzdqfMeMGUP6UtkcWVZB5TM2Cz9xGS1h4TJawsJltISFy2hJl5y6Y3VFsMyXKkmqnPsrC4yooE02F5YKQGRBCVXGpeb+ynxlwaTKqllqHGRlcpWSusrqbKqkzmf5Mv9JWLiMlrBwGS1h4TJawsJltKRLSr5dBRX1qhzfJMtAUBG5LFNAtRGMDa5VUMkUUMh8qayNyp5vwRgzs/ATl9ESFi6jJSxcRktYuIyWaBWcUWfbqpxLK+uvSvlSZRyoUqesNEsFNiqBp8pnIWtX5SQdq3OuueTL/Cdh4TJawsJltISFy2gJC5fRki6ZSG4V2apZCpVoWuU8YVm7KsccWT3/WGVyuEppVuUeZBtGq2RzeJUvw9yEhctoCQuX0RIWLqMlt+0sX5Wg5OLFi6Z9g3GaT1eVs62OmdX3AtROvKGCtvr6etKX2uA6GFtymYWfuIyWsHAZLWHhMlrCwmW0hIXLaImlrILKvl0q5VZq82RArcxotl/A7Z8Qb7UPlK/KBHUVZFmFyMhI0210RdaGn7iMlrBwGS1h4TJawsJltMR0cEYFUSrn6Mp8KXtDQwPpa/kLfRDmrKqUUClk7aoEXFZRCVIbGxtN+wbjbGez8BOX0RIWLqMlLFxGS1i4jJawcBktMZ1VoCLvYOwVRdmpPbcAtdW4ViNZlX3GVKJ/lc2PfT4f6auyapZ6P5VNoGUrrrtq4rtZ+InLaAkLl9ESFi6jJSxcRkssbcEUjBW2lF0WnKkEfSqB3O1EFsipBHiUryyQo+yy8aV8ZVswWe2v1XI2P3EZLWHhMlrCwmW0hIXLaAkLl9ES01kFlSjQ6iTjYJRQb2cGQWUlbTD6pRKlU32TZRUoX5VNtmXj0BUT4vmJy2gJC5fREhYuoyUsXEZLLAVnsi/jKifIqLRrNUAMBsFY0Wv1vajxkb2X1fOEvV6v5T5Qvla3huInLqMlLFxGS1i4jJawcBktYeEyWmIpqxCMTIFKmVFlArXVjZLvZqgxk43D9evXDTbZ50a1IfssVPpA2WW+ZuEnLqMlLFxGS1i4jJawcBktsbTKV6XEpxLIXb161XS7KufSyuiq4MxqQKsyvrJgR6XkSwVyss9CJeCi7LzKl/lPwsJltISFy2gJC5fREhYuoyWmswoqk4GpKF3mS0WcsuOiqAhZZcWqDKoPKtkKlfKlSulb5kvds0pWQTZmlF02kVwlq8ATyRnmJixcRktYuIyWsHAZLbEUnKnMhVUJSmSBETU3VOarsqGxyik0FCqrkmWlTmpltErARZVrAXrMgtGu1ZIvB2fMfxIWLqMlLFxGS1i4jJawcBkt6ZKSb1dNMpaVHylUImTq/WTHW1G+KmVnWbtUVkGl5Cu7N2oiuMyXer/GxkbSVyVTwFkFhrkJC5fREhYuoyUsXEZL7sqSb0REBOlLBWeyQK6pqcm0r91uN+1L9VdlVbLs7OGQkBBT1wNqwRl1Fq/Mt2fPngabrL8qeuD5uAxzExYuoyUsXEZLWLiMlrBwGS2xlFWQQWUVVEq+Q4YMIX2pTIEsonc4HO11MQAqg0BF2DJUNlVWmbQuG3NqHDweD+nb3NxssMlK51Rmo1+/fqQvTyRnmE7AwmW0hIXLaAkLl9ES05GCyjZDVElSVjqkgitZALN//36DLTIykvQtLi427Tt27FiDbfDgwaQvdR+yAJGyy4JGp9Npul0qEKupqSF9x48fb7DFx8eTvnv37jXYoqOjSV8q8JTNS+bgjGFuwsJltISFy2gJC5fREhYuoyU2YfXcHoa5A/ATl9ESFi6jJSxcRktYuIyWsHAZLWHhMlrCwmW0hIXLaAkLl9GS/wEQex4H+ZrregAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 3 -> Prediction: 3 -> Pred. set: ['3']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 3 -> Prediction: 6 -> Pred. set: ['3' '6']\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAK4AAADECAYAAAAGYxrSAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8g+/7EAAAACXBIWXMAAA9hAAAPYQGoP6dpAAATeElEQVR4nO2de0xUxxfHvwvKLixleSgPKSwCFYoV8dWmmkpVhKporEGb+oeP2hb/aKLG2tA2jY0mJlZFjdpqtNJaqSY+qpg0RWJoTKwSmtpKbIhYUFqsD1TeoHSZ3x+V/bncM7rDLurU80lI3MO5c+fOfr3cM+fMXJMQQoBhNMPncXeAYXoDC5fREhYuoyUsXEZLWLiMlrBwGS1h4TJawsJltISFy2gJC/c/SFxcHLKzs73apslkwqeffurVNj2hV8L96quvYDKZ8PPPP3u7P08M7e3tWLRoEV544QXYbDYEBgZi+PDh2Lx5Mzo7O716rkuXLsFkMmH9+vVebfdJ4fz585g9ezbi4+MREBCAAQMGYPz48Th27Fiv2+znxf79p2hvb8f58+cxdepUxMXFwcfHBz/99BOWLVuGsrIyfPvtt4+7i9pw+fJlNDc3Y/78+Rg0aBDa2tpw6NAhzJgxAzt27MC7776r3CYLV0JoaCjOnDnjYlu8eDFsNhu2bt2K/Px8REZGPqbe6cXUqVMxdepUF9t7772HUaNGIT8/v1fC9doz7oIFCxAYGIja2lpkZ2cjMDAQ0dHR2LZtGwCgoqICEydOhNVqhd1uN9yxbt26hffffx/Dhg1DYGAggoKCMGXKFPz222+Gc12+fBkzZsyA1WpFeHg4li1bhuLiYphMJvz4448uvmVlZXjttddgs9kQEBCA9PR0nDp1qtfXGRcXBwBoaGjodRu9paCgABMnTkR4eDjMZjNSUlLwxRdfSP2PHz+OtLQ0WCwWpKSk4PDhwwafhoYGLF26FDExMTCbzUhMTMTatWvR1dX10P5UVlaitra2V9fi6+uLmJiY3o+j6AUFBQUCgCgvL3fa5s+fLywWi0hJSRGLFy8W27ZtE2PHjhUAREFBgRg0aJBYsWKF2LJlixg6dKjw9fUV1dXVzuPLy8tFQkKCyMvLEzt27BCrVq0S0dHRwmazibq6OqdfS0uLiI+PF/7+/iIvL09s2rRJvPjii2L48OECgCgtLXX6njhxQvj5+YmXX35ZbNiwQWzcuFGkpqYKPz8/UVZW5ta13rlzR9y4cUPU1taKw4cPi8jISGG320VnZ2dvho6kpqZGABDr1q17oN+YMWPEggULxMaNG8WWLVtEZmamACC2bt3q4me328WQIUNEcHCwyMvLE/n5+WLYsGHCx8dHHD9+3OnX2toqUlNTRVhYmPjoo4/E9u3bxbx584TJZBJLlixxaROAWLlypcGWnp7u9nW2tLSIGzduiIsXL4r8/Hzh6+sr5s6d6/bxLufuzUEy4QIQa9ascdpu374t/P39hclkEvv373faKysrDQPR0dEhHA6Hy3lqamqE2WwWq1atcto2bNggAIgjR444be3t7SI5OdlFuF1dXeK5554TWVlZoqury+nb1tYmBg8eLCZPnuzWte7bt08AcP6MHj1anDt3zq1j3cVd4ba1tRlsWVlZIj4+3sVmt9sFAHHo0CGnrbGxUURFRYkRI0Y4batXrxZWq1VcuHDB5fi8vDzh6+sramtrnTZvCDc3N9c5jj4+PiInJ0fcunXL7ePvx+vTYW+//bbz38HBwUhKSoLVasWcOXOc9qSkJAQHB6O6utppM5vN8PH5tzsOhwM3b95EYGAgkpKS8Msvvzj9fvjhB0RHR2PGjBlOm8ViwTvvvOPSj19//RVVVVWYO3cubt68ifr6etTX16O1tRWTJk3CyZMn3fpzOGHCBJSUlODAgQNYvHgx+vfvj9bWVvWB8QL+/v7Ofzc2NqK+vh7p6emorq5GY2Oji++gQYPw+uuvOz8HBQVh3rx5OHv2LK5evQoAOHDgAF555RWEhIQ4x6e+vh4ZGRlwOBw4efLkA/sjhDA8mj2IpUuXoqSkBF9//TWmTJkCh8OBu3fvun38/Xg1OLNYLBg4cKCLzWaz4dlnn4XJZDLYb9++7fzc1dWFzZs34/PPP0dNTQ0cDofzd2FhYc5/X758GQkJCYb2EhMTXT5XVVUBAObPny/tb2NjI0JCQh54TREREYiIiAAA5OTkYM2aNZg8eTKqqqqkwZnD4cCNGzdcbKGhofDz83vguR7GqVOnsHLlSpw+fRptbW0uv2tsbITNZnN+TkxMNIzRkCFDAPw7/RYZGYmqqiqcO3fO8J11c/36dY/625Pk5GQkJycDAObNm4fMzExMnz4dZWVlhr4+DK8K19fXV8ku7ls1tGbNGnzyySd46623sHr1aoSGhsLHxwdLly51687Yk+5j1q1bh7S0NNInMDBQud2cnBx8/PHHOHr0KHJzc0mfP//8E4MHD3axlZaW4tVXX1U+Xzd//PEHJk2ahOTkZOTn5yMmJgZ+fn74/vvvsXHjxl6P0eTJk/HBBx+Qv+8Wel+Rk5OD3NxcXLhwAUlJSUrHPjHTYQcPHsSECRPw5ZdfutgbGhowYMAA52e73Y7ff/8dQgiX/6UXL150OS4hIQHAv38iMzIyvNbP9vZ2ADD8ab6fyMhIlJSUuNiGDx/u0XmPHTuGO3fuoKioCLGxsU57aWkp6X/x4kXDGF24cAHA/2dGEhIS0NLS4tXxUcGdsZTxxKR8fX19Xe7AwL/PYHV1dS62rKws1NXVoaioyGnr6OjAzp07XfxGjRqFhIQErF+/Hi0tLYbz9fxT3pP6+npDfwBg165dAIDRo0dLj7VYLMjIyHD5edgjycPo/qt1f58aGxtRUFBA+l+5cgXfffed83NTUxP27NmDtLQ05yPOnDlzcPr0aRQXFxuOb2howD///PPAPrk7HUY9cnR2dmLPnj3w9/dHSkrKQ9voyRNzx83OzsaqVauwcOFCjB07FhUVFSgsLER8fLyLX25uLrZu3Yo333wTS5YsQVRUFAoLC2GxWADAeYfx8fHBrl27MGXKFAwdOhQLFy5EdHQ06urqUFpaiqCgoAemHPfu3Yvt27dj5syZiI+PR3NzM4qLi1FSUoLp06dj4sSJXh+DEydOoKOjw2CfOXMmMjMz4efnh+nTpyM3NxctLS3YuXMnwsPD8ffffxuOGTJkCBYtWoTy8nJERERg9+7duHbtmovQV6xYgaKiImRnZ2PBggUYNWoUWltbUVFRgYMHD+LSpUsuf+168vzzzyM9Pf2hAVpubi6ampowfvx4REdH4+rVqygsLERlZSU2bNjQq0c2r06HWa1Wg296eroYOnSowW6328W0adOcnzs6OsTy5ctFVFSU8Pf3F+PGjROnT58W6enphimX6upqMW3aNOHv7y8GDhwoli9fLg4dOiQAiDNnzrj4nj17VsyaNUuEhYUJs9ks7Ha7mDNnjjhx4sQDr7G8vFzMnj1bxMbGCrPZLKxWqxg5cqTIz8/36hyuEP+fDpP9fPPNN0IIIYqKikRqaqqwWCwiLi5OrF27VuzevVsAEDU1Nc72use2uLhYpKamCrPZLJKTk8WBAwcM525ubhYffvihSExMFH5+fmLAgAFi7NixYv369eLu3btOP3gwHbZv3z6RkZEhIiIiRL9+/URISIjIyMgQR48e7dV4CSGE6V4HtGfTpk1YtmwZ/vrrL0RHRz/u7jB9jJbCbW9vd5nT7OjowIgRI+BwOJwBCPPf5ol5xlVh1qxZiI2NRVpaGhobG7F3715UVlaisLDwcXeNeURoKdysrCzs2rULhYWFcDgcSElJwf79+/HGG2887q4xjwgtHxUY5omZx2UYFVi4jJawcBktcTs4u3TpkkcnUq3+cbcNWbuUXVbs06+fcRi6SyzdsauECSrjcH+F3P1QBTWy9Czl25uCnJ6oXLOKb3cdxcPgOy6jJSxcRktYuIyWsHAZLXnsmbNr164ZbC+99BLp+9lnnxls48aNI3379+9vsMmCkubmZoMtODiY9KVK8GQBl9lsNthka6y6i6rvR1ZgTQWIVDkkAHJ50ZUrV0jfI0eOGGyZmZmk78iRIw02WRBGjY+neS++4zJawsJltISFy2gJC5fREhYuoyVulzWqpHxV0pqUb/dK2p6MGTPGYJMttKNSoM888wzp273Q8n5kMxBUGlaWmlWJ/t3tFwByY5GeG4R0Q61mlqW+qQ3oZHtBULKRSUllBsFut7vlx3dcRktYuIyWsHAZLWHhMlriUcrX0yAMoAOYuXPnkr499wcD5AEXBZWClfVNFpzdvyz+QcfL7LKtmKiaYCoNDNDBoCxIpQIxWa2xSora050nOeXLPJWwcBktYeEyWsLCZbSEhctoySMrJPdG5E3twiiL/lUKyakoW9ZfKqKnzgXQUbpsNa7KC1GoiFxlxkSWdqZmCgICAkjfvlo97C58x2W0hIXLaAkLl9ESFi6jJW4HZ56md1WCM5kvFQTJUodULSu1mld2Plmw426/ADrok9XCUm3IfKkATzYOVIpaFgjGxMQYbLL0MHU+mS8VtHHKl3kqYeEyWsLCZbSEhctoCQuX0ZI+KSRXmSnwNN0qi2Q7OzsNNpWiaFm7VKR/584d0pdKrcrapfqg8jZ6WbqVWv1LFa0DdKpdtoKZmhWQ9UFFD+7Cd1xGS1i4jJawcBktYeEyWvLYN3amkD24h4WFGWyy+lYqvStLM3r6Jh1ZcEYFg0FBQaSvLBCjUHmjELU5dGxsLOlLjYM3Ai5vvHGpJ3zHZbSEhctoCQuX0RIWLqMlLFxGSzwqJPdGytfd4wGgqanJYJNtaEy1oZLGVUlRy14tdf36dYNNtteZSrE11V/Z7IrNZjPYvPG98awCw/QCFi6jJSxcRktYuIyWPPaUr8qDO1UveuvWLdKXCmyoFa+q/aKCKFm6lUpReyNApJAFZ9Tm0OHh4aSvSm207DooqO+C63GZpxIWLqMlLFxGS1i4jJawcBkt0WpWgdoPbNiwYaRvRUWFwdbS0uJ2H2TF4VQ0LVs1S7Ur27dL1gYFNQ6y1DeV8pUVs6usolaZKeCUL8Pcg4XLaAkLl9ESFi6jJY/sXb6PGio1K9usmdr+SOarshKWCmCowAqggzOVYEdWu0ttGP0k1ONyypd5KmHhMlrCwmW0hIXLaAkLl9GSR/a6qL7i6tWrpD0iIsJgk63GpSJ62SugKGRF3Far1WCjZjAAelZAtqkytSdZaGgo6UutNKZWSwNAYGCgwdZX3yXPKjBPJSxcRktYuIyWsHAZLXlk9bjeCO6o1Cr1ZhuATq3Kgh3qfNRbewC1d9hSNb0qfZCNA1V7Kwv6qHG4efMm6SvbHspTuB6XYe7BwmW0hIXLaAkLl9ESFi6jJVq9y7ehocFtXyplKyu2piJ9WcqXSu/KInoKWR9U9jqjrlk2u0K1IUv5UrMVsoJ6T2cKOOXLPJWwcBktYeEyWsLCZbTE7eCMCqJkqc6+WgFK1aGqbH8kS7fKAhAKKhCT9YEKuGQrgilk40ulcW/fvk36UuMgW2lMbVGlMjYyOOXLMPdg4TJawsJltISFy2gJC5fRkkf2Ll+VGQhq9kDmK4voVVLJ1EwB9ZolQG31r8oeX5SvbAZC5dqoVLAsRS27ZneR9YFfF8Uw92DhMlrCwmW0hIXLaMkT+dYdWXBGPeSrpFtVAkRZEEWdT5ZKVkFlCybqOmT9pVBZaSz7Ljztg6fwHZfREhYuoyUsXEZLWLiMlrBwGS3pk1kFT9PDslQn5evr60v6Um2o+KpE9CopatnMBtUHb0T0nhazy/ZQUykw74vZBr7jMlrCwmW0hIXLaAkLl9ESt4MzlQdsylcleJAFBCo1nNRWSbLgjArEVNK4Ml+V9DBllwVyFLKxocZSZRxkbxSianpVvmNPAza+4zJawsJltISFy2gJC5fREhYuoyUepXxVVqyqtCGLZKkoWyX6V4m8VdpQ2TBahsosiMqqWaoPKu1Sr7wC6A2jeVaBYR4CC5fREhYuoyUsXEZL+qQeV+VhnHr4Vwn6ZMEDlZKU1aFSda8yXyrYUalvlV0b1YYsaKSCK1l6WBboutuu7HhP0/ocnDFPJSxcRktYuIyWsHAZLWHhMlriUSG5N1ZvUpGzyqyCbCUstQpVFiFTEb3KtclmFSi7LPpXKST3dK8y2ZhRaVzZO4JVxoxnFRjmHixcRktYuIyWsHAZLXlkwZnMl3rTi8yXCq5kaVEqeJAFZ5RdJTBSSfnKoPog27SaCq5kbwNSCWiplK/su6De+0sFd7I2ODhjnkpYuIyWsHAZLWHhMlrCwmW05LHPKni6wlYFWaqUmtkICAggfVVWJVOo7Nslm61QmQVRee8vNQ6y/lKzChaLhfTljZ0Z5h4sXEZLWLiMlrBwGS3pky2YVHyph3xqhS5ABzBUQAEAoaGhBptKECVLt1J2b2zBRPmqBKMqb+iR9UtlGyiux2WYXsDCZbSEhctoCQuX0RIWLqMlffK6KJXjqWhYlr6k9gmjZiUAoKmpyWCTRenULAZ1vGofrFarwSaL6Kl2VdLDbW1tpC81CyJbuUvN0Mi+i7CwMINNlqLmWQWGuQcLl9ESFi6jJSxcRkseWXAme3BPTEw02FTeYSur55Vt+ExBBUGtra1ut9vc3Ez6UtcsC7hUNpemvouGhgbSNyQkxGCLiooifal6Wm/UD3NwxjD3YOEyWsLCZbSEhctoCQuX0RKT6IslmAzTx/Adl9ESFi6jJSxcRktYuIyWsHAZLWHhMlrCwmW0hIXLaAkLl9GS/wE1fLl6dR/7uwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(4):\n",
    "    print(f\"True label: {y_calib[i]} -> Prediction: {y_pred[i]} -> Pred. set: {le.classes_[prediction_sets[i]]}\")\n",
    "    print_image_by_id(id=i, images=X_calib, labels=y_calib)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## MAPIE package"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mapie.classification import MapieClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(560, 10)"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mapie_classifier = MapieClassifier(estimator=model, cv='prefit', method='score', n_jobs=-1)\n",
    "\n",
    "mapie_classifier = mapie_classifier.fit(X_calib, y_calib)\n",
    "\n",
    "#alpha = [0.05, 0.1, 0.2]\n",
    "alpha = [0.05]\n",
    "y_pred, y_set = mapie_classifier.predict(X_new, alpha=alpha)\n",
    "\n",
    "y_pred_proba_max = np.max(y_score, axis=1)\n",
    "\n",
    "y_set = np.squeeze(y_set)\n",
    "y_set.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n = len(mapie_classifier.conformity_scores_)\n",
    "\n",
    "scores = mapie_classifier.conformity_scores_\n",
    "\n",
    "quantiles = mapie_classifier.quantiles_\n",
    "\n",
    "plot_scores(n, alpha, scores, quantiles, bins=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mapie.metrics import classification_coverage_score_v2, classification_mean_width_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Marginal coverage: 0.941\n"
     ]
    }
   ],
   "source": [
    "coverage_scores = classification_coverage_score_v2(y_new, y_set)\n",
    "\n",
    "print(f\"Marginal coverage: {coverage_scores[0]:.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Avg. set size: 1.27\n"
     ]
    }
   ],
   "source": [
    "setsize = classification_mean_width_score(y_set)\n",
    "\n",
    "print(f\"Avg. set size: {setsize:.2f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "View predictions alongside prediction sets:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 6 -> Prediction: 4 -> Pred. set: ['4' '6']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 0 -> Prediction: 0 -> Pred. set: ['0']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 0 -> Prediction: 0 -> Pred. set: ['0']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True label: 6 -> Prediction: 4 -> Pred. set: ['4' '6']\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 200x200 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(4):\n",
    "    print(f\"True label: {y_new[i]} -> Prediction: {y_pred[i]} -> Pred. set: {le.classes_[y_set[i]]}\")\n",
    "    print_image_by_id(id=i, images=X_new, labels=y_new)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Check sizes:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    427\n",
       "2    117\n",
       "3     16\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(y_set.sum(axis=1)).value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "env_py",
   "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}