examples/ensemble_heteroscedastic_regression/demo_ensemble_heteroscedastic_regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble Heteroscedastic Regression Demo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Import Statements"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.append(\"../../\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from uq360.algorithms.ensemble_heteroscedastic_regression import EnsembleHeteroscedasticRegression\n",
    "from uq360.algorithms.heteroscedastic_regression import HeteroscedasticRegression\n",
    "from uq360.metrics import picp, mpiw, compute_regression_metrics\n",
    "from uq360.metrics import UncertaintyCharacteristicsCurve as ucc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import torch"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load Boston Housing Prices Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "house_prices_dataset = datasets.load_boston()\n",
    "house_prices_df = pd.DataFrame(house_prices_dataset['data'])\n",
    "house_prices_df.columns = house_prices_dataset['feature_names']\n",
    "\n",
    "all_features = ['RM','CRIM','PTRATIO', 'DIS']\n",
    "X = house_prices_df[all_features].values\n",
    "y = house_prices_dataset['target']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train test split of the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Training data')"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(X_train[:,0], y_train)\n",
    "plt.xlabel('#of rooms')\n",
    "plt.ylabel('house price')\n",
    "plt.title('Training data')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler_y = StandardScaler()\n",
    "y_train = scaler_y.fit_transform(y_train.reshape(-1, 1))\n",
    "scaler_X = StandardScaler()\n",
    "X_train = scaler_X.fit_transform(X_train)\n",
    "X_test = scaler_X.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Ensemble of Heteroscedastic Regressions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = y_train.reshape(-1, 1)\n",
    "y_test = y_test.reshape(-1, 1)\n",
    "\n",
    "# define config for Heteroscedastic Regression\n",
    "config_HR = {\"num_features\": X_train.shape[1], \"num_outputs\": y_train.shape[1], \"num_hidden\": 50, \"batch_size\": 32, \"num_epochs\": 50,\n",
    "                  \"lr\": 0.001}\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "\n",
    "HR_kwargs = {\"model_type\":'mlp',\n",
    "                                   \"config\": config_HR,\n",
    "                                   \"device\": device}\n",
    "\n",
    "#define config for ensemble of Heteroscedastic Regression\n",
    "config = {\"num_models\": 5, \n",
    "          \"model_kwargs\":HR_kwargs, }\n",
    "\n",
    "uq_model = EnsembleHeteroscedasticRegression(model_type='ensembleheteroscedasticregression', \n",
    "                                             config=config, \n",
    "                                             device=device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Training model 0\n",
      "\n",
      "Epoch: 0, loss = 1.459425965944926\n",
      "Epoch: 1, loss = 1.2974590311447778\n",
      "Epoch: 2, loss = 1.1947528372208278\n",
      "Epoch: 3, loss = 1.123661324381828\n",
      "Epoch: 4, loss = 1.070436492562294\n",
      "Epoch: 5, loss = 1.027777373790741\n",
      "Epoch: 6, loss = 0.9917417267958323\n",
      "Epoch: 7, loss = 0.9600538512070973\n",
      "Epoch: 8, loss = 0.9311446895202\n",
      "Epoch: 9, loss = 0.9038868645826975\n",
      "Epoch: 10, loss = 0.8776373863220216\n",
      "Epoch: 11, loss = 0.852043315768242\n",
      "Epoch: 12, loss = 0.82683128118515\n",
      "Epoch: 13, loss = 0.8017907192309699\n",
      "Epoch: 14, loss = 0.776911124587059\n",
      "Epoch: 15, loss = 0.7522002359231313\n",
      "Epoch: 16, loss = 0.728042721748352\n",
      "Epoch: 17, loss = 0.7043740202983221\n",
      "Epoch: 18, loss = 0.6817639420429867\n",
      "Epoch: 19, loss = 0.6601356491446495\n",
      "Epoch: 20, loss = 0.6401551167170207\n",
      "Epoch: 21, loss = 0.6218055884043375\n",
      "Epoch: 22, loss = 0.6054740746816\n",
      "Epoch: 23, loss = 0.590736210346222\n",
      "Epoch: 24, loss = 0.577766587336858\n",
      "Epoch: 25, loss = 0.5663900921742121\n",
      "Epoch: 26, loss = 0.5563266575336456\n",
      "Epoch: 27, loss = 0.5472660486896833\n",
      "Epoch: 28, loss = 0.5391053383549055\n",
      "Epoch: 29, loss = 0.5315107380350431\n",
      "Epoch: 30, loss = 0.5245703980326653\n",
      "Epoch: 31, loss = 0.5178951472043991\n",
      "Epoch: 32, loss = 0.5117233991622925\n",
      "Epoch: 33, loss = 0.5060414920250574\n",
      "Epoch: 34, loss = 0.5005507270495096\n",
      "Epoch: 35, loss = 0.4952728475133578\n",
      "Epoch: 36, loss = 0.4901792034506798\n",
      "Epoch: 37, loss = 0.48524660368760425\n",
      "Epoch: 38, loss = 0.48030299693346024\n",
      "Epoch: 39, loss = 0.47558893511692685\n",
      "Epoch: 40, loss = 0.471492275595665\n",
      "Epoch: 41, loss = 0.46731551115711534\n",
      "Epoch: 42, loss = 0.46336371451616293\n",
      "Epoch: 43, loss = 0.4591771562894185\n",
      "Epoch: 44, loss = 0.45520447691281635\n",
      "Epoch: 45, loss = 0.4514981520672639\n",
      "Epoch: 46, loss = 0.4478612927099069\n",
      "Epoch: 47, loss = 0.443877600133419\n",
      "Epoch: 48, loss = 0.44032202536861104\n",
      "Epoch: 49, loss = 0.43675287937124563\n",
      "\n",
      "Training model 1\n",
      "\n",
      "Epoch: 0, loss = 1.3922461469968161\n",
      "Epoch: 1, loss = 1.2596955299377441\n",
      "Epoch: 2, loss = 1.1690905491511028\n",
      "Epoch: 3, loss = 1.1015110413233438\n",
      "Epoch: 4, loss = 1.0477188676595688\n",
      "Epoch: 5, loss = 1.002597416440646\n",
      "Epoch: 6, loss = 0.9631237983703613\n",
      "Epoch: 7, loss = 0.9273099104563395\n",
      "Epoch: 8, loss = 0.8937771171331407\n",
      "Epoch: 9, loss = 0.8614128778378167\n",
      "Epoch: 10, loss = 0.8298585067192713\n",
      "Epoch: 11, loss = 0.798678124944369\n",
      "Epoch: 12, loss = 0.767591545979182\n",
      "Epoch: 13, loss = 0.7369264215230943\n",
      "Epoch: 14, loss = 0.7072292864322661\n",
      "Epoch: 15, loss = 0.679402602215608\n",
      "Epoch: 16, loss = 0.654103306432565\n",
      "Epoch: 17, loss = 0.6320192217826842\n",
      "Epoch: 18, loss = 0.61342354118824\n",
      "Epoch: 19, loss = 0.5980628704031309\n",
      "Epoch: 20, loss = 0.5851480960845947\n",
      "Epoch: 21, loss = 0.5739423384269078\n",
      "Epoch: 22, loss = 0.5641720965504647\n",
      "Epoch: 23, loss = 0.5549236511190733\n",
      "Epoch: 24, loss = 0.5462071746587752\n",
      "Epoch: 25, loss = 0.5378635103503863\n",
      "Epoch: 26, loss = 0.5302660266558329\n",
      "Epoch: 27, loss = 0.5229076420267422\n",
      "Epoch: 28, loss = 0.5162645230690638\n",
      "Epoch: 29, loss = 0.5098871166507404\n",
      "Epoch: 30, loss = 0.5033780684073765\n",
      "Epoch: 31, loss = 0.4971353660027186\n",
      "Epoch: 32, loss = 0.49102014179031056\n",
      "Epoch: 33, loss = 0.4851145495971043\n",
      "Epoch: 34, loss = 0.4795866981148719\n",
      "Epoch: 35, loss = 0.4741823089619478\n",
      "Epoch: 36, loss = 0.4688098033269247\n",
      "Epoch: 37, loss = 0.46360648423433304\n",
      "Epoch: 38, loss = 0.4586530762414138\n",
      "Epoch: 39, loss = 0.4536480767031511\n",
      "Epoch: 40, loss = 0.4489533963302771\n",
      "Epoch: 41, loss = 0.4441815664370855\n",
      "Epoch: 42, loss = 0.4397908846537273\n",
      "Epoch: 43, loss = 0.43538880099852884\n",
      "Epoch: 44, loss = 0.43119228134552634\n",
      "Epoch: 45, loss = 0.4275206079085668\n",
      "Epoch: 46, loss = 0.42259160429239273\n",
      "Epoch: 47, loss = 0.4893317942818005\n",
      "Epoch: 48, loss = 0.5524873125056425\n",
      "Epoch: 49, loss = 0.4425210493306319\n",
      "\n",
      "Training model 2\n",
      "\n",
      "Epoch: 0, loss = 1.3278434177239737\n",
      "Epoch: 1, loss = 1.238789404431979\n",
      "Epoch: 2, loss = 1.1772491186857224\n",
      "Epoch: 3, loss = 1.1278623392184575\n",
      "Epoch: 4, loss = 1.084568942586581\n",
      "Epoch: 5, loss = 1.0444449583689372\n",
      "Epoch: 6, loss = 1.0061522771914801\n",
      "Epoch: 7, loss = 0.9689363539218903\n",
      "Epoch: 8, loss = 0.9325167238712312\n",
      "Epoch: 9, loss = 0.8963273664315542\n",
      "Epoch: 10, loss = 0.8600370933612188\n",
      "Epoch: 11, loss = 0.8236811657746633\n",
      "Epoch: 12, loss = 0.7873370697100959\n",
      "Epoch: 13, loss = 0.7512826025485994\n",
      "Epoch: 14, loss = 0.7165682514508566\n",
      "Epoch: 15, loss = 0.6846622427304586\n",
      "Epoch: 16, loss = 0.6568249563376108\n",
      "Epoch: 17, loss = 0.6343020995457966\n",
      "Epoch: 18, loss = 0.6159059132138888\n",
      "Epoch: 19, loss = 0.6006258899966874\n",
      "Epoch: 20, loss = 0.5870559612909951\n",
      "Epoch: 21, loss = 0.5747611994544666\n",
      "Epoch: 22, loss = 0.5634356091419855\n",
      "Epoch: 23, loss = 0.5526477719346682\n",
      "Epoch: 24, loss = 0.5429349367817242\n",
      "Epoch: 25, loss = 0.5338201547662417\n",
      "Epoch: 26, loss = 0.5271200935045879\n",
      "Epoch: 27, loss = 0.5241357857982318\n",
      "Epoch: 28, loss = 0.5335411404569944\n",
      "Epoch: 29, loss = 0.559286003311475\n",
      "Epoch: 30, loss = 0.5483831216891608\n",
      "Epoch: 31, loss = 0.50990995268027\n",
      "Epoch: 32, loss = 0.49419424434502923\n",
      "Epoch: 33, loss = 0.488176442682743\n",
      "Epoch: 34, loss = 0.48243577405810356\n",
      "Epoch: 35, loss = 0.47652534022927284\n",
      "Epoch: 36, loss = 0.47239890694618225\n",
      "Epoch: 37, loss = 0.46796507885058714\n",
      "Epoch: 38, loss = 0.46395835652947426\n",
      "Epoch: 39, loss = 0.4598600628475348\n",
      "Epoch: 40, loss = 0.45585419361790025\n",
      "Epoch: 41, loss = 0.45130591342846554\n",
      "Epoch: 42, loss = 0.44694073249896366\n",
      "Epoch: 43, loss = 0.4429003906746705\n",
      "Epoch: 44, loss = 0.43921868875622755\n",
      "Epoch: 45, loss = 0.43604414537549024\n",
      "Epoch: 46, loss = 0.4371031746268272\n",
      "Epoch: 47, loss = 0.4783757428328196\n",
      "Epoch: 48, loss = 0.5238680218656858\n",
      "Epoch: 49, loss = 0.47683953369657195\n",
      "\n",
      "Training model 3\n",
      "\n",
      "Epoch: 0, loss = 1.3529571841160457\n",
      "Epoch: 1, loss = 1.2356265932321548\n",
      "Epoch: 2, loss = 1.155227094888687\n",
      "Epoch: 3, loss = 1.0929226974646251\n",
      "Epoch: 4, loss = 1.0414386441310246\n",
      "Epoch: 5, loss = 0.9974718689918518\n",
      "Epoch: 6, loss = 0.9589815984169642\n",
      "Epoch: 7, loss = 0.9240361899137497\n",
      "Epoch: 8, loss = 0.891183207432429\n",
      "Epoch: 9, loss = 0.8598755995432534\n",
      "Epoch: 10, loss = 0.8298288186391194\n",
      "Epoch: 11, loss = 0.8007048765818278\n",
      "Epoch: 12, loss = 0.7724058230717977\n",
      "Epoch: 13, loss = 0.7448627402385075\n",
      "Epoch: 14, loss = 0.7186810274918874\n",
      "Epoch: 15, loss = 0.6939155186216036\n",
      "Epoch: 16, loss = 0.6708587259054184\n",
      "Epoch: 17, loss = 0.6493490387996037\n",
      "Epoch: 18, loss = 0.62999111165603\n",
      "Epoch: 19, loss = 0.6120904137690862\n",
      "Epoch: 20, loss = 0.5975037366151811\n",
      "Epoch: 21, loss = 0.5850068827470144\n",
      "Epoch: 22, loss = 0.5870728914936384\n",
      "Epoch: 23, loss = 0.6100796113411586\n",
      "Epoch: 24, loss = 0.6108613734443983\n",
      "Epoch: 25, loss = 0.558445858458678\n",
      "Epoch: 26, loss = 0.5376242498556772\n",
      "Epoch: 27, loss = 0.5274405926465988\n",
      "Epoch: 28, loss = 0.5159034083286921\n",
      "Epoch: 29, loss = 0.5088501224915186\n",
      "Epoch: 30, loss = 0.5008390843868256\n",
      "Epoch: 31, loss = 0.4946453546484311\n",
      "Epoch: 32, loss = 0.48753096784154576\n",
      "Epoch: 33, loss = 0.48225405067205435\n",
      "Epoch: 34, loss = 0.47584517796834314\n",
      "Epoch: 35, loss = 0.47935723513364786\n",
      "Epoch: 36, loss = 0.4865008133153121\n",
      "Epoch: 37, loss = 0.48087143525481224\n",
      "Epoch: 38, loss = 0.46405114481846493\n",
      "Epoch: 39, loss = 0.4565367996692658\n",
      "Epoch: 40, loss = 0.4494714637597402\n",
      "Epoch: 41, loss = 0.4454454729954401\n",
      "Epoch: 42, loss = 0.44160209844509757\n",
      "Epoch: 43, loss = 0.43733516956369084\n",
      "Epoch: 44, loss = 0.43378984679778415\n",
      "Epoch: 45, loss = 0.42978949844837194\n",
      "Epoch: 46, loss = 0.42640674486756325\n",
      "Epoch: 47, loss = 0.4225499331951142\n",
      "Epoch: 48, loss = 0.41968092073996865\n",
      "Epoch: 49, loss = 0.41587354491154355\n",
      "\n",
      "Training model 4\n",
      "\n",
      "Epoch: 0, loss = 1.422783484061559\n",
      "Epoch: 1, loss = 1.3289557198683422\n",
      "Epoch: 2, loss = 1.259637340903282\n",
      "Epoch: 3, loss = 1.2027983615795772\n",
      "Epoch: 4, loss = 1.1535203456878662\n",
      "Epoch: 5, loss = 1.1098044315973916\n",
      "Epoch: 6, loss = 1.0706793814897537\n",
      "Epoch: 7, loss = 1.0356728186209998\n",
      "Epoch: 8, loss = 1.0038485825061798\n",
      "Epoch: 9, loss = 0.9742681632439295\n",
      "Epoch: 10, loss = 0.94574869175752\n",
      "Epoch: 11, loss = 0.9175616304079692\n",
      "Epoch: 12, loss = 0.8888697524865469\n",
      "Epoch: 13, loss = 0.8593940635522206\n",
      "Epoch: 14, loss = 0.8290184984604517\n",
      "Epoch: 15, loss = 0.7978785336017609\n",
      "Epoch: 16, loss = 0.7661615908145905\n",
      "Epoch: 17, loss = 0.7342470735311508\n",
      "Epoch: 18, loss = 0.7029431412617365\n",
      "Epoch: 19, loss = 0.6731106291214625\n",
      "Epoch: 20, loss = 0.6454509422183037\n",
      "Epoch: 21, loss = 0.6207683483759563\n",
      "Epoch: 22, loss = 0.5998177776734033\n",
      "Epoch: 23, loss = 0.5822738831241925\n",
      "Epoch: 24, loss = 0.5677824790279071\n",
      "Epoch: 25, loss = 0.5548820594946544\n",
      "Epoch: 26, loss = 0.5492111916343372\n",
      "Epoch: 27, loss = 0.5713565175731976\n",
      "Epoch: 28, loss = 0.5521967485547066\n",
      "Epoch: 29, loss = 0.527326464653015\n",
      "Epoch: 30, loss = 0.5185572952032089\n",
      "Epoch: 31, loss = 0.5080018763740858\n",
      "Epoch: 32, loss = 0.5025775333245595\n",
      "Epoch: 33, loss = 0.4965160861611366\n",
      "Epoch: 34, loss = 0.49162967875599856\n",
      "Epoch: 35, loss = 0.4868839358290037\n",
      "Epoch: 36, loss = 0.48219655329982436\n",
      "Epoch: 37, loss = 0.47786381964882213\n",
      "Epoch: 38, loss = 0.47384978954990703\n",
      "Epoch: 39, loss = 0.46992064391573274\n",
      "Epoch: 40, loss = 0.46615597233176226\n",
      "Epoch: 41, loss = 0.46242659042278933\n",
      "Epoch: 42, loss = 0.4589013097186883\n",
      "Epoch: 43, loss = 0.45543017610907555\n",
      "Epoch: 44, loss = 0.4521737719575564\n",
      "Epoch: 45, loss = 0.44885135938723886\n",
      "Epoch: 46, loss = 0.4457805491983891\n",
      "Epoch: 47, loss = 0.442691296339035\n",
      "Epoch: 48, loss = 0.4397575495143732\n",
      "Epoch: 49, loss = 0.436941626171271\n"
     ]
    }
   ],
   "source": [
    "uq_model = uq_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_mean, y_lower, y_upper = uq_model.predict(X_test)\n",
    "y_mean, y_lower, y_upper = scaler_y.inverse_transform(y_mean), scaler_y.inverse_transform(y_lower), scaler_y.inverse_transform(y_upper)\n",
    "X_test = scaler_X.inverse_transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_dict = {'rooms': X_test[:,0],\n",
    "             'y': y_test[:,0],\n",
    "             'y_mean': y_mean[:,0],\n",
    "             'y_upper': y_upper[:,0],\n",
    "             'y_lower': y_lower[:,0]\n",
    "            }\n",
    "\n",
    "pred_df = pd.DataFrame(data=pred_dict)\n",
    "\n",
    "pred_df_sorted = pred_df.sort_values(by='rooms')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Test data PICP: 0.93 MPIW: 17.10')"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(pred_df_sorted['rooms'], pred_df_sorted['y'], 'o', label='ground truth')\n",
    "plt.plot(pred_df_sorted['rooms'], pred_df_sorted['y_mean'], '-', lw=2, label='mean')\n",
    "plt.fill_between(pred_df_sorted['rooms'],\n",
    "                         pred_df_sorted['y_upper'],\n",
    "                         pred_df_sorted['y_lower'],\n",
    "                         alpha=0.3, label='total uncertianty')\n",
    "plt.legend()\n",
    "plt.xlabel('#of rooms')\n",
    "plt.ylabel('house price in $1000')\n",
    "plt.title('Test data PICP: {:.2f} MPIW: {:.2f}'.format(\n",
    "    picp(y_test, y_lower, y_upper),\n",
    "    mpiw(y_lower, y_upper))      \n",
    "         )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Let us now plot the AUUCC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def form_D(yhat, zhatl, zhatu):\n",
    "    # a handy routine to format data as needed by the UCC fit() method\n",
    "    D = np.zeros([yhat.shape[0],3])\n",
    "    D[:, 0] = yhat\n",
    "    D[:, 1] = zhatl\n",
    "    D[:, 2] = zhatu\n",
    "    return D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<uq360.metrics.uncertainty_characteristics_curve.uncertainty_characteristics_curve.UncertaintyCharacteristicsCurve at 0x7f9e903f3b00>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# form matrix for ucc:\n",
    "X_for_ucc = form_D(y_mean[:,0], y_mean[:,0] - y_lower[:,0], y_upper[:,0] - y_mean[:,0])\n",
    "\n",
    "# create an instance of ucc and fit data\n",
    "u = ucc()\n",
    "u.fit(X_for_ucc, y_test[:,0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUUCC is 0.38, OP = (0.10, 0.07)\n"
     ]
    }
   ],
   "source": [
    "u.set_coordinates(x_axis_name='bandwidth', y_axis_name='missrate', normalize=True)\n",
    "auucc, op_info = u.plot_UCC(titlestr=('AUUCC'))\n",
    "print(\"AUUCC is %.2f, OP = (%.2f, %.2f)\" % (auucc, op_info[0]/op_info[2], op_info[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [default]",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}