tutorial/tutorial_prediction.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"figure/clairvoyance_logo.png\">\n",
    "\n",
    "# Clairvoyance: Time-series prediction\n",
    "\n",
    "## ML-AIM (http://vanderschaar-lab.com/)\n",
    "\n",
    "This notebook describes the user-guide of a time-series predictions application using Clairvoyance framework. Time-series prediction is defined as following: utilize both static and temporal features to predict certain labels in the future. For instance, using the temporal data (vitals, lab tests) and static data (demographic information), we predict 'whether the patient will die at the end of hospital stay' or 'whether the patient will get ventilator after 4 hours'. \n",
    "- One-shot prediction: Predict the patient state at the end of the time-series at certain time point.\n",
    "  - Example: Predict patient mortality (at the end of the hospital stays) after 24 hours from the admission.\n",
    "- Rolling window (online) prediction:\n",
    "  - Example: Predict ventilator after 24 hours from the current time point.\n",
    " \n",
    "<img src=\"figure/time-series-prediction-definition.png\">\n",
    "\n",
    "To run this tutorial, you need:\n",
    "### Temporal and static datasets for training and testing\n",
    "\n",
    "If users come with their own temporal and static datasets for training and testing, the users should save those files as 'data_name_temporal_train_data_eav.csv.gz', 'data_name_static_train_data.csv.gz', 'data_name_temporal_test_data_eav.csv.gz', 'data_name_static_test_data.csv.gz' in '../datasets/data/data_name/' directory.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Time-series prediction pipeline summary\n",
    "\n",
    "<img src=\"figure/time-series-prediction-block-diagram.png\">\n",
    "\n",
    "### Step 1: Load dataset\n",
    "  - Extract csv files from the original raw datasets in ../datasets/data/data_name/ directory.  \n",
    "  \n",
    "### Step 2: Preprocess dataset\n",
    "  - Preprocessing the raw data using various filters such as (1) replacing negative values to NaN, (2) do one-hot encidng for certain features, (3) do normalization.  \n",
    "  \n",
    "### Step 3: Define problem\n",
    "  - Set the time-series prediction problem that we want to solve. Set the problem (whether it is one-shot or online prediction), set the label, set the maximum sequence length, and set the treatment features. We also define the metric for evaluation and the task itself (whether classification or regression).\n",
    "\n",
    "### Step 4: Impute dataset\n",
    "  - Impute missing values in the preprocessed static and temporal datasets and return complete datasets.\n",
    "  \n",
    "### Step 5: Feature selection\n",
    "  - Select the relevant static and temporal features to the labels. You can skip the feature selection (set feature selection method = None).\n",
    "  \n",
    "### Step 6: Time-series model fit and predict\n",
    "  - After finishing the data preparation, we define the predictive models and train the model using the training dataset. After training, we use the trained model to predict the labels of the testing dataset.\n",
    "  \n",
    "### Step 7: Estimate uncertainty\n",
    "  - Estimate uncertainty of the predictor models and returns the uncertainty of the predictions.\n",
    "\n",
    "### Step 8: Interpret predictions\n",
    "  - Interpret the trained predictor model and return the instance-wise feature and temporal importance.\n",
    "\n",
    "### Step 9: Visualize results\n",
    "  - Visualize the various results such as performance, predictions, uncertainties, and interpretations."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 0: Import necessary packages\n",
    "\n",
    "Import necessary packages for the entire tutorials."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Necessary packages\n",
    "from __future__ import absolute_import\n",
    "from __future__ import division\n",
    "from __future__ import print_function\n",
    "\n",
    "import numpy as np\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import sys\n",
    "sys.path.append('../')\n",
    "\n",
    "from utils import PipelineComposer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 1: Load dataset\n",
    "\n",
    "Extract temporal and static datasets from 'data_name_temporal_train_data_eav.csv.gz', 'data_name_static_train_data.csv.gz', 'data_name_temporal_test_data_eav.csv.gz', 'data_name_static_test_data.csv.gz' in '../datasets/data/data_name/' directory.\n",
    "\n",
    "- CSVLoader: Load csv files from the original raw datasets in ../datasets/data/data_name/ directory.\n",
    "- file_names: mimic in this tutorial."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish data loading.\n"
     ]
    }
   ],
   "source": [
    "from datasets import CSVLoader\n",
    "\n",
    "# Define data name\n",
    "data_name = 'mimic'\n",
    "# Define data dictionary\n",
    "data_directory = '../datasets/data/'+data_name + '/' + data_name + '_'\n",
    "\n",
    "# Load train and test datasets\n",
    "data_loader_training = CSVLoader(static_file=data_directory + 'static_train_data.csv.gz',\n",
    "                                 temporal_file=data_directory + 'temporal_train_data_eav.csv.gz')\n",
    "\n",
    "data_loader_testing = CSVLoader(static_file=data_directory + 'static_test_data.csv.gz',\n",
    "                                temporal_file=data_directory + 'temporal_test_data_eav.csv.gz')\n",
    "\n",
    "dataset_training = data_loader_training.load()\n",
    "dataset_testing = data_loader_testing.load()\n",
    "\n",
    "print('Finish data loading.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: Preprocess dataset\n",
    "\n",
    "Preprocess the raw data using multiple filters. In this tutorial, we replace all the negative values to NaN (using NegativeFilter), do one-hot encoding on 'admission_type' feature (using OneHotEncoder), and do MinMax Normalization (using Normalization). Preprocessing is done for both training and testing datasets. \n",
    "  - NegativeFilter: Replace negative values to NaN\n",
    "  - OneHotEncoder: One hot encoding certain features\n",
    "    - one_hot_encoding: input features that need to be one-hot encoded\n",
    "  - Normalization (3 options): MinMax, Standard, None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish preprocessing.\n"
     ]
    }
   ],
   "source": [
    "from preprocessing import FilterNegative, OneHotEncoder, Normalizer\n",
    "\n",
    "# (1) filter out negative values\n",
    "negative_filter = FilterNegative()\n",
    "# (2) one-hot encode categorical features\n",
    "one_hot_encoding = 'admission_type'\n",
    "onehot_encoder = OneHotEncoder(one_hot_encoding_features=[one_hot_encoding])\n",
    "# (3) Normalize features: 3 options (minmax, standard, none)\n",
    "normalization = 'minmax'\n",
    "normalizer = Normalizer(normalization)\n",
    "\n",
    "# Data preprocessing\n",
    "filter_pipeline = PipelineComposer(negative_filter, onehot_encoder, normalizer)\n",
    "\n",
    "dataset_training = filter_pipeline.fit_transform(dataset_training)\n",
    "dataset_testing = filter_pipeline.transform(dataset_testing)\n",
    "\n",
    "print('Finish preprocessing.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 3: Define problem   \n",
    "\n",
    "Set the time-series prediction problem that we want to solve. Set the problem (whether it is one-shot or online prediction), set the label, set the maximum sequence length, and set the treatment features. We also define the metric for evaluation and the task itself (whether classification or regression). In this tutorial, we predict whether the patients will get ventilator after 4 hours (online setting).\n",
    "  - problem: 'one-shot'(one time prediction) or 'online'(rolling window prediction)\n",
    "    - 'one-shot': one time prediction at the end of the time-series \n",
    "    - 'online': preditcion at every time stamps of the time-series\n",
    "  - max_seq_len: maximum sequence length of time-series sequence\n",
    "  - label_name: the column name for the label(s)\n",
    "  - treatment: the column name for treatments\n",
    "  - window: x-hour ahead prediction.\n",
    "  \n",
    "  - other parameters:\n",
    "    - metric_name: auc, apr, mse, mae\n",
    "    - task: classification or regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 18490/18490 [00:50<00:00, 365.54it/s]\n",
      "100%|██████████| 18490/18490 [00:47<00:00, 385.65it/s]\n",
      "100%|██████████| 18490/18490 [00:51<00:00, 357.71it/s]\n",
      "100%|██████████| 4610/4610 [00:08<00:00, 545.16it/s]\n",
      "100%|██████████| 4610/4610 [00:07<00:00, 576.70it/s]\n",
      "100%|██████████| 4610/4610 [00:08<00:00, 532.08it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish defining problem.\n"
     ]
    }
   ],
   "source": [
    "from preprocessing import ProblemMaker\n",
    "\n",
    "# Define parameters\n",
    "problem = 'online'\n",
    "max_seq_len = 24\n",
    "label_name = 'ventilator'\n",
    "treatment = None\n",
    "window = 4\n",
    "\n",
    "# Define problem \n",
    "problem_maker = ProblemMaker(problem=problem, label=[label_name],\n",
    "                             max_seq_len=max_seq_len, treatment=treatment, window = window)\n",
    "\n",
    "dataset_training = problem_maker.fit_transform(dataset_training)\n",
    "dataset_testing = problem_maker.fit_transform(dataset_testing)\n",
    "\n",
    "# Set other parameters\n",
    "metric_name = 'auc'\n",
    "task = 'classification'\n",
    "\n",
    "metric_sets = [metric_name]\n",
    "metric_parameters =  {'problem': problem, 'label_name': [label_name]}\n",
    "\n",
    "print('Finish defining problem.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 4: Impute dataset\n",
    "\n",
    "Impute missing values in the preprocessed static and temporal datasets and return complete datasets.\n",
    "  - Static imputation (6 options): mean, median, mice, missforest, knn, gain\n",
    "  - Temporal imputation (8 options): mean, median, linear, quadratic, cubic, spline, mrnn, tgain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish imputation.\n"
     ]
    }
   ],
   "source": [
    "from imputation import Imputation\n",
    "\n",
    "# Set imputation models\n",
    "static_imputation_model = 'median'\n",
    "temporal_imputation_model = 'median'\n",
    "\n",
    "# Impute the missing data\n",
    "static_imputation = Imputation(imputation_model_name = static_imputation_model, data_type = 'static')\n",
    "temporal_imputation = Imputation(imputation_model_name = temporal_imputation_model, data_type = 'temporal')\n",
    "\n",
    "imputation_pipeline = PipelineComposer(static_imputation, temporal_imputation)\n",
    "\n",
    "dataset_training = imputation_pipeline.fit_transform(dataset_training)\n",
    "dataset_testing = imputation_pipeline.transform(dataset_testing)\n",
    "\n",
    "print('Finish imputation.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 5: Feature selection\n",
    "\n",
    "Select the relevant static and temporal features to the labels. If you do not want, we can skip the feature selection (set feature selection method = None).\n",
    "  - feature selection method: \n",
    "      - feature_selection_model: greedy-addition, greedy-deletion, recursive-addition, recursive-deletion, None\n",
    "      - feature_number: selected featuer number"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish feature selection.\n"
     ]
    }
   ],
   "source": [
    "from feature_selection import FeatureSelection\n",
    "\n",
    "# Set feature selection parameters\n",
    "static_feature_selection_model = None\n",
    "temporal_feature_selection_model = None\n",
    "static_feature_selection_number = None\n",
    "temporal_feature_selection_number = None\n",
    "\n",
    "# Select relevant features\n",
    "static_feature_selection = \\\n",
    "FeatureSelection(feature_selection_model_name = static_feature_selection_model,\n",
    "                 feature_type = 'static', feature_number = static_feature_selection_number,\n",
    "                 task = task, metric_name = metric_name,\n",
    "                 metric_parameters = metric_parameters)\n",
    "\n",
    "temporal_feature_selection = \\\n",
    "FeatureSelection(feature_selection_model_name = temporal_feature_selection_model,\n",
    "                 feature_type = 'temporal', feature_number = temporal_feature_selection_number,\n",
    "                 task = task, metric_name = metric_name,\n",
    "                 metric_parameters = metric_parameters)\n",
    "\n",
    "feature_selection_pipeline = PipelineComposer(static_feature_selection, temporal_feature_selection)\n",
    "\n",
    "dataset_training = feature_selection_pipeline.fit_transform(dataset_training)\n",
    "dataset_testing = feature_selection_pipeline.transform(dataset_testing)\n",
    "\n",
    "print('Finish feature selection.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 6: Time-series model fit and predict\n",
    "\n",
    "After finishing the data preparation, we define the predictive models (6 options, RNN, GRU, LSTM, Attention, Temporal CNN, and Transformer), and train the model using the training dataset. We set validation set as the 20% of the training set for early stopping and best model saving. After training, we use the trained model to predict the labels of the testing dataset.\n",
    "\n",
    "- predictor_parameters:\n",
    "  - model_name: rnn, gru, lstm, attention, tcn, transformer\n",
    "  - model_parameters: network parameters such as numer of layers\n",
    "    - h_dim: hidden dimensions\n",
    "    - n_layer: layer number\n",
    "    - n_head: head number (only for transformer model)\n",
    "    - batch_size: number of samples in mini-batch\n",
    "    - epochs: number of epochs\n",
    "    - learning_rate: learning rate\n",
    "  - static_mode: how to utilize static features (concatenate or None)\n",
    "  - time_mode: how to utilize time information (concatenate or None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 14792 samples, validate on 3698 samples\n",
      "Epoch 1/20\n",
      "14792/14792 [==============================] - 5s 339us/sample - loss: 0.3446 - val_loss: 0.2807\n",
      "Epoch 2/20\n",
      "14792/14792 [==============================] - 4s 273us/sample - loss: 0.2759 - val_loss: 0.2720\n",
      "Epoch 3/20\n",
      "14792/14792 [==============================] - 4s 267us/sample - loss: 0.2583 - val_loss: 0.2439\n",
      "Epoch 4/20\n",
      "14792/14792 [==============================] - 4s 270us/sample - loss: 0.2479 - val_loss: 0.2400\n",
      "Epoch 5/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2416 - val_loss: 0.2338\n",
      "Epoch 6/20\n",
      "14792/14792 [==============================] - 4s 267us/sample - loss: 0.2370 - val_loss: 0.2280\n",
      "Epoch 7/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2344 - val_loss: 0.2296\n",
      "Epoch 8/20\n",
      "14792/14792 [==============================] - 4s 264us/sample - loss: 0.2310 - val_loss: 0.2255\n",
      "Epoch 9/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2280 - val_loss: 0.2214\n",
      "Epoch 10/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2271 - val_loss: 0.2209\n",
      "Epoch 11/20\n",
      "14792/14792 [==============================] - 4s 268us/sample - loss: 0.2248 - val_loss: 0.2188\n",
      "Epoch 12/20\n",
      "14792/14792 [==============================] - 4s 268us/sample - loss: 0.2240 - val_loss: 0.2262\n",
      "Epoch 13/20\n",
      "14792/14792 [==============================] - 4s 269us/sample - loss: 0.2233 - val_loss: 0.2162\n",
      "Epoch 14/20\n",
      "14792/14792 [==============================] - 4s 270us/sample - loss: 0.2222 - val_loss: 0.2194\n",
      "Epoch 15/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2190 - val_loss: 0.2138\n",
      "Epoch 16/20\n",
      "14792/14792 [==============================] - 4s 264us/sample - loss: 0.2175 - val_loss: 0.2145\n",
      "Epoch 17/20\n",
      "14792/14792 [==============================] - 4s 264us/sample - loss: 0.2170 - val_loss: 0.2139\n",
      "Epoch 18/20\n",
      "14792/14792 [==============================] - 4s 268us/sample - loss: 0.2152 - val_loss: 0.2120\n",
      "Epoch 19/20\n",
      "14792/14792 [==============================] - 4s 267us/sample - loss: 0.2141 - val_loss: 0.2107\n",
      "Epoch 20/20\n",
      "14792/14792 [==============================] - 4s 265us/sample - loss: 0.2123 - val_loss: 0.2100\n",
      "Finish predictor model training and testing.\n"
     ]
    }
   ],
   "source": [
    "from prediction import prediction\n",
    "\n",
    "# Set predictive model\n",
    "model_name = 'gru'\n",
    "\n",
    "# Set model parameters\n",
    "model_parameters = {'h_dim': 100,\n",
    "                    'n_layer': 2,\n",
    "                    'n_head': 2,\n",
    "                    'batch_size': 128,\n",
    "                    'epoch': 20,\n",
    "                    'model_type': model_name,\n",
    "                    'learning_rate': 0.001,\n",
    "                    'static_mode': 'concatenate',\n",
    "                    'time_mode': 'concatenate',\n",
    "                    'verbose': True}\n",
    "\n",
    "# Set up validation for early stopping and best model saving\n",
    "dataset_training.train_val_test_split(prob_val=0.2, prob_test = 0.0)\n",
    "\n",
    "# Train the predictive model\n",
    "pred_class = prediction(model_name, model_parameters, task)\n",
    "pred_class.fit(dataset_training)\n",
    "# Return the predictions on the testing set\n",
    "test_y_hat = pred_class.predict(dataset_testing)\n",
    "\n",
    "print('Finish predictor model training and testing.')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 7: Estimate uncertainty\n",
    "\n",
    "Estimate uncertainty of the predictor models and returns the uncertainty of the predictions (test_ci_hat).\n",
    "\n",
    "- uncertainty_parameters:\n",
    "  - uncertainty estimation model name (ensemble)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 14792 samples, validate on 3698 samples\n",
      "Epoch 1/20\n",
      "14792/14792 [==============================] - 2s 162us/sample - loss: 0.3353 - val_loss: 0.2710\n",
      "Epoch 2/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2743 - val_loss: 0.2540\n",
      "Epoch 3/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2593 - val_loss: 0.2453\n",
      "Epoch 4/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2513 - val_loss: 0.2420\n",
      "Epoch 5/20\n",
      "14792/14792 [==============================] - 2s 111us/sample - loss: 0.2491 - val_loss: 0.2391\n",
      "Epoch 6/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2436 - val_loss: 0.2365\n",
      "Epoch 7/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2420 - val_loss: 0.2321\n",
      "Epoch 8/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2399 - val_loss: 0.2345\n",
      "Epoch 9/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2374 - val_loss: 0.2306\n",
      "Epoch 10/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2338 - val_loss: 0.2299\n",
      "Epoch 11/20\n",
      "14792/14792 [==============================] - 2s 110us/sample - loss: 0.2326 - val_loss: 0.2256\n",
      "Epoch 12/20\n",
      "14792/14792 [==============================] - 2s 110us/sample - loss: 0.2320 - val_loss: 0.2304\n",
      "Epoch 13/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2314 - val_loss: 0.2248\n",
      "Epoch 14/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2288 - val_loss: 0.2213\n",
      "Epoch 15/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2271 - val_loss: 0.2234\n",
      "Epoch 16/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2270 - val_loss: 0.2233\n",
      "Epoch 17/20\n",
      "14792/14792 [==============================] - 2s 107us/sample - loss: 0.2257 - val_loss: 0.2328\n",
      "Epoch 18/20\n",
      "14792/14792 [==============================] - 2s 108us/sample - loss: 0.2267 - val_loss: 0.2253\n",
      "Epoch 19/20\n",
      "14792/14792 [==============================] - 2s 111us/sample - loss: 0.2250 - val_loss: 0.2203\n",
      "Epoch 20/20\n",
      "14792/14792 [==============================] - 2s 109us/sample - loss: 0.2230 - val_loss: 0.2223\n",
      "Train on 14792 samples, validate on 3698 samples\n",
      "Epoch 1/20\n",
      "14792/14792 [==============================] - 6s 407us/sample - loss: 0.3566 - val_loss: 0.3205\n",
      "Epoch 2/20\n",
      "14792/14792 [==============================] - 4s 294us/sample - loss: 0.2895 - val_loss: 0.2627\n",
      "Epoch 3/20\n",
      "14792/14792 [==============================] - 4s 303us/sample - loss: 0.2693 - val_loss: 0.2635\n",
      "Epoch 4/20\n",
      "14792/14792 [==============================] - 4s 300us/sample - loss: 0.2583 - val_loss: 0.2536\n",
      "Epoch 5/20\n",
      "14792/14792 [==============================] - 4s 294us/sample - loss: 0.2554 - val_loss: 0.2470\n",
      "Epoch 6/20\n",
      "14792/14792 [==============================] - 4s 296us/sample - loss: 0.2490 - val_loss: 0.2420\n",
      "Epoch 7/20\n",
      "14792/14792 [==============================] - 4s 293us/sample - loss: 0.2459 - val_loss: 0.2492\n",
      "Epoch 8/20\n",
      "14792/14792 [==============================] - 4s 295us/sample - loss: 0.2446 - val_loss: 0.2356\n",
      "Epoch 9/20\n",
      "14792/14792 [==============================] - 4s 302us/sample - loss: 0.2409 - val_loss: 0.2329\n",
      "Epoch 10/20\n",
      "14792/14792 [==============================] - 4s 304us/sample - loss: 0.2363 - val_loss: 0.2302\n",
      "Epoch 11/20\n",
      "14792/14792 [==============================] - 4s 296us/sample - loss: 0.2353 - val_loss: 0.2330\n",
      "Epoch 12/20\n",
      "14792/14792 [==============================] - 5s 311us/sample - loss: 0.2330 - val_loss: 0.2278\n",
      "Epoch 13/20\n",
      "14792/14792 [==============================] - 4s 302us/sample - loss: 0.2318 - val_loss: 0.2264\n",
      "Epoch 14/20\n",
      "14792/14792 [==============================] - 5s 305us/sample - loss: 0.2295 - val_loss: 0.2239\n",
      "Epoch 15/20\n",
      "14792/14792 [==============================] - 4s 303us/sample - loss: 0.2274 - val_loss: 0.2259\n",
      "Epoch 16/20\n",
      "14792/14792 [==============================] - 4s 299us/sample - loss: 0.2262 - val_loss: 0.2251\n",
      "Epoch 17/20\n",
      "14792/14792 [==============================] - 4s 300us/sample - loss: 0.2264 - val_loss: 0.2206\n",
      "Epoch 18/20\n",
      "14792/14792 [==============================] - 4s 297us/sample - loss: 0.2239 - val_loss: 0.2213\n",
      "Epoch 19/20\n",
      "14792/14792 [==============================] - 4s 298us/sample - loss: 0.2227 - val_loss: 0.2205\n",
      "Epoch 20/20\n",
      "14792/14792 [==============================] - 4s 298us/sample - loss: 0.2210 - val_loss: 0.2178\n",
      "Finish uncertainty estimation\n"
     ]
    }
   ],
   "source": [
    "from uncertainty import uncertainty\n",
    "\n",
    "# Set uncertainty model\n",
    "uncertainty_model_name = 'ensemble'\n",
    "\n",
    "# Train uncertainty model\n",
    "uncertainty_model = uncertainty(uncertainty_model_name, model_parameters, pred_class, task)\n",
    "uncertainty_model.fit(dataset_training)\n",
    "# Return uncertainty of the trained predictive model\n",
    "test_ci_hat = uncertainty_model.predict(dataset_testing)\n",
    "\n",
    "print('Finish uncertainty estimation')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 8: Interpret predictions.\n",
    "\n",
    "Interpret the trained predictor model and return the instance-wise feature and temporal importance.\n",
    "\n",
    "- interpretor_parameters:\n",
    "  - interpretation_model_name: interpretation model name (tinvase)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 14792 samples, validate on 3698 samples\n",
      "Epoch 1/20\n",
      "14792/14792 [==============================] - 13s 888us/sample - loss: 0.1852 - select_loss: 0.1424 - predict_loss: 0.1836 - val_loss: 0.1234 - val_select_loss: 0.1415 - val_predict_loss: 0.1219\n",
      "Epoch 2/20\n",
      "14792/14792 [==============================] - 10s 683us/sample - loss: 0.1184 - select_loss: 0.1453 - predict_loss: 0.1169 - val_loss: 0.1141 - val_select_loss: 0.1424 - val_predict_loss: 0.1127\n",
      "Epoch 3/20\n",
      "14792/14792 [==============================] - 10s 674us/sample - loss: 0.1084 - select_loss: 0.1421 - predict_loss: 0.1070 - val_loss: 0.0985 - val_select_loss: 0.1418 - val_predict_loss: 0.0971\n",
      "Epoch 4/20\n",
      "14792/14792 [==============================] - 10s 695us/sample - loss: 0.1014 - select_loss: 0.1402 - predict_loss: 0.1000 - val_loss: 0.0932 - val_select_loss: 0.1394 - val_predict_loss: 0.0919\n",
      "Epoch 5/20\n",
      "14792/14792 [==============================] - 11s 720us/sample - loss: 0.0961 - select_loss: 0.1375 - predict_loss: 0.0948 - val_loss: 0.0810 - val_select_loss: 0.1331 - val_predict_loss: 0.0797\n",
      "Epoch 6/20\n",
      "14792/14792 [==============================] - 10s 697us/sample - loss: 0.0662 - select_loss: 0.1366 - predict_loss: 0.0649 - val_loss: 0.0562 - val_select_loss: 0.1361 - val_predict_loss: 0.0549\n",
      "Epoch 7/20\n",
      "14792/14792 [==============================] - 10s 695us/sample - loss: 0.0548 - select_loss: 0.1371 - predict_loss: 0.0534 - val_loss: 0.0505 - val_select_loss: 0.1368 - val_predict_loss: 0.0491\n",
      "Epoch 8/20\n",
      "14792/14792 [==============================] - 11s 712us/sample - loss: 0.0499 - select_loss: 0.1385 - predict_loss: 0.0485 - val_loss: 0.0429 - val_select_loss: 0.1385 - val_predict_loss: 0.0415\n",
      "Epoch 9/20\n",
      "14792/14792 [==============================] - 11s 713us/sample - loss: 0.0378 - select_loss: 0.1387 - predict_loss: 0.0364 - val_loss: 0.0340 - val_select_loss: 0.1409 - val_predict_loss: 0.0326\n",
      "Epoch 10/20\n",
      "14792/14792 [==============================] - 11s 720us/sample - loss: 0.0309 - select_loss: 0.1396 - predict_loss: 0.0295 - val_loss: 0.0266 - val_select_loss: 0.1379 - val_predict_loss: 0.0252\n",
      "Epoch 11/20\n",
      "14792/14792 [==============================] - 11s 711us/sample - loss: 0.0267 - select_loss: 0.1372 - predict_loss: 0.0253 - val_loss: 0.0247 - val_select_loss: 0.1357 - val_predict_loss: 0.0233\n",
      "Epoch 12/20\n",
      "14792/14792 [==============================] - 10s 705us/sample - loss: 0.0242 - select_loss: 0.1338 - predict_loss: 0.0228 - val_loss: 0.0218 - val_select_loss: 0.1307 - val_predict_loss: 0.0205\n",
      "Epoch 13/20\n",
      "14792/14792 [==============================] - 10s 667us/sample - loss: 0.0211 - select_loss: 0.1305 - predict_loss: 0.0198 - val_loss: 0.0196 - val_select_loss: 0.1289 - val_predict_loss: 0.0183\n",
      "Epoch 14/20\n",
      "14792/14792 [==============================] - 10s 681us/sample - loss: 0.0197 - select_loss: 0.1282 - predict_loss: 0.0184 - val_loss: 0.0182 - val_select_loss: 0.1267 - val_predict_loss: 0.0169\n",
      "Epoch 15/20\n",
      "14792/14792 [==============================] - 10s 690us/sample - loss: 0.0192 - select_loss: 0.1252 - predict_loss: 0.0179 - val_loss: 0.0202 - val_select_loss: 0.1240 - val_predict_loss: 0.0189\n",
      "Epoch 16/20\n",
      "14792/14792 [==============================] - 10s 673us/sample - loss: 0.0173 - select_loss: 0.1219 - predict_loss: 0.0160 - val_loss: 0.0162 - val_select_loss: 0.1204 - val_predict_loss: 0.0150\n",
      "Epoch 17/20\n",
      "14792/14792 [==============================] - 10s 674us/sample - loss: 0.0173 - select_loss: 0.1187 - predict_loss: 0.0161 - val_loss: 0.0147 - val_select_loss: 0.1152 - val_predict_loss: 0.0136\n",
      "Epoch 18/20\n",
      "14792/14792 [==============================] - 10s 673us/sample - loss: 0.0153 - select_loss: 0.1140 - predict_loss: 0.0142 - val_loss: 0.0162 - val_select_loss: 0.1105 - val_predict_loss: 0.0151\n",
      "Epoch 19/20\n",
      "14792/14792 [==============================] - 10s 676us/sample - loss: 0.0150 - select_loss: 0.1100 - predict_loss: 0.0140 - val_loss: 0.0157 - val_select_loss: 0.1065 - val_predict_loss: 0.0146\n",
      "Epoch 20/20\n",
      "14792/14792 [==============================] - 10s 682us/sample - loss: 0.0146 - select_loss: 0.1055 - predict_loss: 0.0135 - val_loss: 0.0136 - val_select_loss: 0.1035 - val_predict_loss: 0.0126\n",
      "Finish model interpretation\n"
     ]
    }
   ],
   "source": [
    "from interpretation import interpretation\n",
    "\n",
    "# Set interpretation model\n",
    "interpretation_model_name = 'tinvase'\n",
    "\n",
    "# Train interpretation model\n",
    "interpretor = interpretation(interpretation_model_name, model_parameters, pred_class, task)\n",
    "interpretor.fit(dataset_training)\n",
    "# Return instance-wise temporal and static feature importance\n",
    "test_s_hat = interpretor.predict(dataset_testing)\n",
    "\n",
    "print('Finish model interpretation')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 9: Visualize results\n",
    "\n",
    "(1) Visualize the performance of the trained model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finish predictor model evaluation.\n",
      "Overall performance\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from evaluation import Metrics\n",
    "from evaluation import print_performance\n",
    "\n",
    "# Evaluate predictor model\n",
    "result = Metrics(metric_sets, metric_parameters).evaluate(dataset_testing.label, test_y_hat)\n",
    "print('Finish predictor model evaluation.')\n",
    "\n",
    "print('Overall performance')\n",
    "print_performance(result, metric_sets, metric_parameters)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 9: Visualize results\n",
    "\n",
    "(2) Visualize the predictions of a certain patient by trained predictive model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Each prediction\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from evaluation import print_prediction\n",
    "\n",
    "# Set the patient index for visualization\n",
    "index = [1]\n",
    "\n",
    "print('Each prediction')\n",
    "print_prediction(test_y_hat[index], metric_parameters)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 9: Visualize results\n",
    "\n",
    "(3) Visualize the uncertainty of a certain patient by trained predictive and uncertainty model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uncertainty estimations\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from evaluation import print_uncertainty\n",
    "\n",
    "print('Uncertainty estimations')\n",
    "print_uncertainty (test_y_hat[index], test_ci_hat[index], metric_parameters)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 9: Visualize results\n",
    "\n",
    "(4) Visualize the interpretation of a certain patient prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model interpretation\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x1152 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from evaluation import print_interpretation\n",
    "\n",
    "print('Model interpretation')\n",
    "print_interpretation (test_s_hat[index], dataset_training.feature_name, metric_parameters, model_parameters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}