{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Quick start"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:20:09.046564Z",
     "start_time": "2018-08-26T00:20:02.540948Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n",
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n",
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n",
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n",
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n",
      "/Users/demaasit/anaconda3/envs/tfp/lib/python3.6/importlib/_bootstrap.py:219: ImportWarning: can't resolve package from __spec__ or __package__, falling back on __name__ and __path__\n",
      "  return f(*args, **kwds)\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "from warnings import filterwarnings\n",
    "filterwarnings(\"ignore\")\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow_probability as tfp\n",
    "tf.enable_eager_execution()\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "np.random.seed(12345)\n",
    "rc = {'xtick.labelsize': 20, 'ytick.labelsize': 20, 'axes.labelsize': 20, 'font.size': 20, \n",
    "      'legend.fontsize': 12.0, 'axes.titlesize': 10, \"figure.figsize\": [12, 6]}\n",
    "sns.set(rc = rc)\n",
    "from IPython.core.interactiveshell import InteractiveShell\n",
    "InteractiveShell.ast_node_interactivity = \"all\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:20:09.052630Z",
     "start_time": "2018-08-26T00:20:09.048753Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.10.1\n"
     ]
    }
   ],
   "source": [
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 1: Prepare the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:06:45.220611Z",
     "start_time": "2018-08-26T00:06:45.198841Z"
    }
   },
   "outputs": [],
   "source": [
    "# Pretend to load synthetic data set.\n",
    "features = tfp.distributions.Normal(loc=0., scale=1.).sample(int(100e3))\n",
    "labels = tfp.distributions.Bernoulli(logits=1.618 * features).sample()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: Build the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:06:54.759750Z",
     "start_time": "2018-08-26T00:06:54.756997Z"
    }
   },
   "outputs": [],
   "source": [
    "# Specify model.\n",
    "model = tfp.glm.Bernoulli()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 3: Perform Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:07:00.110726Z",
     "start_time": "2018-08-26T00:07:00.004677Z"
    }
   },
   "outputs": [],
   "source": [
    "# Fit model given data.\n",
    "coeffs, linear_response, is_converged, num_iter = tfp.glm.fit(\n",
    "    model_matrix=features[:, tf.newaxis],\n",
    "    response=tf.to_float(labels),\n",
    "    model=model)\n",
    "# ==> coeffs is approximately [1.618] (We're golden!)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-26T00:07:15.140038Z",
     "start_time": "2018-08-26T00:07:15.134457Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor([1.618256], shape=(1,), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "print(coeffs)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:tfp]",
   "language": "python",
   "name": "conda-env-tfp-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}