Chapter 21 - Saving and Loading Trained Models.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Chapter 21\n",
    "## Saving and Loading Trained Models\n",
    "\n",
    "### 21.0 Introduction\n",
    "In the last 20 chapters around 200 recipies, we have convered how to take raw data nad usem achine learning to create well-performing predictive models. However, for all our work to be worthwhile we eventually need to do something with our model, such as integrating it with an existing software application. To accomplish this goal, we need to be able to bot hsave our models after training and load them when they are needed by an application. This is the focus of the final chapter\n",
    "\n",
    "### 21.1 Saving and Loading a scikit-learn Model\n",
    "#### Problem\n",
    "You have trained a scikit-learn model and want to save it and load it elsewhere.\n",
    "\n",
    "#### Solution\n",
    "Save the model as a pickle file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/f00/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/sklearn/ensemble/weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n",
      "  from numpy.core.umath_tests import inner1d\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['model.pkl']"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load libraries\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn import datasets\n",
    "from sklearn.externals import joblib\n",
    "\n",
    "# load data\n",
    "iris = datasets.load_iris()\n",
    "features = iris.data\n",
    "target = iris.target\n",
    "\n",
    "# create decision tree classifier object\n",
    "classifier = RandomForestClassifier()\n",
    "\n",
    "# train model\n",
    "model = classifier.fit(features, target)\n",
    "\n",
    "# save model as pickle file\n",
    "joblib.dump(model, \"model.pkl\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once the model is saved we can use scikit-learn in our destination application (e.g., web application) to load the model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load model from file\n",
    "classifier = joblib.load(\"model.pkl\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And use it to make predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0])"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# create new observation\n",
    "new_observation = [[ 5.2, 3.2, 1.1, 0.1]]\n",
    "\n",
    "# predict obserrvation's class\n",
    "classifier.predict(new_observation)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Discussion\n",
    "The first step in using a model in production is to save that model as a file that can be loaded by another application or workflow. We can accomplish this by saving the model as a pickle file, a Python-specific data format. Specifically, to save the model we use `joblib`, which is a library extending pickle for cases when we have large NumPy arrays--a common occurance for trained models in scikit-learn.\n",
    "\n",
    "When saving scikit-learn models, be aware that saved models might not be compatible between versions of scikit-learn; therefore, it can be helpful to include the version of scikit-learn used in the model in the filename:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['model_(version).pkl']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# import library\n",
    "import sklearn\n",
    "\n",
    "# get scikit-learn version\n",
    "scikit_version = joblib.__version__\n",
    "\n",
    "# save model as pickle file\n",
    "joblib.dump(model, \"model_(version).pkl\".format(version=scikit_version))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 21.2 Saving and Loading a Keras Model\n",
    "#### Problem\n",
    "You have a trained Keras model and want to save it and load it elsewhere.\n",
    "\n",
    "#### Solution\n",
    "Save the model as HDF5:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using Theano backend.\n"
     ]
    },
    {
     "ename": "ModuleNotFoundError",
     "evalue": "No module named 'theano'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-1-ab685cfa2361>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# load libraries\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mnumpy\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdatasets\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mimdb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpreprocessing\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtext\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mTokenizer\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mmodels\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/keras/__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0m__future__\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mabsolute_import\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mutils\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mactivations\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mapplications\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/keras/utils/__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mdata_utils\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mio_utils\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 6\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mconv_utils\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      7\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;31m# Globally-importable utils.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/keras/utils/conv_utils.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msix\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmoves\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mnumpy\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 9\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m\u001b[0;34m.\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mbackend\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mK\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     10\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/keras/backend/__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     84\u001b[0m \u001b[0;32melif\u001b[0m \u001b[0m_BACKEND\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'theano'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     85\u001b[0m     \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstderr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwrite\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'Using Theano backend.\\n'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 86\u001b[0;31m     \u001b[0;32mfrom\u001b[0m \u001b[0;34m.\u001b[0m\u001b[0mtheano_backend\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     87\u001b[0m \u001b[0;32melif\u001b[0m \u001b[0m_BACKEND\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'tensorflow'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     88\u001b[0m     \u001b[0msys\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstderr\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwrite\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'Using TensorFlow backend.\\n'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda/envs/machine_learning_cookbook/lib/python3.6/site-packages/keras/backend/theano_backend.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mcollections\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mdefaultdict\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mcontextlib\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mcontextmanager\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m \u001b[0;32mimport\u001b[0m \u001b[0mtheano\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtheano\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mtensor\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mT\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtheano\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msandbox\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mrng_mrg\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mMRG_RandomStreams\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mRandomStreams\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'theano'"
     ]
    }
   ],
   "source": [
    "# load libraries\n",
    "import numpy as np\n",
    "from keras.datasets import imdb\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras import models\n",
    "from keras import layers\n",
    "from keras.models import load_model\n",
    "\n",
    "# set random seed\n",
    "np.random.seed(0)\n",
    "\n",
    "# set the number of features we want\n",
    "number_of_features = 1000\n",
    "\n",
    "# load data and target vector from movie review data\n",
    "(train_Data, train_target), (test_data, test_target) = imdb.load_data(num_words=number_of_features)\n",
    "\n",
    "# convert movie review data to a one-hot encoded feature matrix\n",
    "tokenizer = Tokenizer(num_words=number_of_features)\n",
    "train_features = tokenizer.sequences_to_matrix(train_data, mode=\"binary\")\n",
    "test_features = tokenizer.sequences_to_matrix(test_data, mode=\"binary\")\n",
    "\n",
    "# start neural network\n",
    "network = models.Sequential()\n",
    "\n",
    "# add fully connected layer with ReLU activation function\n",
    "network.add(layers.Dense(units=16, activation=\"relu\", input_shape=(number_of_features,)))\n",
    "\n",
    "# add fully connected layer with a sigmoid activation function\n",
    "network.add(layers.Dense(units=1, activation=\"sigmoid\"))\n",
    "\n",
    "# compile neural network\n",
    "network.compile(loss=\"binary_crossentropy\", optimizer=\"rmsprop\", metrics=[\"accuracy\"])\n",
    "\n",
    "# train neural network\n",
    "history = network.fit(train_features, train_target, epochs=3, verbose=0, batch_size=100, validation_data=(test_features, test_target))\n",
    "\n",
    "# save neural network\n",
    "network.save(\"model.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can then load the model either in another application or for additional training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'load_model' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-7-114bd8857354>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# load neural network\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mnetwork\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mload_model\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"model.h5\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'load_model' is not defined"
     ]
    }
   ],
   "source": [
    "# load neural network\n",
    "network = load_model(\"model.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Discussion\n",
    "Unlike scikit-learn, Keras does not recommend you save models using pickle. Instead, models are saved as an HDF5 file. The HDF5 file contains everything you need to not only load the model to make predicitons (i.e., achitecture and trained parameters), but also to restart training (i.e. loss and optimizer settings and the current state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:machine_learning_cookbook]",
   "language": "python",
   "name": "conda-env-machine_learning_cookbook-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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}