Introduction to Neural Networks Part 1 - Workbook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Introduction to Neural Networks\n",
    "## Part 1: Perceptrons\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Why Neural Networks? Why Machine Learning?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Recent developments in neural networks have \"created\" these art works.\n",
    "![A Neural Algorithm of Artistic Style, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge](images/artistic_neural_networks.jpg)\n",
    "taken from a paper on [A Neural Algorithm of Artistic Style by Leon A. Gatys, Alexander S. Ecker, Matthias Bethge](http://arxiv.org/pdf/1508.06576.pdf)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Consider a single neuron cell"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![A Neuron](https://upload.wikimedia.org/wikipedia/commons/b/bd/Neuron.jpg)\n",
    "Source: [Wikipedia](https://upload.wikimedia.org/wikipedia/commons/b/bd/Neuron.jpg)\n",
    "\n",
    "If the stimulus to a neuron cell is above a \"threshold\", the cell is activated and will release a tiny electrical impulse to the next cells which it is connected with."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### How could this be represented mathematically?\n",
    "![Structure of a single perceptron](https://upload.wikimedia.org/wikipedia/commons/8/8c/Perceptron_moj.png)\n",
    "Source: [Wikipedia](https://upload.wikimedia.org/wikipedia/commons/8/8c/Perceptron_moj.png)\n",
    "\n",
    "$$ f(x) = \\begin{cases}1 & \\text{if }\\vec{x} \\cdot \\vec{w} + w_0 > 0\\\\0 & \\text{otherwise}\\end{cases} $$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Mathematical Representation of a Single Layer Perceptron\n",
    "\n",
    "$$ w = \\begin{bmatrix}\n",
    "         w_1 \\\\\n",
    "         \\vdots \\\\\n",
    "         w_n\n",
    "        \\end{bmatrix} \n",
    "   x = \\begin{bmatrix}\n",
    "         x_1 \\\\\n",
    "         \\vdots \\\\\n",
    "         x_n\n",
    "        \\end{bmatrix}\n",
    "$$\n",
    "\n",
    "#### Linear Activation Function\n",
    "\n",
    "\\begin{equation*}\n",
    "z = x^T * w + w_0\n",
    "\\end{equation*}\n",
    "\n",
    "\\begin{equation*}\n",
    "\\phi(z) = \\begin{cases}\n",
    "1\\ if\\ z > 0 \\\\\n",
    "0\\ otherwise\n",
    "\\end{cases}\n",
    "\\end{equation*}\n",
    "\n",
    "\n",
    "#### How does the single layer perceptron learn?\n",
    "\n",
    "1. Initialize the weights with zero values\n",
    "2. For each training sample, compute the predicted $y$ and update the weights accordingly\n",
    "\n",
    "\\begin{equation*} \n",
    "\\Delta w = \\eta(y_{training set}^{(i)} - y_{predicted}^{(i)}) x^{(i)} \\\\\n",
    "w = w + \\Delta w \\\\\n",
    "\\end{equation*}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/hannes/.virtualenvs/PDX-data-science-intro-perceptron/lib/python2.7/site-packages/matplotlib/font_manager.py:273: UserWarning: Matplotlib is building the font cache using fc-list. This may take a moment.\n",
      "  warnings.warn('Matplotlib is building the font cache using fc-list. This may take a moment.')\n"
     ]
    }
   ],
   "source": [
    "# install all required packages with 'pip install -r requirements.txt'\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class Perceptron(object):\n",
    "    \"\"\"\n",
    "    \n",
    "    Perceptron classifier.\n",
    "\n",
    "    Adopted from Python Machine Learning by Sebastian Raschka (see below)\n",
    "\n",
    "    \"\"\"\n",
    "    def __init__(self, learning_rate=0.01, epoches=10):\n",
    "        self.learning_rate = learning_rate\n",
    "        self.epoches = epoches\n",
    "        \n",
    "    def fit(self, X, y):\n",
    "        \"\"\"This method will fit the perceptron activation function to match the training set\"\"\"\n",
    "        \n",
    "        # set self.weights to a zero value\n",
    "        # check out a numpy method, get the dimension of the input vector X \n",
    "        # with `X.shape[1]`\n",
    "        \n",
    "        ### TASK: create an numpy zero array with with the dimension x, 1\n",
    "        self.weights = ...\n",
    "        self.errors = []\n",
    "\n",
    "        for _ in range(self.epoches):\n",
    "            errors = 0\n",
    "            for xi, expected_y in zip(X, y):\n",
    "                ### TASK: implement the update rule (expected - predicted). \n",
    "                ### Don't forget to apply the learning rate\n",
    "                update = ...\n",
    "                self.weights[1:] += ...\n",
    "                self.weights[0] += ...\n",
    "                errors += int(update != 0.0)\n",
    "            self.errors.append(errors)\n",
    "        return self    \n",
    "    \n",
    "    def net_input(self, X):\n",
    "        \"\"\"This method will calculate the dot product of the input vector x multiplied with the vector of the weights.\n",
    "           Don't forget the bias w_0\n",
    "        \"\"\"\n",
    "        \n",
    "        ### TASK: Calculate this sum of the multipled vectors x and w. You can either do this by using the numpy.dot\n",
    "        ### method or by using a for loop (the numpy.dot method will be faster but less intuitive)\n",
    "        _sum = ...\n",
    "        return _sum\n",
    "\n",
    "    def predict(self, X):\n",
    "        \"\"\"Return class label after unit step\"\"\"\n",
    "        ### TASK: Take a result from self.net_input(X) and classify it. \n",
    "        ### Return 1 if the result is greater than 0, other return 1\n",
    "        y_predicted = ...\n",
    "        return y_predicted\n",
    "      \n",
    "    def feature_scaling(self, X):\n",
    "        \"\"\" Scales features between 0 and 1 \"\"\"\n",
    "        # perform feature scaling on each column\n",
    "        for col in list(X.columns.values):\n",
    "            X[col] = X[col] / X[col].max()\n",
    "        return X\n",
    "        \n",
    "    def plot(self, X, y, \n",
    "             cols=[0, 1], \n",
    "             axis_labels=['x', 'y'], \n",
    "             _max_samples=100,\n",
    "             save=False\n",
    "            ):\n",
    "        max_samples = _max_samples if _max_samples < len(y) else len(y)\n",
    "        X = X[0:max_samples, cols]\n",
    "        for i in range(0, max_samples):\n",
    "            if int(y[i]) == 1:\n",
    "                color='red'\n",
    "                marker='o'\n",
    "            else:\n",
    "                color='blue'\n",
    "                marker='x'\n",
    "            plt.scatter(\n",
    "                X[i][cols[0]], X[i][cols[1]],\n",
    "                color=color, marker=marker, )\n",
    "\n",
    "        plt.xlabel(axis_labels[0])\n",
    "        plt.ylabel(axis_labels[1])\n",
    "        plt.show()\n",
    " \n",
    "    def plot_errors(self):\n",
    "        plt.plot(range(1, len(p.errors) + 1), p.errors, marker='o')\n",
    "        plt.xlabel('Epochs')\n",
    "        plt.ylabel('Number of misclassifications')\n",
    "        plt.show()\n",
    "\n",
    "        \n",
    "    def cross_val_score(self, X, y):\n",
    "        \"\"\"Returns the CV matrix for a model and its expected results\"\"\"\n",
    "        # set variables\n",
    "        _pred_y = self.predict(X)\n",
    "        cov_results = numpy.zeros((2,2))\n",
    "        # Check for positives and negatives\n",
    "        positions = (\n",
    "            list(numpy.where(y != 1))[0],\n",
    "            list(numpy.where(y == 1))[0]\n",
    "        )\n",
    "        for res in [0, 1]:\n",
    "            # get total number of pos/neg samples\n",
    "            total = len(positions[res])\n",
    "            # loop through sample positions\n",
    "            for pos in list(positions[res]):\n",
    "                # check postives\n",
    "                if y[pos] == 1:\n",
    "                    if _pred_y[pos] == 1:\n",
    "                        cov_results[1, 1] += 1\n",
    "                    else:\n",
    "                        cov_results[1, 0] += 1\n",
    "                # check not positives (could be -1 or 0)\n",
    "                else:\n",
    "                    if _pred_y[pos] != 1:\n",
    "                        cov_results[0, 0] += 1\n",
    "                    else:\n",
    "                        cov_results[0, 1] += 1\n",
    "            # divide results by the number of samples in the category (positive/negative)\n",
    "            cov_results[res, :] /= total\n",
    "        return cov_results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OR Examples, solved by a single Perceptron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Prediction: ', array([0, 1, 1, 1]))\n",
      "Cross validation: \n",
      "[[ 1.  0.]\n",
      " [ 0.  1.]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEPCAYAAABsj5JaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x105be8c50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEPCAYAAABGP2P1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x105bdd0d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# OR test\n",
    "\n",
    "data_set = numpy.array([\n",
    "    [0, 0, 0],\n",
    "    [0, 1, 1],\n",
    "    [1, 0, 1],\n",
    "    [1, 1, 1]\n",
    "])\n",
    "\n",
    "p = Perceptron(0.01, 10)\n",
    "p.fit(data_set[:, :2], data_set[:, 2:])\n",
    "print ('Prediction: ', p.predict(data_set[:, :2]))\n",
    "print ('Cross validation: ')\n",
    "print p.cross_val_score(data_set[:, :2], data_set[:, 2:])\n",
    "p.plot(data_set[:, :2], data_set[:, 2])\n",
    "p.predict(data_set[:, :2])\n",
    "p.plot_errors()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Iris dataset\n",
    "Determine the type of Irises (in this case, we classify between setosa and versicolor). The classic data originates from the [UCI dataset server](http://archive.ics.uci.edu/ml/) and is one of the most downloaded sets so far. An overview of the different input variables can be found [here](https://en.wikipedia.org/wiki/Iris_flower_data_set#/media/File:Iris_dataset_scatterplot.svg).\n",
    "For this tutorial, let's classify the Irises by the sepal and petal length."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.6</td>\n",
       "      <td>3.1</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>Iris-setosa</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     0    1    2    3            4\n",
       "0  5.1  3.5  1.4  0.2  Iris-setosa\n",
       "1  4.9  3.0  1.4  0.2  Iris-setosa\n",
       "2  4.7  3.2  1.3  0.2  Iris-setosa\n",
       "3  4.6  3.1  1.5  0.2  Iris-setosa\n",
       "4  5.0  3.6  1.4  0.2  Iris-setosa"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('https://archive.ics.uci.edu/ml/'\n",
    "        'machine-learning-databases/iris/iris.data', header=None)\n",
    "df.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy\n",
    "\n",
    "# select setosa and versicolor\n",
    "y = df.iloc[0:100, 4].values\n",
    "y = numpy.where(y == 'Iris-setosa', 0, 1)\n",
    "\n",
    "# extract sepal length and petal length\n",
    "X = df.iloc[0:100, [0, 2]].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAEPCAYAAABFpK+YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x105e10d50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEPCAYAAABGP2P1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x105e10290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "p = Perceptron(learning_rate=0.1, epoches=10)\n",
    "\n",
    "p.fit(X, y)\n",
    "p.plot(X, y)\n",
    "p.plot_errors()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.,  0.],\n",
       "       [ 0.,  1.]])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p.cross_val_score(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      " 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      " 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n",
      "[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
      " 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
      " 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]\n"
     ]
    }
   ],
   "source": [
    "print y\n",
    "print p.predict(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  \n",
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Credit card fraud dectection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# https://onlinecourses.science.psu.edu/stat857/node/215\n",
    "import pandas as pd\n",
    "\n",
    "training_set = pd.read_csv('german_credit_dataset/Training50.csv')\n",
    "test_set = pd.read_csv('german_credit_dataset/Test50.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index([u'Creditability', u'Account.Balance', u'Duration.of.Credit..month.',\n",
      "       u'Payment.Status.of.Previous.Credit', u'Purpose', u'Credit.Amount',\n",
      "       u'Value.Savings.Stocks', u'Length.of.current.employment',\n",
      "       u'Instalment.per.cent', u'Sex...Marital.Status', u'Guarantors',\n",
      "       u'Duration.in.Current.address', u'Most.valuable.available.asset',\n",
      "       u'Age..years.', u'Concurrent.Credits', u'Type.of.apartment',\n",
      "       u'No.of.Credits.at.this.Bank', u'Occupation', u'No.of.dependents',\n",
      "       u'Telephone', u'Foreign.Worker'],\n",
      "      dtype='object')\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Creditability</th>\n",
       "      <th>Account.Balance</th>\n",
       "      <th>Duration.of.Credit..month.</th>\n",
       "      <th>Payment.Status.of.Previous.Credit</th>\n",
       "      <th>Purpose</th>\n",
       "      <th>Credit.Amount</th>\n",
       "      <th>Value.Savings.Stocks</th>\n",
       "      <th>Length.of.current.employment</th>\n",
       "      <th>Instalment.per.cent</th>\n",
       "      <th>Sex...Marital.Status</th>\n",
       "      <th>...</th>\n",
       "      <th>Duration.in.Current.address</th>\n",
       "      <th>Most.valuable.available.asset</th>\n",
       "      <th>Age..years.</th>\n",
       "      <th>Concurrent.Credits</th>\n",
       "      <th>Type.of.apartment</th>\n",
       "      <th>No.of.Credits.at.this.Bank</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>No.of.dependents</th>\n",
       "      <th>Telephone</th>\n",
       "      <th>Foreign.Worker</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2108</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>29</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>950</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>33</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>42</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>7174</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>30</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>36</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>7980</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>27</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>2028</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>30</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Creditability  Account.Balance  Duration.of.Credit..month.  \\\n",
       "0              1                3                           6   \n",
       "1              0                1                          15   \n",
       "2              0                1                          42   \n",
       "3              0                3                          36   \n",
       "4              1                3                          24   \n",
       "\n",
       "   Payment.Status.of.Previous.Credit  Purpose  Credit.Amount  \\\n",
       "0                                  2        3           2108   \n",
       "1                                  1        4            950   \n",
       "2                                  2        3           7174   \n",
       "3                                  3        4           7980   \n",
       "4                                  3        2           2028   \n",
       "\n",
       "   Value.Savings.Stocks  Length.of.current.employment  Instalment.per.cent  \\\n",
       "0                     1                             3                    2   \n",
       "1                     1                             4                    4   \n",
       "2                     4                             3                    4   \n",
       "3                     4                             1                    4   \n",
       "4                     1                             3                    2   \n",
       "\n",
       "   Sex...Marital.Status       ...        Duration.in.Current.address  \\\n",
       "0                     3       ...                                  2   \n",
       "1                     2       ...                                  3   \n",
       "2                     1       ...                                  3   \n",
       "3                     2       ...                                  4   \n",
       "4                     2       ...                                  2   \n",
       "\n",
       "   Most.valuable.available.asset  Age..years.  Concurrent.Credits  \\\n",
       "0                              1           29                   2   \n",
       "1                              3           33                   2   \n",
       "2                              3           30                   2   \n",
       "3                              3           27                   2   \n",
       "4                              2           30                   2   \n",
       "\n",
       "   Type.of.apartment  No.of.Credits.at.this.Bank  Occupation  \\\n",
       "0                  1                           1           1   \n",
       "1                  1                           2           1   \n",
       "2                  2                           1           1   \n",
       "3                  1                           2           1   \n",
       "4                  2                           2           1   \n",
       "\n",
       "   No.of.dependents  Telephone  Foreign.Worker  \n",
       "0                 1          1               1  \n",
       "1                 2          1               1  \n",
       "2                 1          2               1  \n",
       "3                 1          2               1  \n",
       "4                 1          1               1  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# drop id column from training/test set\n",
    "training_set.drop('ID', axis=1, inplace=True)\n",
    "\n",
    "# inspect dataset\n",
    "print training_set.columns\n",
    "# training_set.head()\n",
    "training_set.head()  # training_set.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Creditability</th>\n",
       "      <th>Account.Balance</th>\n",
       "      <th>Duration.of.Credit..month.</th>\n",
       "      <th>Payment.Status.of.Previous.Credit</th>\n",
       "      <th>Purpose</th>\n",
       "      <th>Credit.Amount</th>\n",
       "      <th>Value.Savings.Stocks</th>\n",
       "      <th>Length.of.current.employment</th>\n",
       "      <th>Instalment.per.cent</th>\n",
       "      <th>Sex...Marital.Status</th>\n",
       "      <th>...</th>\n",
       "      <th>Duration.in.Current.address</th>\n",
       "      <th>Most.valuable.available.asset</th>\n",
       "      <th>Age..years.</th>\n",
       "      <th>Concurrent.Credits</th>\n",
       "      <th>Type.of.apartment</th>\n",
       "      <th>No.of.Credits.at.this.Bank</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>No.of.dependents</th>\n",
       "      <th>Telephone</th>\n",
       "      <th>Foreign.Worker</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.10</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.114416</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.386667</td>\n",
       "      <td>1</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.051563</td>\n",
       "      <td>0.25</td>\n",
       "      <td>1.00</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>...</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.440000</td>\n",
       "      <td>1</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.70</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.389383</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.75</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>...</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>1</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.60</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.433131</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.25</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>...</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.360000</td>\n",
       "      <td>1</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.40</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.110074</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>...</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.50</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>1</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Creditability  Account.Balance  Duration.of.Credit..month.  \\\n",
       "0              1         1.000000                        0.10   \n",
       "1              0         0.333333                        0.25   \n",
       "2              0         0.333333                        0.70   \n",
       "3              0         1.000000                        0.60   \n",
       "4              1         1.000000                        0.40   \n",
       "\n",
       "   Payment.Status.of.Previous.Credit  Purpose  Credit.Amount  \\\n",
       "0                           0.666667     0.75       0.114416   \n",
       "1                           0.333333     1.00       0.051563   \n",
       "2                           0.666667     0.75       0.389383   \n",
       "3                           1.000000     1.00       0.433131   \n",
       "4                           1.000000     0.50       0.110074   \n",
       "\n",
       "   Value.Savings.Stocks  Length.of.current.employment  Instalment.per.cent  \\\n",
       "0                  0.25                          0.75                  0.5   \n",
       "1                  0.25                          1.00                  1.0   \n",
       "2                  1.00                          0.75                  1.0   \n",
       "3                  1.00                          0.25                  1.0   \n",
       "4                  0.25                          0.75                  0.5   \n",
       "\n",
       "   Sex...Marital.Status       ...        Duration.in.Current.address  \\\n",
       "0              1.000000       ...                               0.50   \n",
       "1              0.666667       ...                               0.75   \n",
       "2              0.333333       ...                               0.75   \n",
       "3              0.666667       ...                               1.00   \n",
       "4              0.666667       ...                               0.50   \n",
       "\n",
       "   Most.valuable.available.asset  Age..years.  Concurrent.Credits  \\\n",
       "0                           0.25     0.386667                   1   \n",
       "1                           0.75     0.440000                   1   \n",
       "2                           0.75     0.400000                   1   \n",
       "3                           0.75     0.360000                   1   \n",
       "4                           0.50     0.400000                   1   \n",
       "\n",
       "   Type.of.apartment  No.of.Credits.at.this.Bank  Occupation  \\\n",
       "0           0.333333                         0.5           1   \n",
       "1           0.333333                         1.0           1   \n",
       "2           0.666667                         0.5           1   \n",
       "3           0.333333                         1.0           1   \n",
       "4           0.666667                         1.0           1   \n",
       "\n",
       "   No.of.dependents  Telephone  Foreign.Worker  \n",
       "0               0.5        0.5             0.5  \n",
       "1               1.0        0.5             0.5  \n",
       "2               0.5        1.0             0.5  \n",
       "3               0.5        1.0             0.5  \n",
       "4               0.5        0.5             0.5  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# perform feature scaling on each column\n",
    "for col in list(training_set.columns.values):\n",
    "    training_set[col] = training_set[col] / training_set[col].max()\n",
    "training_set.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# extract the creditability column as y vector\n",
    "y = training_set['Creditability'].values\n",
    "# drop the creditability column from the dataset\n",
    "training_set.drop('Creditability', axis=1, inplace=True)\n",
    "# remaining dataset is used as input matrix\n",
    "X = training_set.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 0])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p = Perceptron(0.0001, 1000)\n",
    "p.fit(X, y)\n",
    "p.predict(X[:3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.47552448  0.52447552]\n",
      " [ 0.10084034  0.89915966]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEKCAYAAAAb7IIBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJKlQBoAkFer/AVTq+6soT6P+AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10768b210>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEPCAYAAABsj5JaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzGTLTFdpAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1073577d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print p.cross_val_score(X, y)\n",
    "p.plot(training_set[['Account.Balance', 'Purpose']].values, y)\n",
    "p.plot_errors()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "source": [
    "### Try the XOR Problem"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Prediction: ', array([1, 1, 0, 0]))\n",
      "Cross validation: \n",
      "[[ 0.5  0.5]\n",
      " [ 0.5  0.5]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAAEPCAYAAABsj5JaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIJKkJhaIJKmJBSJJavL/AElbqCfDczyQAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x106e95610>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# XOR problem\n",
    "\n",
    "data_set = numpy.array([\n",
    "    [0, 0, 0],\n",
    "    [0, 1, 1],\n",
    "    [1, 0, 1],\n",
    "    [1, 1, 0]\n",
    "])\n",
    "\n",
    "p = Perceptron(0.01, 10)\n",
    "p.fit(data_set[:, :2], data_set[:, 2])\n",
    "print ('Prediction: ', p.predict(data_set[:, :2]))\n",
    "print ('Cross validation: ')\n",
    "print p.cross_val_score(data_set[:, :2], data_set[:, 2])\n",
    "p.plot(data_set[:, :2], data_set[:, 2], axis_labels=['x1', 'x2'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Prediction: ', array([0, 1, 1, 0]))\n",
      "Cross validation: \n",
      "[[ 1.  0.]\n",
      " [ 0.  1.]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAEPCAYAAABGP2P1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10765cb50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "array(0)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# XOR problem solved with a single layer perceptron\n",
    "\n",
    "data_set = numpy.array([\n",
    "    # index, (x1+x2)^2, output\n",
    "    [0, 0, 0],\n",
    "    [1, 1, 1],\n",
    "    [2, 1, 1],\n",
    "    [3, 4, 0]\n",
    "])\n",
    "\n",
    "p = Perceptron(0.001, 1000)\n",
    "p.fit(data_set[:, :2], data_set[:, 2])\n",
    "print ('Prediction: ', p.predict(data_set[:, :2]))\n",
    "print ('Cross validation: ')\n",
    "print p.cross_val_score(data_set[:, :2], data_set[:, 2])\n",
    "p.plot(data_set[:, [0,1]], data_set[:, 2], axis_labels=['x1', 'x2'])\n",
    "p.predict([3,4])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Further Readings\n",
    "\n",
    "### General Introduction\n",
    "![The Master Algorithm](http://t2.gstatic.com/images?q=tbn:ANd9GcTvctDHIzSuyffTPDOL22iOH_p9JJRTUBNy6eT8qKsC9Z6Gx7pp)\n",
    "\n",
    "### Theoretical Background\n",
    "![Rojas](http://dcis.inf.fu-berlin.de/rojas/wp-content/uploads/2015/07/raul_rojas-512x372.jpg)\n",
    "[Neural Networks: A Systematic Introduction by Rojas, Free book](https://page.mi.fu-berlin.de/rojas/neural/neuron.pdf)\n",
    "\n",
    "![Neural Networks & Learning Machines 3rd Edition by Simon Haykin](http://ecx.images-amazon.com/images/I/51b1PWBu6uL._SY344_BO1,204,203,200_.jpg)\n",
    "[Neural Networks & Learning Machines 3rd Edition by Simon Haykin](http://www.powells.com/book/neural-networks-learning-machines-3rd-edition-9780131471399/72-0)\n",
    "### Machine Learning and Python\n",
    "![Machine Learning: An Algorithmic Perspective by Stephen Marsland](http://ecx.images-amazon.com/images/I/41xC5yM5FLL._SY344_BO1,204,203,200_.jpg)\n",
    "[Machine Learning: An Algorithmic Perspective by Stephen Marsland](http://www.powells.com/book/machine-learning-9781420067187/1-1)\n",
    "![Python Machine Learning by Sebastian Raschka](http://ecx.images-amazon.com/images/I/51KGnFRB-zL._SX258_BO1,204,203,200_.jpg)\n",
    "[Python Machine Learning by Sebastian Raschka](http://www.powells.com/book/python-machine-learning-9781783555130/61-0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}