Merge branch 'master' of https://github.com/nfmcclure/tensorflow_cookbook

Author: nfmcclure

04_Support_Vector_Machines/02_Working_with_Linear_SVMs/02_linear_svm.ipynb

@@ -27,8 +27,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
-   "metadata": {},
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "import matplotlib.pyplot as plt\n",
@@ -48,8 +50,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 21,
-   "metadata": {},
+   "execution_count": 2,
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "np.random.seed(42)\n",
@@ -66,7 +70,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -85,8 +89,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 23,
-   "metadata": {},
+   "execution_count": 6,
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "train_indices = np.random.choice(len(x_vals),\n",
@@ -108,8 +114,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 24,
-   "metadata": {},
+   "execution_count": 7,
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Declare batch size\n",
@@ -132,15 +140,17 @@
     "\n",
     "SVM linear model is given by the equation:\n",
     "\n",
-    "$$\\left[ \\frac{1}{n} \\sum_{i=1}^{n} \\max(0, 1 - A \\cdot x - b) \\right] + \\alpha \\cdot ||A||^{2}$$\n",
+    "$$\\left[ \\frac{1}{n} \\sum_{i=1}^{n} \\max(0, 1 - y_i(A \\cdot x + b)) \\right] + \\alpha \\cdot ||A||^{2}$$\n",
     "\n",
     "Our loss function will be the above quantity and we will tell TensorFlow to minimize it. Note that $n$ is the number of points (in a batch), $A$ is the hyperplane-normal vector (to solve for), $b$ is the hyperplane-offset (to solve for), and $\\alpha$ is the soft-margin parameter."
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 25,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Declare model operations\n",
@@ -162,7 +172,9 @@
   {
    "cell_type": "code",
    "execution_count": 26,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Declare loss function\n",
@@ -188,7 +200,9 @@
   {
    "cell_type": "code",
    "execution_count": 27,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Declare prediction function\n",
@@ -281,7 +295,9 @@
   {
    "cell_type": "code",
    "execution_count": 29,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "# Extract coefficients\n",
@@ -381,7 +397,9 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": []
   }
@@ -402,7 +420,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.1"
+   "version": "3.6.5"
   }
  },
  "nbformat": 4,

04_Support_Vector_Machines/02_Working_with_Linear_SVMs/02_linear_svm.py

@@ -35,7 +35,7 @@
 
 # Split data into train/test sets
 train_indices = np.random.choice(len(x_vals),
-                                 round(len(x_vals)*0.9),
+                                 int(round(len(x_vals)*0.9)),
                                  replace=False)
 test_indices = np.array(list(set(range(len(x_vals))) - set(train_indices)))
 x_vals_train = x_vals[train_indices]

04_Support_Vector_Machines/03_Reduction_to_Linear_Regression/03_support_vector_regression.py

@@ -87,7 +87,7 @@
 # Extract Coefficients
 [[slope]] = sess.run(A)
 [[y_intercept]] = sess.run(b)
-[width] = sess.run(epsilon)
+width = sess.run(epsilon)
 
 # Get best fit line
 best_fit = []