diff --git a/08_dimensionality_reduction.ipynb b/08_dimensionality_reduction.ipynb
index da3ebc1..e26025e 100644
--- a/08_dimensionality_reduction.ipynb
+++ b/08_dimensionality_reduction.ipynb
@@ -2,40 +2,28 @@
  "cells": [
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "**Chapter 8 – Dimensionality Reduction**"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "_This notebook contains all the sample code and solutions to the exercices in chapter 8._"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# Setup"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "First, let's make sure this notebook works well in both python 2 and 3, import a few common modules, ensure MatplotLib plots figures inline and prepare a function to save the figures:"
    ]
@@ -44,9 +32,7 @@
    "cell_type": "code",
    "execution_count": 1,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -82,10 +68,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# Projection methods\n",
     "Build 3D dataset:"
@@ -94,11 +77,7 @@
   {
    "cell_type": "code",
    "execution_count": 2,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.random.seed(4)\n",
@@ -115,10 +94,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "## PCA using SVD decomposition"
    ]
@@ -127,9 +103,7 @@
    "cell_type": "code",
    "execution_count": 3,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -142,11 +116,7 @@
   {
    "cell_type": "code",
    "execution_count": 4,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "m, n = X.shape\n",
@@ -158,11 +128,7 @@
   {
    "cell_type": "code",
    "execution_count": 5,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(X_centered, U.dot(S).dot(V))"
@@ -171,11 +137,7 @@
   {
    "cell_type": "code",
    "execution_count": 6,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "W2 = V.T[:, :2]\n",
@@ -186,9 +148,7 @@
    "cell_type": "code",
    "execution_count": 7,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -197,20 +157,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "## PCA using Scikit-Learn"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "With Scikit-Learn, PCA is really trivial. It even takes care of mean centering for you:"
    ]
@@ -218,11 +172,7 @@
   {
    "cell_type": "code",
    "execution_count": 8,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.decomposition import PCA\n",
@@ -234,11 +184,7 @@
   {
    "cell_type": "code",
    "execution_count": 9,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "X2D[:5]"
@@ -247,11 +193,7 @@
   {
    "cell_type": "code",
    "execution_count": 10,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "X2D_using_svd[:5]"
@@ -260,9 +202,7 @@
   {
    "cell_type": "markdown",
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "source": [
     "Notice that running PCA multiple times on slightly different datasets may result in different results. In general the only difference is that some axes may be flipped. In this example, PCA using Scikit-Learn gives the same projection as the one given by the SVD approach, except both axes are flipped:"
@@ -271,11 +211,7 @@
   {
    "cell_type": "code",
    "execution_count": 11,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(X2D, -X2D_using_svd)"
@@ -283,10 +219,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Recover the 3D points projected on the plane (PCA 2D subspace)."
    ]
@@ -295,9 +228,7 @@
    "cell_type": "code",
    "execution_count": 12,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -306,10 +237,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Of course, there was some loss of information during the projection step, so the recovered 3D points are not exactly equal to the original 3D points:"
    ]
@@ -317,11 +245,7 @@
   {
    "cell_type": "code",
    "execution_count": 13,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(X3D_inv, X)"
@@ -329,10 +253,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "We can compute the reconstruction error:"
    ]
@@ -340,11 +261,7 @@
   {
    "cell_type": "code",
    "execution_count": 14,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.mean(np.sum(np.square(X3D_inv - X), axis=1))"
@@ -352,10 +269,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "The inverse transform in the SVD approach looks like this:"
    ]
@@ -363,11 +277,7 @@
   {
    "cell_type": "code",
    "execution_count": 15,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "X3D_inv_using_svd = X2D_using_svd.dot(V[:2, :])"
@@ -375,10 +285,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "The reconstructions from both methods are not identical because Scikit-Learn's `PCA` class automatically takes care of reversing the mean centering, but if we subtract the mean, we get the same reconstruction:"
    ]
@@ -386,11 +293,7 @@
   {
    "cell_type": "code",
    "execution_count": 16,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(X3D_inv_using_svd, X3D_inv - pca.mean_)"
@@ -398,10 +301,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "The `PCA` object gives access to the principal components that it computed:"
    ]
@@ -409,11 +309,7 @@
   {
    "cell_type": "code",
    "execution_count": 17,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca.components_"
@@ -421,10 +317,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Compare to the first two principal components computed using the SVD method:"
    ]
@@ -432,11 +325,7 @@
   {
    "cell_type": "code",
    "execution_count": 18,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "V[:2]"
@@ -444,20 +333,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Notice how the axes are flipped."
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Now let's look at the explained variance ratio:"
    ]
@@ -465,11 +348,7 @@
   {
    "cell_type": "code",
    "execution_count": 19,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca.explained_variance_ratio_"
@@ -477,20 +356,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "The first dimension explains 84.2% of the variance, while the second explains 14.6%."
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "By projecting down to 2D, we lost about 1.1% of the variance:"
    ]
@@ -498,11 +371,7 @@
   {
    "cell_type": "code",
    "execution_count": 20,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "1 - pca.explained_variance_ratio_.sum()"
@@ -510,10 +379,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Here is how to compute the explained variance ratio using the SVD approach (recall that `s` is the diagonal of the matrix `S`):"
    ]
@@ -521,11 +387,7 @@
   {
    "cell_type": "code",
    "execution_count": 21,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.square(s) / np.square(s).sum()"
@@ -533,20 +395,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Next, let's generate some nice figures! :)"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Utility class to draw 3D arrows (copied from http://stackoverflow.com/questions/11140163)"
    ]
@@ -555,9 +411,7 @@
    "cell_type": "code",
    "execution_count": 22,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -578,10 +432,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Express the plane as a function of x and y."
    ]
@@ -589,11 +440,7 @@
   {
    "cell_type": "code",
    "execution_count": 23,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "axes = [-1.8, 1.8, -1.3, 1.3, -1.0, 1.0]\n",
@@ -609,10 +456,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Plot the 3D dataset, the plane and the projections on that plane."
    ]
@@ -620,11 +464,7 @@
   {
    "cell_type": "code",
    "execution_count": 24,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from mpl_toolkits.mplot3d import Axes3D\n",
@@ -666,11 +506,7 @@
   {
    "cell_type": "code",
    "execution_count": 25,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "fig = plt.figure()\n",
@@ -690,10 +526,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# Manifold learning\n",
     "Swiss roll:"
@@ -703,9 +536,7 @@
    "cell_type": "code",
    "execution_count": 26,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -716,11 +547,7 @@
   {
    "cell_type": "code",
    "execution_count": 27,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "axes = [-11.5, 14, -2, 23, -12, 15]\n",
@@ -744,11 +571,7 @@
   {
    "cell_type": "code",
    "execution_count": 28,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "plt.figure(figsize=(11, 4))\n",
@@ -773,11 +596,7 @@
   {
    "cell_type": "code",
    "execution_count": 29,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from matplotlib import gridspec\n",
@@ -857,10 +676,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# PCA"
    ]
@@ -868,11 +684,7 @@
   {
    "cell_type": "code",
    "execution_count": 30,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "angle = np.pi / 5\n",
@@ -937,10 +749,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# MNIST compression"
    ]
@@ -949,9 +758,7 @@
    "cell_type": "code",
    "execution_count": 31,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -964,9 +771,7 @@
    "cell_type": "code",
    "execution_count": 32,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -981,11 +786,7 @@
   {
    "cell_type": "code",
    "execution_count": 33,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca = PCA()\n",
@@ -997,11 +798,7 @@
   {
    "cell_type": "code",
    "execution_count": 34,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "d"
@@ -1010,11 +807,7 @@
   {
    "cell_type": "code",
    "execution_count": 35,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca = PCA(n_components=0.95)\n",
@@ -1024,11 +817,7 @@
   {
    "cell_type": "code",
    "execution_count": 36,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca.n_components_"
@@ -1037,11 +826,7 @@
   {
    "cell_type": "code",
    "execution_count": 37,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.sum(pca.explained_variance_ratio_)"
@@ -1050,11 +835,7 @@
   {
    "cell_type": "code",
    "execution_count": 38,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "pca = PCA(n_components = 154)\n",
@@ -1066,9 +847,7 @@
    "cell_type": "code",
    "execution_count": 39,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1091,11 +870,7 @@
   {
    "cell_type": "code",
    "execution_count": 40,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "plt.figure(figsize=(7, 4))\n",
@@ -1113,9 +888,7 @@
    "cell_type": "code",
    "execution_count": 41,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1124,10 +897,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "## Incremental PCA"
    ]
@@ -1135,11 +905,7 @@
   {
    "cell_type": "code",
    "execution_count": 42,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.decomposition import IncrementalPCA\n",
@@ -1157,9 +923,7 @@
    "cell_type": "code",
    "execution_count": 43,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1169,11 +933,7 @@
   {
    "cell_type": "code",
    "execution_count": 44,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "plt.figure(figsize=(7, 4))\n",
@@ -1188,9 +948,7 @@
    "cell_type": "code",
    "execution_count": 45,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1199,10 +957,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Let's compare the results of transforming MNIST using regular PCA and incremental PCA. First, the means are equal: "
    ]
@@ -1210,11 +965,7 @@
   {
    "cell_type": "code",
    "execution_count": 46,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(pca.mean_, inc_pca.mean_)"
@@ -1222,10 +973,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "But the results are not exactly identical. Incremental PCA gives a very good approximate solution, but it's not perfect:"
    ]
@@ -1233,11 +981,7 @@
   {
    "cell_type": "code",
    "execution_count": 47,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "np.allclose(X_reduced_pca, X_reduced_inc_pca)"
@@ -1245,20 +989,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "### Using `memmap()`"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Let's create the `memmap()` structure and copy the MNIST data into it. This would typically be done by a first program:"
    ]
@@ -1266,11 +1004,7 @@
   {
    "cell_type": "code",
    "execution_count": 48,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "filename = \"my_mnist.data\"\n",
@@ -1282,10 +1016,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Now deleting the `memmap()` object will trigger its Python finalizer, which ensures that the data is saved to disk."
    ]
@@ -1293,11 +1024,7 @@
   {
    "cell_type": "code",
    "execution_count": 49,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "del X_mm"
@@ -1305,10 +1032,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Next, another program would load the data and use it for training:"
    ]
@@ -1316,11 +1040,7 @@
   {
    "cell_type": "code",
    "execution_count": 50,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "X_mm = np.memmap(filename, dtype=\"float32\", mode=\"readonly\", shape=(m, n))\n",
@@ -1333,11 +1053,7 @@
   {
    "cell_type": "code",
    "execution_count": 51,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "rnd_pca = PCA(n_components=154, svd_solver=\"randomized\", random_state=42)\n",
@@ -1346,20 +1062,14 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "## Time complexity"
    ]
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Let's time regular PCA against Incremental PCA and Randomized PCA, for various number of principal components:"
    ]
@@ -1367,11 +1077,7 @@
   {
    "cell_type": "code",
    "execution_count": 52,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "import time\n",
@@ -1391,10 +1097,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "Now let's compare PCA and Randomized PCA for datasets of different sizes (number of instances):"
    ]
@@ -1402,11 +1105,7 @@
   {
    "cell_type": "code",
    "execution_count": 53,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "times_rpca = []\n",
@@ -1435,10 +1134,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "And now let's compare their performance on datasets of 2,000 instances with various numbers of features:"
    ]
@@ -1447,9 +1143,6 @@
    "cell_type": "code",
    "execution_count": 54,
    "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true,
     "scrolled": true
    },
    "outputs": [],
@@ -1480,10 +1173,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# Kernel PCA"
    ]
@@ -1492,9 +1182,7 @@
    "cell_type": "code",
    "execution_count": 55,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1505,9 +1193,7 @@
    "cell_type": "code",
    "execution_count": 56,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1520,11 +1206,7 @@
   {
    "cell_type": "code",
    "execution_count": 57,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.decomposition import KernelPCA\n",
@@ -1558,11 +1240,7 @@
   {
    "cell_type": "code",
    "execution_count": 58,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "plt.figure(figsize=(6, 5))\n",
@@ -1586,11 +1264,7 @@
   {
    "cell_type": "code",
    "execution_count": 59,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "X_reduced = rbf_pca.fit_transform(X)\n",
@@ -1606,11 +1280,7 @@
   {
    "cell_type": "code",
    "execution_count": 60,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.model_selection import GridSearchCV\n",
@@ -1634,11 +1304,7 @@
   {
    "cell_type": "code",
    "execution_count": 61,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "print(grid_search.best_params_)"
@@ -1647,11 +1313,7 @@
   {
    "cell_type": "code",
    "execution_count": 62,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "rbf_pca = KernelPCA(n_components = 2, kernel=\"rbf\", gamma=0.0433,\n",
@@ -1663,11 +1325,7 @@
   {
    "cell_type": "code",
    "execution_count": 63,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.metrics import mean_squared_error\n",
@@ -1677,10 +1335,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# LLE"
    ]
@@ -1689,9 +1344,7 @@
    "cell_type": "code",
    "execution_count": 64,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1701,11 +1354,7 @@
   {
    "cell_type": "code",
    "execution_count": 65,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.manifold import LocallyLinearEmbedding\n",
@@ -1717,11 +1366,7 @@
   {
    "cell_type": "code",
    "execution_count": 66,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "plt.title(\"Unrolled swiss roll using LLE\", fontsize=14)\n",
@@ -1737,10 +1382,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "source": [
     "# MDS, Isomap and t-SNE"
    ]
@@ -1748,11 +1390,7 @@
   {
    "cell_type": "code",
    "execution_count": 67,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.manifold import MDS\n",
@@ -1764,11 +1402,7 @@
   {
    "cell_type": "code",
    "execution_count": 68,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.manifold import Isomap\n",
@@ -1781,9 +1415,7 @@
    "cell_type": "code",
    "execution_count": 69,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": [
@@ -1796,11 +1428,7 @@
   {
    "cell_type": "code",
    "execution_count": 70,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis\n",
@@ -1815,11 +1443,7 @@
   {
    "cell_type": "code",
    "execution_count": 71,
-   "metadata": {
-    "collapsed": false,
-    "deletable": true,
-    "editable": true
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "titles = [\"MDS\", \"Isomap\", \"t-SNE\"]\n",
@@ -1843,31 +1467,326 @@
   {
    "cell_type": "markdown",
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "source": [
     "# Exercise solutions"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 1. to 8."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "See appendix A."
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "source": [
-    "**Coming soon**"
+    "## 9."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "*Exercise: Load the MNIST dataset (introduced in chapter 3) and split it into a training set and a test set (take the first 60,000 instances for training, and the remaining 10,000 for testing).*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 72,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "from sklearn.datasets import fetch_mldata\n",
+    "mnist = fetch_mldata('MNIST original')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 73,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "X_train = mnist['data'][:60000]\n",
+    "y_train = mnist['target'][:60000]\n",
+    "\n",
+    "X_test = mnist['data'][60000:]\n",
+    "y_test = mnist['target'][60000:]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "*Exercise: Train a Random Forest classifier on the dataset and time how long it takes, then evaluate the resulting model on the test set.*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 74,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "from sklearn.ensemble import RandomForestClassifier\n",
+    "\n",
+    "rnd_clf = RandomForestClassifier(random_state=42)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 75,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "import time\n",
+    "\n",
+    "t0 = time.time()\n",
+    "rnd_clf.fit(X_train, y_train)\n",
+    "t1 = time.time()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 76,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(\"Training took {:.2f}s\".format(t1 - t0))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 77,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from sklearn.metrics import accuracy_score\n",
+    "\n",
+    "y_pred = rnd_clf.predict(X_test)\n",
+    "accuracy_score(y_test, y_pred)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "*Exercise: Next, use PCA to reduce the dataset's dimensionality, with an explained variance ratio of 95%.*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 78,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "from sklearn.decomposition import PCA\n",
+    "\n",
+    "pca = PCA(n_components=0.95)\n",
+    "X_train_reduced = pca.fit_transform(X_train)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "*Exercise: Train a new Random Forest classifier on the reduced dataset and see how long it takes. Was training much faster?*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 79,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "rnd_clf2 = RandomForestClassifier(random_state=42)\n",
+    "t0 = time.time()\n",
+    "rnd_clf2.fit(X_train_reduced, y_train)\n",
+    "t1 = time.time()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 80,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(\"Training took {:.2f}s\".format(t1 - t0))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Oh no! Training is actually 3 times slower now! How can that be? Well, as we saw in this chapter, dimensionality reduction does not always lead to faster training time: it depends on the dataset, the model and the training algorithm. See figure 8-6 (the `manifold_decision_boundary_plot*` plots above). If you try a softmax classifier instead of a random forest classifier, you will find that training time is reduced by a factor of 3 when using PCA. Actually, we will do this in a second, but first let's check the precision of the new random forest classifier."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "*Exercise: Next evaluate the classifier on the test set: how does it compare to the previous classifier?*"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 81,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "X_test_reduced = pca.transform(X_test)\n",
+    "\n",
+    "y_pred = rnd_clf2.predict(X_test_reduced)\n",
+    "accuracy_score(y_test, y_pred)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "It is common for performance to drop slightly when reducing dimensionality, because we do lose some useful signal in the process. However, the performance drop is rather severe in this case. So PCA really did not help: it slowed down training and reduced performance. :(\n",
+    "\n",
+    "Let's see if it helps when using softmax regression:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 82,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "from sklearn.linear_model import LogisticRegression\n",
+    "\n",
+    "log_clf = LogisticRegression(multi_class=\"multinomial\", solver=\"lbfgs\", random_state=42)\n",
+    "t0 = time.time()\n",
+    "log_clf.fit(X_train, y_train)\n",
+    "t1 = time.time()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 83,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(\"Training took {:.2f}s\".format(t1 - t0))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 84,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "y_pred = log_clf.predict(X_test)\n",
+    "accuracy_score(y_test, y_pred)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Okay, so softmax regression takes much longer to train on this dataset than the random forest classifier, plus it performs worse on the test set. But that's not what we are interested in right now, we want to see how much PCA can help softmax regression. Let's train the softmax regression model using the reduced dataset:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 85,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "log_clf2 = LogisticRegression(multi_class=\"multinomial\", solver=\"lbfgs\", random_state=42)\n",
+    "t0 = time.time()\n",
+    "log_clf2.fit(X_train_reduced, y_train)\n",
+    "t1 = time.time()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 86,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(\"Training took {:.2f}s\".format(t1 - t0))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Nice! Reducing dimensionality led to a ×4 speedup. :)  Let's the model's accuracy:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 87,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "y_pred = log_clf2.predict(X_test_reduced)\n",
+    "accuracy_score(y_test, y_pred)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "A very slight drop in performance, which might be a reasonable price to pay for a ×4 speedup, depending on the application."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "So there you have it: PCA can give you a formidable speedup... but not always!"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 10."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Coming soon."
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
-    "collapsed": true,
-    "deletable": true,
-    "editable": true
+    "collapsed": true
    },
    "outputs": [],
    "source": []
@@ -1906,5 +1825,5 @@
   }
  },
  "nbformat": 4,
- "nbformat_minor": 0
+ "nbformat_minor": 1
 }