Pushing the docs to dev/ for branch: master, commit 6ee390d3fec73a6a57bd1761ff6e69db2e722caa

Author: sklearn-ci

dev/_downloads/3409d9766d352cc9f9b169d4a799a87a/auto_examples_python.zip

@@ -69,7 +69,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "Empirical validation\n====================\n\nWe validate the above bounds on the digits dataset or on the 20 newsgroups\ntext document (TF-IDF word frequencies) dataset:\n\n- for the digits dataset, some 8x8 gray level pixels data for 500\n  handwritten digits pictures are randomly projected to spaces for various\n  larger number of dimensions ``n_components``.\n\n- for the 20 newsgroups dataset some 500 documents with 100k\n  features in total are projected using a sparse random matrix to smaller\n  euclidean spaces with various values for the target number of dimensions\n  ``n_components``.\n\nThe default dataset is the digits dataset. To run the example on the twenty\nnewsgroups dataset, pass the --twenty-newsgroups command line argument to\nthis script.\n\n"
+        "Empirical validation\n====================\n\nWe validate the above bounds on the 20 newsgroups text document\n(TF-IDF word frequencies) dataset or on the digits dataset:\n\n- for the 20 newsgroups dataset some 500 documents with 100k\n  features in total are projected using a sparse random matrix to smaller\n  euclidean spaces with various values for the target number of dimensions\n  ``n_components``.\n\n- for the digits dataset, some 8x8 gray level pixels data for 500\n  handwritten digits pictures are randomly projected to spaces for various\n  larger number of dimensions ``n_components``.\n\nThe default dataset is the 20 newsgroups dataset. To run the example on the\ndigits dataset, pass the ``--use-digits-dataset`` command line argument to\nthis script.\n\n"
       ]
     },
     {
@@ -80,7 +80,7 @@
       },
       "outputs": [],
       "source": [
-        "if '--twenty-newsgroups' in sys.argv:\n    # Need an internet connection hence not enabled by default\n    data = fetch_20newsgroups_vectorized().data[:500]\nelse:\n    data = load_digits().data[:500]"
+        "if '--use-digits-dataset' in sys.argv:\n    data = load_digits().data[:500]\nelse:\n    data = fetch_20newsgroups_vectorized().data[:500]"
       ]
     },
     {
@@ -98,7 +98,7 @@
       },
       "outputs": [],
       "source": [
-        "n_samples, n_features = data.shape\nprint(\"Embedding %d samples with dim %d using various random projections\"\n      % (n_samples, n_features))\n\nn_components_range = np.array([300, 1000, 10000])\ndists = euclidean_distances(data, squared=True).ravel()\n\n# select only non-identical samples pairs\nnonzero = dists != 0\ndists = dists[nonzero]\n\nfor n_components in n_components_range:\n    t0 = time()\n    rp = SparseRandomProjection(n_components=n_components)\n    projected_data = rp.fit_transform(data)\n    print(\"Projected %d samples from %d to %d in %0.3fs\"\n          % (n_samples, n_features, n_components, time() - t0))\n    if hasattr(rp, 'components_'):\n        n_bytes = rp.components_.data.nbytes\n        n_bytes += rp.components_.indices.nbytes\n        print(\"Random matrix with size: %0.3fMB\" % (n_bytes / 1e6))\n\n    projected_dists = euclidean_distances(\n        projected_data, squared=True).ravel()[nonzero]\n\n    plt.figure()\n    plt.hexbin(dists, projected_dists, gridsize=100, cmap=plt.cm.PuBu)\n    plt.xlabel(\"Pairwise squared distances in original space\")\n    plt.ylabel(\"Pairwise squared distances in projected space\")\n    plt.title(\"Pairwise distances distribution for n_components=%d\" %\n              n_components)\n    cb = plt.colorbar()\n    cb.set_label('Sample pairs counts')\n\n    rates = projected_dists / dists\n    print(\"Mean distances rate: %0.2f (%0.2f)\"\n          % (np.mean(rates), np.std(rates)))\n\n    plt.figure()\n    plt.hist(rates, bins=50, range=(0., 2.), edgecolor='k', **density_param)\n    plt.xlabel(\"Squared distances rate: projected / original\")\n    plt.ylabel(\"Distribution of samples pairs\")\n    plt.title(\"Histogram of pairwise distance rates for n_components=%d\" %\n              n_components)\n\n    # TODO: compute the expected value of eps and add them to the previous plot\n    # as vertical lines / region\n\nplt.show()"
+        "n_samples, n_features = data.shape\nprint(\"Embedding %d samples with dim %d using various random projections\"\n      % (n_samples, n_features))\n\nn_components_range = np.array([300, 1000, 10000])\ndists = euclidean_distances(data, squared=True).ravel()\n\n# select only non-identical samples pairs\nnonzero = dists != 0\ndists = dists[nonzero]\n\nfor n_components in n_components_range:\n    t0 = time()\n    rp = SparseRandomProjection(n_components=n_components)\n    projected_data = rp.fit_transform(data)\n    print(\"Projected %d samples from %d to %d in %0.3fs\"\n          % (n_samples, n_features, n_components, time() - t0))\n    if hasattr(rp, 'components_'):\n        n_bytes = rp.components_.data.nbytes\n        n_bytes += rp.components_.indices.nbytes\n        print(\"Random matrix with size: %0.3fMB\" % (n_bytes / 1e6))\n\n    projected_dists = euclidean_distances(\n        projected_data, squared=True).ravel()[nonzero]\n\n    plt.figure()\n    min_dist = min(projected_dists.min(), dists.min())\n    max_dist = max(projected_dists.max(), dists.max())\n    plt.hexbin(dists, projected_dists, gridsize=100, cmap=plt.cm.PuBu,\n               extent=[min_dist, max_dist, min_dist, max_dist])\n    plt.xlabel(\"Pairwise squared distances in original space\")\n    plt.ylabel(\"Pairwise squared distances in projected space\")\n    plt.title(\"Pairwise distances distribution for n_components=%d\" %\n              n_components)\n    cb = plt.colorbar()\n    cb.set_label('Sample pairs counts')\n\n    rates = projected_dists / dists\n    print(\"Mean distances rate: %0.2f (%0.2f)\"\n          % (np.mean(rates), np.std(rates)))\n\n    plt.figure()\n    plt.hist(rates, bins=50, range=(0., 2.), edgecolor='k', **density_param)\n    plt.xlabel(\"Squared distances rate: projected / original\")\n    plt.ylabel(\"Distribution of samples pairs\")\n    plt.title(\"Histogram of pairwise distance rates for n_components=%d\" %\n              n_components)\n\n    # TODO: compute the expected value of eps and add them to the previous plot\n    # as vertical lines / region\n\nplt.show()"
       ]
     },
     {

dev/_downloads/36b58500501fbf3f06587ee0039d1985/plot_johnson_lindenstrauss_bound.ipynb

@@ -102,27 +102,26 @@
 # Empirical validation
 # ====================
 #
-# We validate the above bounds on the digits dataset or on the 20 newsgroups
-# text document (TF-IDF word frequencies) dataset:
-#
-# - for the digits dataset, some 8x8 gray level pixels data for 500
-#   handwritten digits pictures are randomly projected to spaces for various
-#   larger number of dimensions ``n_components``.
+# We validate the above bounds on the 20 newsgroups text document
+# (TF-IDF word frequencies) dataset or on the digits dataset:
 #
 # - for the 20 newsgroups dataset some 500 documents with 100k
 #   features in total are projected using a sparse random matrix to smaller
 #   euclidean spaces with various values for the target number of dimensions
 #   ``n_components``.
 #
-# The default dataset is the digits dataset. To run the example on the twenty
-# newsgroups dataset, pass the --twenty-newsgroups command line argument to
+# - for the digits dataset, some 8x8 gray level pixels data for 500
+#   handwritten digits pictures are randomly projected to spaces for various
+#   larger number of dimensions ``n_components``.
+#
+# The default dataset is the 20 newsgroups dataset. To run the example on the
+# digits dataset, pass the ``--use-digits-dataset`` command line argument to
 # this script.
 
-if '--twenty-newsgroups' in sys.argv:
-    # Need an internet connection hence not enabled by default
-    data = fetch_20newsgroups_vectorized().data[:500]
-else:
+if '--use-digits-dataset' in sys.argv:
     data = load_digits().data[:500]
+else:
+    data = fetch_20newsgroups_vectorized().data[:500]
 
 ##########################################################
 # For each value of ``n_components``, we plot:
@@ -158,7 +157,10 @@
         projected_data, squared=True).ravel()[nonzero]
 
     plt.figure()
-    plt.hexbin(dists, projected_dists, gridsize=100, cmap=plt.cm.PuBu)
+    min_dist = min(projected_dists.min(), dists.min())
+    max_dist = max(projected_dists.max(), dists.max())
+    plt.hexbin(dists, projected_dists, gridsize=100, cmap=plt.cm.PuBu,
+               extent=[min_dist, max_dist, min_dist, max_dist])
     plt.xlabel("Pairwise squared distances in original space")
     plt.ylabel("Pairwise squared distances in projected space")
     plt.title("Pairwise distances distribution for n_components=%d" %