Merge remote-tracking branch 'upstream/master'

Conflicts:
	sklearn/metrics/pairwise.py

.mailmap

@@ -1,8 +1,12 @@
 Gael Varoquaux <gael.varoquaux@normalesup.org> gvaroquaux <gael.varoquaux@normalesup.org>
 Gael Varoquaux <gael.varoquaux@normalesup.org> Gael varoquaux <gael.varoquaux@normalesup.org>
 Gael Varoquaux <gael.varoquaux@normalesup.org> GaelVaroquaux <gael.varoquaux@normalesup.org>
+Gael Varoquaux <gael.varoquaux@normalesup.org> Varoquaux <varoquau@normalesup.org>
 Olivier Grisel <olivier.grisel@ensta.org> ogrisel <olivier.grisel@ensta.org>
+Olivier Grisel <olivier.grisel@ensta.org> Olivier Grisel <ogrisel@turingcarpet.(none)>
 Alexandre Gramfort <alexandre.gramfort@inria.fr> Alexandre Gramfort <alexandre.gramfort@gmail.com>
+Alexandre Gramfort <alexandre.gramfort@inria.fr> Alexandre Gramfort <alexandre.gramfort@m4x.org>
+Alexandre Gramfort <alexandre.gramfort@inria.fr> Alexandre Gramfort <gramfort@localhost.(none)>
 Matthieu Perrot <matthieu.perrot@cea.fr> Matthieu Perrot <revilyo@earth.(none)>
 Matthieu Perrot <matthieu.perrot@cea.fr> revilyo <revilyo@earth.(none)>
 Vincent Michel <vincent.michel@inria.fr> vincent <vincent@vincent.org>
@@ -12,6 +16,7 @@ Vincent Michel <vincent.michel@inria.fr> Vincent M <vm.michel@gmail.com>
 Vincent Michel <vincent.michel@inria.fr> Vincent Michel <vincent.michel@logilab.fr>
 Vincent Michel <vincent.michel@inria.fr> Vincent M <vincent.michel@logilab.fr>
 Vincent Michel <vincent.michel@inria.fr> Vincent michel <vmic@crater2.logilab.fr>
+Vincent Michel <vincent.michel@inria.fr> Vincent Michel <vm.michel@gmail.com>
 Ariel Rokem <arokem@berkeley.edu> arokem <arokem@berkeley.edu>
 Bertrand Thirion <bertrand.thirion@inria.fr> bthirion <bertrand.thirion@inria.fr>
 Peter Prettenhofer <peter.prettenhofer@gmail.com> pprett <peter.prettenhofer@gmail.com>
@@ -23,19 +28,41 @@ James Bergstra <james.bergstra@gmail.com> james.bergstra <james.bergstra@gmail.c
 Xinfan Meng <mxf3306@gmail.com> mxf <mxf@chomsky.localdomain>
 Jan Schlüter <scikit-learn@jan-schlueter.de> f0k <scikit-learn@jan-schlueter.de>
 Vlad Niculae <vlad@vene.ro> vene <vlad@vene.ro>
-Andreas Müller <amueller@ais.uni-bonn.de> amueller <amueller@ais.uni-bonn.de>
 Virgile Fritsch <virgile.fritsch@gmail.com> VirgileFritsch <virgile.fritsch@gmail.com>
 Virgile Fritsch <virgile.fritsch@gmail.com> Virgile <virgile.fritsch@gmail.com>
+Virgile Fritsch <virgile.fritsch@gmail.com> Virgile <virgile@virgile-Precision-M4400.(none)>
 Jean Kossaifi <jean.kossaifi@gmail.com> Jean  KOSSAIFI <jkossaifi@is208616.intra.cea.fr>
 Jean Kossaifi <jean.kossaifi@gmail.com> JeanKossaifi <jean.kossaifi@gmail.com>
-Jake Vanderplas <vanderplas@astro.washington.edu> Jacob Vanderplas <jakevdp@yahoo.com>
+Jean Kossaifi <jean.kossaifi@gmail.com> Jean Kossaifi <kossaifi@is208616.intra.cea.fr>
+Jake VanderPlas <vanderplas@astro.washington.edu> Jacob Vanderplas <jakevdp@yahoo.com>
+Jake VanderPlas <vanderplas@astro.washington.edu> Jake Vanderplas <jakevdp@yahoo.com>
+Jake VanderPlas <vanderplas@astro.washington.edu> Jake Vanderplas <vanderplas@astro.washington.edu>
 Andreas Mueller <amueller@ais.uni-bonn.de> Andy <amueller@ais.uni-bonn.de>
 Andreas Mueller <amueller@ais.uni-bonn.de> andy <andy@marvin>
 Andreas Mueller <amueller@ais.uni-bonn.de> Andreas Mueller <amueller@templateimage.ista.local>
+Andreas Mueller <amueller@ais.uni-bonn.de> Andreas Müller <amueller@ais.uni-bonn.de>
 Brian Holt <bh00038@cvplws63.eps.surrey.ac.uk> bdholt1 <bdholt1@gmail.com>
+Brian Holt <bh00038@cvplws63.eps.surrey.ac.uk> Brian Holt <bdholt1@gmail.com>
 Robert Layton <robertlayton@gmail.com> robertlayton <robertlayton@gmail.com>
+Robert Layton <robertlayton@gmail.com> = <robertlayton@gmail.com>
 Fabian Pedregosa <fabian@fseoane.net> Fabian Pedregosa <fabian.pedregosa@inria.fr>
 Lars Buitinck <L.J.Buitinck@uva.nl> Lars Buitinck <larsmans@gmail.com>
 Lars Buitinck <L.J.Buitinck@uva.nl> unknown <Lars@.(none)>
 Lars Buitinck <L.J.Buitinck@uva.nl> Lars Buitinck <l.j.buitinck@uva.nl>
 DraXus <draxus@gmail.com> draxus <draxus@hammer.ugr>
+Edouard DUCHESNAY <ed203246@is206877.intra.cea.fr>  Edouard Duchesnay <duchesnay@is143433.(none)>
+Edouard DUCHESNAY <ed203246@is206877.intra.cea.fr> Edouard Duchesnay <edouard.duchesnay@gmail.com>
+Edouard DUCHESNAY <ed203246@is206877.intra.cea.fr> duchesnay <edouard.duchesnay@gmail.com>
+Edouard DUCHESNAY <ed203246@is206877.intra.cea.fr> duchesnay <edouard@is2206219.(none)> 
+Emmanuelle Gouillart <emmanuelle.gouillart@nsup.org> Emmanuelle Gouillart <emma@aleph.(none)>
+Emmanuelle Gouillart <emmanuelle.gouillart@nsup.org> emmanuelle <emmanuelle.gouillart@nsup.org> 
+Gilles Louppe <g.louppe@gmail.com> Gilles Louppe <g.louppe@ulg.ac.be>
+Nelle Varoquaux <nelle.varoquaux@gmail.com> Nelle Varoquaux <nelle@phgroup.com>
+Nicolas Pinto <pinto@alum.mit.edu> Nicolas Pinto <pinto@mit.edu>
+Olivier Hervieu <olivier.hervieu@gmail.com> Olivier Hervieu <olivier.hervieu@tinyclues.com>
+Satrajit Ghosh <satra@mit.edu> Satrajit Ghosh <satrajit.ghosh@gmail.com>
+Shiqiao Du <lucidfrontier.45@gmail.com> Shiqiao Du <s.du@freebit.net>
+Shiqiao Du <lucidfrontier.45@gmail.com> Shiqiao <lucidfrontier.45@gmail.com>
+Tim Sheerman-Chase <t.sheerman-chase@surrey.ac.uk> Tim Sheerman-Chase <ts00051@ts00051-desktop.(none)>
+Vincent Schut <schut@sarvision.nl> Vincent Schut <vincent@TIMO.(none)>
+iBayer <mane.desk@gmail.com> ibayer <mane.desk@gmail.com>

doc/developers/index.rst

@@ -75,29 +75,24 @@ repository <http://github.com/scikit-learn/scikit-learn/>`__ on GitHub:
 
         $ git clone git@github.com:YourLogin/scikit-learn.git
 
- 4. Work on this copy, on your computer, using Git to do the version
-    control::
+ 4. Create a branch to hold your changes::
 
-        $ git add modified_files
-        $ git commit
-        $ git push origin master
-
-    and so on.
+        $ git checkout -b my-feature
 
-If your changes are not just trivial fixes, it is better to directly
-work in a branch with the name of the feature you are working on. In
-this case, replace step 4 with step 5:
+    and start making changes. Never work in the ``master`` branch!
 
-  5. Create a branch to host your changes and publish it on your public
-     repo::
+ 5. Work on this copy, on your computer, using Git to do the version
+    control. When you're done editing, do::
 
-        $ git checkout -b my-feature
         $ git add modified_files
         $ git commit
-        $ git push origin my-feature
 
-When you are ready, and you have pushed your changes to your GitHub repo, go
-the web page of the repo, and click on 'Pull request' to send us a pull
+    to record your changes in Git, then push them to GitHub with::
+
+        $ git push -u origin my-feature
+
+Finally, go to the web page of the your fork of the scikit-learn repo,
+and click 'Pull request' to send your changes to the maintainers for review.
 request. This will send an email to the committers, but might also send an
 email to the mailing list in order to get more visibility.
 
@@ -109,8 +104,7 @@ email to the mailing list in order to get more visibility.
   to use instead of ``origin``. If we choose the name ``upstream`` for it, the
   command will be::
 
-        $ git remote add upstream git@github.com:scikit-learn/scikit-learn.git
-
+        $ git remote add upstream https://github.com/scikit-learn/scikit-learn.git
 
 (If any of the above seems like magic to you, then look up the
 `Git documentation <http://git-scm.com/documentation>`_ on the web.)

doc/modules/clustering.rst

@@ -259,6 +259,19 @@ function of the gradient of the image.
    the spectral clustering solver chooses an arbitrary one, putting 
    the first sample alone in one bin. 
 
+.. warning:: Transforming distance to well-behaved similarities
+
+    Note that if the values of your similarity matrix are not well
+    distributed, e.g. with negative values or with a distance matrix
+    rather than a similarity, the spectral problem will be singular and
+    the problem not solvable. In which case it is advised to apply a
+    transformation to the entries of the matrix. For instance, in the
+    case of a signed distance matrix, is common to apply a heat kernel::
+
+        similarity = np.exp(-beta * distance / distance.std())
+
+    See the examples for such an application.
+
 .. topic:: Examples:
 
  * :ref:`example_cluster_plot_segmentation_toy.py`: Segmenting objects
@@ -894,7 +907,7 @@ cluster analysis.
 
   >>> import numpy as np
   >>> from sklearn.cluster import KMeans
-  >>> kmeans_model = KMeans(k=3, random_state=1).fit(X)
+  >>> kmeans_model = KMeans(n_clusters=3, random_state=1).fit(X)
   >>> labels = kmeans_model.labels_
   >>> metrics.silhouette_score(X, labels, metric='euclidean')  
   ...                                                      # doctest: +ELLIPSIS

doc/modules/linear_model.rst

@@ -392,7 +392,7 @@ orthogonal matching pursuit can approximate the optimum solution vector with a
 fixed number of non-zero elements:
 
 .. math:: \text{arg\,min\,} ||y - X\gamma||_2^2 \text{ subject to } \
-    ||\gamma||_0 \leq n_{nonzero_coefs}
+    ||\gamma||_0 \leq n_{nonzero\_coefs}
 
 Alternatively, orthogonal matching pursuit can target a specific error instead
 of a specific number of non-zero coefficients. This can be expressed as:

doc/modules/sgd.rst

@@ -145,6 +145,9 @@ array of shape [n_classes, n_features] and `intercept_` is a one
 dimensional array of shape [n_classes]. The i-th row of `coef_` holds
 the weight vector of the OVA classifier for the i-th class; classes are
 indexed in ascending order (see attribute `classes`).
+Note that, in principle, since they allow to create a probability model,
+`loss="log"` and `loss="modified_huber"` are more suitable for
+one-vs-all classification.
 
 :class:`SGDClassifier` supports both weighted classes and weighted
 instances via the fit parameters `class_weight` and `sample_weight`. See

doc/modules/svm.rst

@@ -352,10 +352,9 @@ Tips on Practical Use
     set `cache_size` to a higher value than the default of 200(MB),
     such as 500(MB) or 1000(MB).
 
-  * **Setting C**: In constrast to the scaling in LibSVM and LibLinear,
-    the ``C`` parameter in `sklearn.svm` is a per sample penalty.
-    Commonly good values for ``C`` often are very large (i.e. ``10**4``)
-    and seldom below ``1``.
+  * **Setting C**: C is ``1`` by default and it's a reasonable default choice.
+    If you have a lot of noisy observations you should decrease it.
+    It corresponds to regularize more the estimation.
 
   * Support Vector Machine algorithms are not scale invariant, so **it
     is highly recommended to scale your data**. For example, scale each

doc/themes/scikit-learn/layout.html

@@ -226,7 +226,7 @@ <h3>{{ _('This page') }}</h3>
       {% trans last_updated=last_updated|e %}Last updated on {{ last_updated }}.{% endtrans %}
     {%- endif %}
     {%- if show_sphinx %}
-      {% trans sphinx_version=sphinx_version|e %}Created using <a href="http://sphinx.pocoo.org/">Sphinx</a> {{ sphinx_version }}{% endtrans %}. Design by <a href="http://webylimonada.com">Web y Limonada</a>.
+      {% trans sphinx_version=sphinx_version|e %}Created using <a href="http://sphinx.pocoo.org/">Sphinx</a> {{ sphinx_version }}{% endtrans %}. Design by <a href="http://desgrana.es">Desgrana</a>.
     {%- endif %}
     {%- if show_source and has_source and sourcename %}
     <span style="padding-left: 5ex;">

doc/tutorial/index.rst

@@ -35,3 +35,15 @@ Tutorials: From the bottom up with scikit-learn
 .. note:: **Videos**
 
     Videos with tutorials can also be found in the :ref:`videos` section.
+
+.. note:: **Doctest Mode**
+
+   The code-examples in the above tutorials are written in a
+   *python-console* format. If you wish to easily execute these examples
+   in **iPython**, use::
+   	
+	%doctest_mode
+
+   in the iPython-console. You can then simply copy and paste the examples 
+   directly into iPython without having to worry about removing the **>>>** 
+   manually.

doc/tutorial/statistical_inference/unsupervised_learning.rst

@@ -37,9 +37,9 @@ algorithms. The simplest clustering algorithm is the
     >>> X_iris = iris.data
     >>> y_iris = iris.target
 
-    >>> k_means = cluster.KMeans(k=3)
+    >>> k_means = cluster.KMeans(n_clusters=3)
     >>> k_means.fit(X_iris) # doctest: +ELLIPSIS
-    KMeans(copy_x=True, init='k-means++', k=3, max_iter=300,...
+    KMeans(copy_x=True, init='k-means++', ...
     >>> print k_means.labels_[::10]
     [1 1 1 1 1 0 0 0 0 0 2 2 2 2 2]
     >>> print y_iris[::10]
@@ -117,9 +117,9 @@ algorithms. The simplest clustering algorithm is the
         ...    from scipy import misc
         ...    lena = misc.lena()
     	>>> X = lena.reshape((-1, 1)) # We need an (n_sample, n_feature) array
-    	>>> k_means = cluster.KMeans(k=5, n_init=1)
+    	>>> k_means = cluster.KMeans(n_clusters=5, n_init=1)
     	>>> k_means.fit(X) # doctest: +ELLIPSIS
-    	KMeans(copy_x=True, init='k-means++', k=5, ...
+    	KMeans(copy_x=True, init='k-means++', ...
     	>>> values = k_means.cluster_centers_.squeeze()
     	>>> labels = k_means.labels_
     	>>> lena_compressed = np.choose(labels, values)

examples/cluster/plot_cluster_comparison.py

@@ -70,9 +70,9 @@
 
     # create clustering estimators
     ms = cluster.MeanShift(bandwidth=bandwidth, bin_seeding=True)
-    two_means = cluster.MiniBatchKMeans(k=2)
+    two_means = cluster.MiniBatchKMeans(n_clusters=2)
     ward_five = cluster.Ward(n_clusters=2, connectivity=connectivity)
-    spectral = cluster.SpectralClustering(k=2, mode='arpack')
+    spectral = cluster.SpectralClustering(n_clusters=2, mode='arpack')
     dbscan = cluster.DBSCAN(eps=.2)
     affinity_propagation = cluster.AffinityPropagation(damping=.9)
 

examples/cluster/plot_cluster_iris.py

@@ -38,9 +38,9 @@
 X = iris.data
 y = iris.target
 
-estimators = {'k_means_iris_3': KMeans(k=3),
-              'k_means_iris_8': KMeans(k=8),
-              'k_means_iris_bad_init': KMeans(k=3, n_init=1, init='random'),
+estimators = {'k_means_iris_3': KMeans(n_clusters=3),
+              'k_means_iris_8': KMeans(n_clusters=8),
+              'k_means_iris_bad_init': KMeans(n_clusters=3, n_init=1, init='random'),
              }
 
 

examples/cluster/plot_color_quantization.py

@@ -51,7 +51,7 @@
 print "Fitting estimator on a small sub-sample of the data"
 t0 = time()
 image_array_sample = shuffle(image_array, random_state=0)[:1000]
-kmeans = KMeans(k=n_colors, random_state=0).fit(image_array_sample)
+kmeans = KMeans(n_clusters=n_colors, random_state=0).fit(image_array_sample)
 print "done in %0.3fs." % (time() - t0)
 
 # Get labels for all points

examples/cluster/plot_dbscan.py

@@ -28,7 +28,7 @@
 
 ##############################################################################
 # Compute DBSCAN
-db = DBSCAN().fit(S, eps=0.95, min_samples=10)
+db = DBSCAN(eps=0.95, min_samples=10).fit(S)
 core_samples = db.core_sample_indices_
 labels = db.labels_
 

examples/cluster/plot_kmeans_digits.py

@@ -72,16 +72,16 @@ def bench_k_means(estimator, name, data):
                                   sample_size=sample_size),
          )
 
-bench_k_means(KMeans(init='k-means++', k=n_digits, n_init=10),
+bench_k_means(KMeans(init='k-means++', n_clusters=n_digits, n_init=10),
               name="k-means++", data=data)
 
-bench_k_means(KMeans(init='random', k=n_digits, n_init=10),
+bench_k_means(KMeans(init='random', n_clusters=n_digits, n_init=10),
               name="random", data=data)
 
 # in this case the seeding of the centers is deterministic, hence we run the
 # kmeans algorithm only once with n_init=1
 pca = PCA(n_components=n_digits).fit(data)
-bench_k_means(KMeans(init=pca.components_, k=n_digits, n_init=1),
+bench_k_means(KMeans(init=pca.components_, n_clusters=n_digits, n_init=1),
               name="PCA-based",
               data=data)
 print 79 * '_'
@@ -90,7 +90,8 @@ def bench_k_means(estimator, name, data):
 # Visualize the results on PCA-reduced data
 
 reduced_data = PCA(n_components=2).fit_transform(data)
-kmeans = KMeans(init='k-means++', k=n_digits, n_init=10).fit(reduced_data)
+kmeans = KMeans(init='k-means++', n_clusters=n_digits, n_init=10)
+kmeans.fit(reduced_data)
 
 # Step size of the mesh. Decrease to increase the quality of the VQ.
 h = .02     # point in the mesh [x_min, m_max]x[y_min, y_max].

examples/cluster/plot_kmeans_stability_low_dim_dense.py

@@ -87,11 +87,8 @@ def make_data(random_state, n_samples_per_center, grid_size, scale):
     for run_id in range(n_runs):
         X, y = make_data(run_id, n_samples_per_center, grid_size, scale)
         for i, n_init in enumerate(n_init_range):
-            km = factory(k=n_clusters,
-                         init=init,
-                         random_state=run_id,
-                         n_init=n_init,
-                         **params).fit(X)
+            km = factory(n_clusters=n_clusters, init=init, random_state=run_id,
+                    n_init=n_init, **params).fit(X)
             inertia[i, run_id] = km.inertia_
     p = pl.errorbar(n_init_range, inertia.mean(axis=1), inertia.std(axis=1))
     plots.append(p[0])
@@ -105,7 +102,7 @@ def make_data(random_state, n_samples_per_center, grid_size, scale):
 # Part 2: Qualitative visual inspection of the convergence
 
 X, y = make_data(random_state, n_samples_per_center, grid_size, scale)
-km = MiniBatchKMeans(k=n_clusters, init='random', n_init=1,
+km = MiniBatchKMeans(n_clusters=n_clusters, init='random', n_init=1,
                      random_state=random_state).fit(X)
 
 fig = pl.figure()

examples/cluster/plot_lena_compress.py

@@ -33,7 +33,7 @@
     from scipy import misc
     lena = misc.lena()
 X = lena.reshape((-1, 1))  # We need an (n_sample, n_feature) array
-k_means = cluster.KMeans(k=n_clusters, n_init=4)
+k_means = cluster.KMeans(n_clusters=n_clusters, n_init=4)
 k_means.fit(X)
 values = k_means.cluster_centers_.squeeze()
 labels = k_means.labels_

examples/cluster/plot_lena_segmentation.py

@@ -41,7 +41,7 @@
 # Apply spectral clustering (this step goes much faster if you have pyamg
 # installed)
 N_REGIONS = 11
-labels = spectral_clustering(graph, k=N_REGIONS)
+labels = spectral_clustering(graph, n_clusters=N_REGIONS)
 labels = labels.reshape(lena.shape)
 
 ###############################################################################

examples/cluster/plot_mini_batch_kmeans.py

@@ -35,7 +35,7 @@
 ##############################################################################
 # Compute clustering with Means
 
-k_means = KMeans(init='k-means++', k=3, n_init=10)
+k_means = KMeans(init='k-means++', n_clusters=3, n_init=10)
 t0 = time.time()
 k_means.fit(X)
 t_batch = time.time() - t0
@@ -46,7 +46,7 @@
 ##############################################################################
 # Compute clustering with MiniBatchKMeans
 
-mbk = MiniBatchKMeans(init='k-means++', k=3, batch_size=batch_size,
+mbk = MiniBatchKMeans(init='k-means++', n_clusters=3, batch_size=batch_size,
                       n_init=10, max_no_improvement=10, verbose=0)
 t0 = time.time()
 mbk.fit(X)

examples/cluster/plot_segmentation_toy.py

@@ -70,7 +70,7 @@
 
 # Force the solver to be arpack, since amg is numerically
 # unstable on this example
-labels = spectral_clustering(graph, k=4, mode='arpack')
+labels = spectral_clustering(graph, n_clusters=4, mode='arpack')
 label_im = -np.ones(mask.shape)
 label_im[mask] = labels
 
@@ -88,7 +88,7 @@
 graph = image.img_to_graph(img, mask=mask)
 graph.data = np.exp(-graph.data / graph.data.std())
 
-labels = spectral_clustering(graph, k=2, mode='arpack')
+labels = spectral_clustering(graph, n_clusters=2, mode='arpack')
 label_im = -np.ones(mask.shape)
 label_im[mask] = labels
 

examples/decomposition/plot_faces_decomposition.py

@@ -93,8 +93,8 @@ def plot_gallery(title, images):
      True, False),
 
     ('Cluster centers - MiniBatchKMeans',
-     MiniBatchKMeans(k=n_components, tol=1e-3, batch_size=20, max_iter=50,
-                     random_state=rng),
+     MiniBatchKMeans(n_cluster=n_components, tol=1e-3, batch_size=20,
+         max_iter=50, random_state=rng),
      True, False)
 ]