Merge branch 'dfsg' into debian

* dfsg:
  Corrected macro ROC in example plot_roc
  BF: FIX OvR decision_function_shape in SVC
  Fix fit_transform, stability issue and scale issue in PLS
  FIX class_weight in LogisticRegression and LogisticRegressionCV
  FIX MaxAbsScaler on sparse matrices with 1 row
  Deprecate residues_ in LinearRegression
  Addressed comments on PR scikit-learn#5451
  MAINT Removed deprecated stuff.
  MAINT disable circle ci on 0.17.X
  MAINT: deprecation warns from StandardScaler std_
  FIX: remove shuffling in LabelKFold
  FIX skip LDA deprecation test on python3.3 that has no reload.
  Fix broken examples using RandomTreeEmbeddings
  MAINT Use the full listing of the rackspace wheelhouse for appveyor

Author: yarikoptic

circle.yml

@@ -0,0 +1,5 @@
+general:
+  # Restric the build to the branch master only
+  branches:
+    only:
+       - master

continuous_integration/appveyor/requirements.txt

@@ -2,7 +2,7 @@
 # Those wheels were collected from http://www.lfd.uci.edu/~gohlke/pythonlibs/
 # This is a temporary solution. As soon as numpy and scipy provide official
 # wheel for windows we ca delete this --find-links line.
---find-links http://28daf2247a33ed269873-7b1aad3fab3cc330e1fd9d109892382a.r6.cf2.rackcdn.com/index.html
+--find-links http://28daf2247a33ed269873-7b1aad3fab3cc330e1fd9d109892382a.r6.cf2.rackcdn.com/
 
 # fix the versions of numpy to force the use of numpy and scipy to use the whl
 # of the rackspace folder instead of trying to install from more recent

doc/whats_new.rst

@@ -64,6 +64,10 @@ New features
      shuffling step in the ``cd`` solver.
      By `Tom Dupre la Tour`_ and `Mathieu Blondel`_.
 
+   - **IndexError** bug `#5495
+     <https://github.com/scikit-learn/scikit-learn/issues/5495>`_ when
+     doing OVR(SVC(decision_function_shape="ovr")). Fixed by `Elvis Dohmatob`_.
+
 Enhancements
 ............
    - :class:`manifold.TSNE` now supports approximate optimization via the
@@ -280,6 +284,10 @@ Bug fixes
       <https://github.com/scikit-learn/scikit-learn/pull/4478>`_)
       By `Andreas Müller`_, `Loic Esteve`_ and `Giorgio Patrini`_.
 
+    - Fixed bug in :class:`cross_decomposition.PLS` that yielded unstable and
+      platform dependent output, and failed on `fit_transform`.
+       By `Arthur Mensch`_.
+
 API changes summary
 -------------------
     - Attribute `data_min`, `data_max` and `data_range` in
@@ -3766,3 +3774,5 @@ David Huard, Dave Morrill, Ed Schofield, Travis Oliphant, Pearu Peterson.
 .. _Jean Kossaifi: https://github.com/JeanKossaifi
 .. _Andrew Lamb: https://github.com/andylamb
 .. _Graham Clenaghan: https://github.com/gclenaghan
+.. _Giorgio Patrini: https://github.com/giorgiop
+.. _Elvis Dohmatob: https://github.com/dohmatob

examples/ensemble/plot_feature_transformation.py

@@ -34,10 +34,10 @@
 from sklearn.linear_model import LogisticRegression
 from sklearn.ensemble import (RandomTreesEmbedding, RandomForestClassifier,
                               GradientBoostingClassifier)
-from sklearn.feature_selection import SelectFromModel
 from sklearn.preprocessing import OneHotEncoder
 from sklearn.cross_validation import train_test_split
 from sklearn.metrics import roc_curve
+from sklearn.pipeline import make_pipeline
 
 n_estimator = 10
 X, y = make_classification(n_samples=80000)
@@ -51,13 +51,13 @@
                                                             test_size=0.5)
 
 # Unsupervised transformation based on totally random trees
-rt = RandomTreesEmbedding(max_depth=3, n_estimators=n_estimator)
-rt_lm = LogisticRegression()
-rt.fit(X_train, y_train)
-rt_lm.fit(SelectFromModel(rt, prefit=True).transform(X_train_lr), y_train_lr)
+rt = RandomTreesEmbedding(max_depth=3, n_estimators=n_estimator,
+	random_state=0)
 
-y_pred_rt = rt_lm.predict_proba(
-	SelectFromModel(rt, prefit=True).transform(X_test))[:, 1]
+rt_lm = LogisticRegression()
+pipeline = make_pipeline(rt, rt_lm)
+pipeline.fit(X_train, y_train)
+y_pred_rt = pipeline.predict_proba(X_test)[:, 1]
 fpr_rt_lm, tpr_rt_lm, _ = roc_curve(y_test, y_pred_rt)
 
 # Supervised transformation based on random forests

examples/ensemble/plot_random_forest_embedding.py

@@ -30,17 +30,14 @@
 from sklearn.datasets import make_circles
 from sklearn.ensemble import RandomTreesEmbedding, ExtraTreesClassifier
 from sklearn.decomposition import TruncatedSVD
-from sklearn.feature_selection import SelectFromModel
 from sklearn.naive_bayes import BernoulliNB
 
 # make a synthetic dataset
 X, y = make_circles(factor=0.5, random_state=0, noise=0.05)
 
 # use RandomTreesEmbedding to transform data
 hasher = RandomTreesEmbedding(n_estimators=10, random_state=0, max_depth=3)
-hasher.fit(X)
-model = SelectFromModel(hasher, prefit=True)
-X_transformed = model.transform(X)
+X_transformed = hasher.fit_transform(X)
 
 # Visualize result using PCA
 pca = TruncatedSVD(n_components=2)

examples/model_selection/plot_roc.py

@@ -44,6 +44,7 @@
 from sklearn.cross_validation import train_test_split
 from sklearn.preprocessing import label_binarize
 from sklearn.multiclass import OneVsRestClassifier
+from scipy import interp
 
 # Import some data to play with
 iris = datasets.load_iris()
@@ -99,10 +100,23 @@
 # Plot ROC curves for the multiclass problem
 
 # Compute macro-average ROC curve and ROC area
-fpr["macro"] = np.mean([fpr[i] for i in range(n_classes)], axis=0)
-tpr["macro"] = np.mean([tpr[i] for i in range(n_classes)], axis=0)
+
+# First aggregate all false positive rates
+all_fpr = np.unique(np.concatenate([fpr[i] for i in range(n_classes)]))
+
+# Then interpolate all ROC curves at this points
+mean_tpr = np.zeros_like(all_fpr)
+for i in range(n_classes):
+    mean_tpr += interp(all_fpr, fpr[i], tpr[i])
+
+# Finally average it and compute AUC
+mean_tpr /= n_classes
+
+fpr["macro"] = all_fpr
+tpr["macro"] = mean_tpr
 roc_auc["macro"] = auc(fpr["macro"], tpr["macro"])
 
+# Plot all ROC curves
 plt.figure()
 plt.plot(fpr["micro"], tpr["micro"],
          label='micro-average ROC curve (area = {0:0.2f})'

sklearn/cluster/_feature_agglomeration.py

@@ -11,9 +11,6 @@
 from ..utils import check_array
 from ..utils.validation import check_is_fitted
 
-import warnings
-
-
 ###############################################################################
 # Mixin class for feature agglomeration.
 
@@ -24,7 +21,7 @@ class AgglomerationTransform(TransformerMixin):
 
     pooling_func = np.mean
 
-    def transform(self, X, pooling_func=None):
+    def transform(self, X):
         """
         Transform a new matrix using the built clustering
 
@@ -34,25 +31,14 @@ def transform(self, X, pooling_func=None):
             A M by N array of M observations in N dimensions or a length
             M array of M one-dimensional observations.
 
-        pooling_func : callable, default=np.mean
-            This combines the values of agglomerated features into a single
-            value, and should accept an array of shape [M, N] and the keyword
-            argument `axis=1`, and reduce it to an array of size [M].
-
         Returns
         -------
         Y : array, shape = [n_samples, n_clusters] or [n_clusters]
             The pooled values for each feature cluster.
         """
         check_is_fitted(self, "labels_")
 
-        if pooling_func is not None:
-            warnings.warn("The pooling_func parameter is deprecated since 0.15 "
-                          "and will be removed in 0.18. "
-                          "Pass it to the constructor instead.",
-                          DeprecationWarning)
-        else:
-            pooling_func = self.pooling_func
+        pooling_func = self.pooling_func
         X = check_array(X)
         nX = []
         if len(self.labels_) != X.shape[1]:

sklearn/cluster/dbscan_.py

@@ -9,8 +9,6 @@
 #
 # License: BSD 3 clause
 
-import warnings
-
 import numpy as np
 from scipy import sparse
 
@@ -24,8 +22,7 @@
 
 
 def dbscan(X, eps=0.5, min_samples=5, metric='minkowski',
-           algorithm='auto', leaf_size=30, p=2, sample_weight=None,
-           random_state=None):
+           algorithm='auto', leaf_size=30, p=2, sample_weight=None):
     """Perform DBSCAN clustering from vector array or distance matrix.
 
     Read more in the :ref:`User Guide <dbscan>`.
@@ -75,10 +72,6 @@ def dbscan(X, eps=0.5, min_samples=5, metric='minkowski',
         weight may inhibit its eps-neighbor from being core.
         Note that weights are absolute, and default to 1.
 
-    random_state: numpy.RandomState, optional
-        Deprecated and ignored as of version 0.16, will be removed in version
-        0.18. DBSCAN does not use random initialization.
-
     Returns
     -------
     core_samples : array [n_core_samples]
@@ -109,11 +102,6 @@ def dbscan(X, eps=0.5, min_samples=5, metric='minkowski',
     """
     if not eps > 0.0:
         raise ValueError("eps must be positive.")
-    if random_state is not None:
-        warnings.warn("The parameter random_state is deprecated in 0.16 "
-                      "and will be removed in version 0.18. "
-                      "DBSCAN is deterministic except for rare border cases.",
-                      category=DeprecationWarning)
 
     X = check_array(X, accept_sparse='csr')
     if sample_weight is not None:
@@ -195,9 +183,6 @@ class DBSCAN(BaseEstimator, ClusterMixin):
         of the construction and query, as well as the memory required
         to store the tree. The optimal value depends
         on the nature of the problem.
-    random_state: numpy.RandomState, optional
-        Deprecated and ignored as of version 0.16, will be removed in version
-        0.18. DBSCAN does not use random initialization.
 
     Attributes
     ----------
@@ -233,14 +218,13 @@ class DBSCAN(BaseEstimator, ClusterMixin):
     """
 
     def __init__(self, eps=0.5, min_samples=5, metric='euclidean',
-                 algorithm='auto', leaf_size=30, p=None, random_state=None):
+                 algorithm='auto', leaf_size=30, p=None):
         self.eps = eps
         self.min_samples = min_samples
         self.metric = metric
         self.algorithm = algorithm
         self.leaf_size = leaf_size
         self.p = p
-        self.random_state = random_state
 
     def fit(self, X, y=None, sample_weight=None):
         """Perform DBSCAN clustering from features or distance matrix.

sklearn/cluster/hierarchical.py

@@ -86,8 +86,7 @@ def _fix_connectivity(X, connectivity, n_components=None,
 ###############################################################################
 # Hierarchical tree building functions
 
-def ward_tree(X, connectivity=None, n_components=None, n_clusters=None,
-              return_distance=False):
+def ward_tree(X, connectivity=None, n_clusters=None, return_distance=False):
     """Ward clustering based on a Feature matrix.
 
     Recursively merges the pair of clusters that minimally increases
@@ -111,12 +110,6 @@ def ward_tree(X, connectivity=None, n_components=None, n_clusters=None,
         be symmetric and only the upper triangular half is used.
         Default is None, i.e, the Ward algorithm is unstructured.
 
-    n_components : int (optional)
-        Number of connected components. If None the number of connected
-        components is estimated from the connectivity matrix.
-        NOTE: This parameter is now directly determined directly
-        from the connectivity matrix and will be removed in 0.18
-
     n_clusters : int (optional)
         Stop early the construction of the tree at n_clusters. This is
         useful to decrease computation time if the number of clusters is
@@ -199,11 +192,6 @@ def ward_tree(X, connectivity=None, n_components=None, n_clusters=None,
         else:
             return children_, 1, n_samples, None
 
-    if n_components is not None:
-        warnings.warn(
-            "n_components is now directly calculated from the connectivity "
-            "matrix and will be removed in 0.18",
-            DeprecationWarning)
     connectivity, n_components = _fix_connectivity(X, connectivity)
     if n_clusters is None:
         n_nodes = 2 * n_samples - 1
@@ -326,12 +314,6 @@ def linkage_tree(X, connectivity=None, n_components=None,
         be symmetric and only the upper triangular half is used.
         Default is None, i.e, the Ward algorithm is unstructured.
 
-    n_components : int (optional)
-        Number of connected components. If None the number of connected
-        components is estimated from the connectivity matrix.
-        NOTE: This parameter is now directly determined directly
-        from the connectivity matrix and will be removed in 0.18
-
     n_clusters : int (optional)
         Stop early the construction of the tree at n_clusters. This is
         useful to decrease computation time if the number of clusters is
@@ -435,11 +417,6 @@ def linkage_tree(X, connectivity=None, n_components=None,
             return children_, 1, n_samples, None, distances
         return children_, 1, n_samples, None
 
-    if n_components is not None:
-        warnings.warn(
-            "n_components is now directly calculated from the connectivity "
-            "matrix and will be removed in 0.18",
-            DeprecationWarning)
     connectivity, n_components = _fix_connectivity(X, connectivity)
 
     connectivity = connectivity.tocoo()
@@ -636,12 +613,6 @@ class AgglomerativeClustering(BaseEstimator, ClusterMixin):
         By default, no caching is done. If a string is given, it is the
         path to the caching directory.
 
-    n_components : int (optional)
-        Number of connected components. If None the number of connected
-        components is estimated from the connectivity matrix.
-        NOTE: This parameter is now directly determined from the connectivity
-        matrix and will be removed in 0.18
-
     compute_full_tree : bool or 'auto' (optional)
         Stop early the construction of the tree at n_clusters. This is
         useful to decrease computation time if the number of clusters is
@@ -689,12 +660,10 @@ class AgglomerativeClustering(BaseEstimator, ClusterMixin):
 
     def __init__(self, n_clusters=2, affinity="euclidean",
                  memory=Memory(cachedir=None, verbose=0),
-                 connectivity=None, n_components=None,
-                 compute_full_tree='auto', linkage='ward',
-                 pooling_func=np.mean):
+                 connectivity=None, compute_full_tree='auto',
+                 linkage='ward', pooling_func=np.mean):
         self.n_clusters = n_clusters
         self.memory = memory
-        self.n_components = n_components
         self.connectivity = connectivity
         self.compute_full_tree = compute_full_tree
         self.linkage = linkage
@@ -760,7 +729,6 @@ def fit(self, X, y=None):
             kwargs['affinity'] = self.affinity
         self.children_, self.n_components_, self.n_leaves_, parents = \
             memory.cache(tree_builder)(X, connectivity,
-                                       n_components=self.n_components,
                                        n_clusters=n_clusters,
                                        **kwargs)
         # Cut the tree
@@ -807,12 +775,6 @@ class FeatureAgglomeration(AgglomerativeClustering, AgglomerationTransform):
         By default, no caching is done. If a string is given, it is the
         path to the caching directory.
 
-    n_components : int (optional)
-        Number of connected components. If None the number of connected
-        components is estimated from the connectivity matrix.
-        NOTE: This parameter is now directly determined from the connectivity
-        matrix and will be removed in 0.18
-
     compute_full_tree : bool or 'auto', optional, default "auto"
         Stop early the construction of the tree at n_clusters. This is
         useful to decrease computation time if the number of clusters is

sklearn/cluster/mean_shift_.py

@@ -93,7 +93,7 @@ def _mean_shift_single_seed(my_mean, X, nbrs, max_iter):
 
 def mean_shift(X, bandwidth=None, seeds=None, bin_seeding=False,
                min_bin_freq=1, cluster_all=True, max_iter=300,
-               max_iterations=None, n_jobs=1):
+               n_jobs=1):
     """Perform mean shift clustering of data using a flat kernel.
 
     Read more in the :ref:`User Guide <mean_shift>`.
@@ -161,12 +161,6 @@ def mean_shift(X, bandwidth=None, seeds=None, bin_seeding=False,
     See examples/cluster/plot_meanshift.py for an example.
 
     """
-    # FIXME To be removed in 0.18
-    if max_iterations is not None:
-        warnings.warn("The `max_iterations` parameter has been renamed to "
-                      "`max_iter` from version 0.16. The `max_iterations` "
-                      "parameter will be removed in 0.18", DeprecationWarning)
-        max_iter = max_iterations
 
     if bandwidth is None:
         bandwidth = estimate_bandwidth(X)