[MRG] Remove preprocessing the data for RCA (#194)

wdevazelhes · bellet · commit 85185175f356 · 2019-06-12T17:19:04.000+02:00
* Remove initialization of the data for RCA * Add deprecated flag for supervised version too * Remove comment saying we'll do PCA * Add ChangedBehaviorWarning and do tests * improve change behavior warning * Update message in case covariance matrix is not invertible * FIX: still ignore testing RCA while fixed in #198 * Some reformatting * Fix test string * TST: add test for warning message when covariance is not definite * Address #194 (comment)
diff --git a/metric_learn/rca.py b/metric_learn/rca.py
@@ -17,6 +17,7 @@
 from six.moves import xrange
 from sklearn import decomposition
 from sklearn.base import TransformerMixin
+from sklearn.exceptions import ChangedBehaviorWarning
 
 from ._util import _check_n_components
 from .base_metric import MahalanobisMixin
@@ -48,7 +49,7 @@ class RCA(MahalanobisMixin, TransformerMixin):
   """
 
   def __init__(self, n_components=None, num_dims='deprecated',
-               pca_comps=None, preprocessor=None):
+               pca_comps='deprecated', preprocessor=None):
     """Initialize the learner.
 
     Parameters
@@ -62,12 +63,10 @@ def __init__(self, n_components=None, num_dims='deprecated',
           `num_dims` was deprecated in version 0.5.0 and will
           be removed in 0.6.0. Use `n_components` instead.
 
-    pca_comps : int, float, None or string
-        Number of components to keep during PCA preprocessing.
-        If None (default), does not perform PCA.
-        If ``0 < pca_comps < 1``, it is used as
-        the minimum explained variance ratio.
-        See sklearn.decomposition.PCA for more details.
+    pca_comps : Not used
+      .. deprecated:: 0.5.0
+         `pca_comps` was deprecated in version 0.5.0 and will
+         be removed in 0.6.0.
 
     preprocessor : array-like, shape=(n_samples, n_features) or callable
         The preprocessor to call to get tuples from indices. If array-like,
@@ -83,8 +82,9 @@ def _check_dimension(self, rank, X):
     if rank < d:
       warnings.warn('The inner covariance matrix is not invertible, '
                     'so the transformation matrix may contain Nan values. '
-                    'You should adjust pca_comps to remove noise and '
-                    'redundant information.')
+                    'You should reduce the dimensionality of your input,'
+                    'for instance using `sklearn.decomposition.PCA` as a '
+                    'preprocessing step.')
 
     dim = _check_n_components(d, self.n_components)
     return dim
@@ -105,25 +105,33 @@ def fit(self, X, chunks):
                     ' It has been deprecated in version 0.5.0 and will be'
                     ' removed in 0.6.0. Use "n_components" instead',
                     DeprecationWarning)
+
+    if self.pca_comps != 'deprecated':
+      warnings.warn(
+        '"pca_comps" parameter is not used. '
+        'It has been deprecated in version 0.5.0 and will be'
+        'removed in 0.6.0. RCA will not do PCA preprocessing anymore. If '
+        'you still want to do it, you could use '
+        '`sklearn.decomposition.PCA` and an `sklearn.pipeline.Pipeline`.',
+        DeprecationWarning)
+
     X, chunks = self._prepare_inputs(X, chunks, ensure_min_samples=2)
 
-    # PCA projection to remove noise and redundant information.
-    if self.pca_comps is not None:
-      pca = decomposition.PCA(n_components=self.pca_comps)
-      X_t = pca.fit_transform(X)
-      M_pca = pca.components_
-    else:
-      X_t = X - X.mean(axis=0)
-      M_pca = None
+    warnings.warn(
+      "RCA will no longer center the data before training. If you want "
+      "to do some preprocessing, you should do it manually (you can also "
+      "use an `sklearn.pipeline.Pipeline` for instance). This warning "
+      "will disappear in version 0.6.0.", ChangedBehaviorWarning)
 
-    chunk_mask, chunked_data = _chunk_mean_centering(X_t, chunks)
+    chunks = np.asanyarray(chunks, dtype=int)
+    chunk_mask, chunked_data = _chunk_mean_centering(X, chunks)
 
     inner_cov = np.atleast_2d(np.cov(chunked_data, rowvar=0, bias=1))
-    dim = self._check_dimension(np.linalg.matrix_rank(inner_cov), X_t)
+    dim = self._check_dimension(np.linalg.matrix_rank(inner_cov), X)
 
     # Fisher Linear Discriminant projection
-    if dim < X_t.shape[1]:
-      total_cov = np.cov(X_t[chunk_mask], rowvar=0)
+    if dim < X.shape[1]:
+      total_cov = np.cov(X[chunk_mask], rowvar=0)
       tmp = np.linalg.lstsq(total_cov, inner_cov)[0]
       vals, vecs = np.linalg.eig(tmp)
       inds = np.argsort(vals)[:dim]
@@ -133,9 +141,6 @@ def fit(self, X, chunks):
     else:
       self.transformer_ = _inv_sqrtm(inner_cov).T
 
-    if M_pca is not None:
-        self.transformer_ = np.atleast_2d(self.transformer_.dot(M_pca))
-
     return self
 
 
@@ -155,7 +160,7 @@ class RCA_Supervised(RCA):
   """
 
   def __init__(self, num_dims='deprecated', n_components=None,
-               pca_comps=None, num_chunks=100, chunk_size=2,
+               pca_comps='deprecated', num_chunks=100, chunk_size=2,
                preprocessor=None):
     """Initialize the supervised version of `RCA`.
 
diff --git a/test/metric_learn_test.py b/test/metric_learn_test.py
@@ -18,9 +18,10 @@
   HAS_SKGGM = False
 else:
   HAS_SKGGM = True
-from metric_learn import (LMNN, NCA, LFDA, Covariance, MLKR, MMC, RCA,
+from metric_learn import (LMNN, NCA, LFDA, Covariance, MLKR, MMC,
                           LSML_Supervised, ITML_Supervised, SDML_Supervised,
-                          RCA_Supervised, MMC_Supervised, SDML, ITML, LSML)
+                          RCA_Supervised, MMC_Supervised, SDML, RCA, ITML,
+                          LSML)
 # Import this specially for testing.
 from metric_learn.constraints import wrap_pairs
 from metric_learn.lmnn import _sum_outer_products
@@ -837,6 +838,63 @@ def test_feature_null_variance(self):
     csep = class_separation(rca.transform(X), self.iris_labels)
     self.assertLess(csep, 0.30)
 
+  def test_deprecation_pca_comps(self):
+    # test that a deprecation message is thrown if pca_comps is set at
+    # initialization
+    # TODO: remove in v.0.6
+    X, y = make_classification(random_state=42, n_samples=100)
+    rca_supervised = RCA_Supervised(pca_comps=X.shape[1], num_chunks=20)
+    msg = ('"pca_comps" parameter is not used. '
+           'It has been deprecated in version 0.5.0 and will be'
+           'removed in 0.6.0. RCA will not do PCA preprocessing anymore. If '
+           'you still want to do it, you could use '
+           '`sklearn.decomposition.PCA` and an `sklearn.pipeline.Pipeline`.')
+    with pytest.warns(ChangedBehaviorWarning) as expected_msg:
+      rca_supervised.fit(X, y)
+    assert str(expected_msg[0].message) == msg
+
+    rca = RCA(pca_comps=X.shape[1])
+    with pytest.warns(ChangedBehaviorWarning) as expected_msg:
+      rca.fit(X, y)
+    assert str(expected_msg[0].message) == msg
+
+  def test_changedbehaviorwarning_preprocessing(self):
+    # test that a ChangedBehaviorWarning is thrown when using RCA
+    # TODO: remove in v.0.6
+
+    msg = ("RCA will no longer center the data before training. If you want "
+           "to do some preprocessing, you should do it manually (you can also "
+           "use an `sklearn.pipeline.Pipeline` for instance). This warning "
+           "will disappear in version 0.6.0.")
+
+    X, y = make_classification(random_state=42, n_samples=100)
+    rca_supervised = RCA_Supervised(num_chunks=20)
+    with pytest.warns(ChangedBehaviorWarning) as expected_msg:
+      rca_supervised.fit(X, y)
+    assert str(expected_msg[0].message) == msg
+
+    rca = RCA()
+    with pytest.warns(ChangedBehaviorWarning) as expected_msg:
+      rca.fit(X, y)
+    assert str(expected_msg[0].message) == msg
+
+  def test_rank_deficient_returns_warning(self):
+    """Checks that if the covariance matrix is not invertible, we raise a
+    warning message advising to use PCA"""
+    X, y = load_iris(return_X_y=True)
+    # we make the fourth column a linear combination of the two first,
+    # so that the covariance matrix will not be invertible:
+    X[:, 3] = X[:, 0] + 3 * X[:, 1]
+    rca = RCA()
+    msg = ('The inner covariance matrix is not invertible, '
+           'so the transformation matrix may contain Nan values. '
+           'You should reduce the dimensionality of your input,'
+           'for instance using `sklearn.decomposition.PCA` as a '
+           'preprocessing step.')
+    with pytest.warns(None) as raised_warnings:
+      rca.fit(X, y)
+    assert any(str(w.message) == msg for w in raised_warnings)
+
 
 @pytest.mark.parametrize('num_dims', [None, 2])
 def test_deprecation_num_dims_rca(num_dims):
diff --git a/test/test_base_metric.py b/test/test_base_metric.py
@@ -89,13 +89,13 @@ def test_rca(self):
     self.assertEqual(remove_spaces(str(metric_learn.RCA())),
                      remove_spaces("RCA(n_components=None, "
                                    "num_dims='deprecated', "
-                                   "pca_comps=None, "
+                                   "pca_comps='deprecated', "
                                    "preprocessor=None)"))
     self.assertEqual(remove_spaces(str(metric_learn.RCA_Supervised())),
                      remove_spaces(
                        "RCA_Supervised(chunk_size=2, "
                        "n_components=None, num_chunks=100, "
-                       "num_dims='deprecated', pca_comps=None, "
+                       "num_dims='deprecated', pca_comps='deprecated', "
                        "preprocessor=None)"))
 
   def test_mlkr(self):
