PERF Skip check_finite with scipy.linalg in sklearn.covariance (scikit-learn#18848)

sadakmed · web-flow · commit a8b1905e8f97 · 2020-12-01T08:32:23.000+01:00
When check_array is run on input, use check_finite=False with scipy.linalg to avoid duplicate checks.
diff --git a/sklearn/covariance/_empirical_covariance.py b/sklearn/covariance/_empirical_covariance.py
@@ -167,7 +167,7 @@ def _set_covariance(self, covariance):
         self.covariance_ = covariance
         # set precision
         if self.store_precision:
-            self.precision_ = linalg.pinvh(covariance)
+            self.precision_ = linalg.pinvh(covariance, check_finite=False)
         else:
             self.precision_ = None
 
@@ -182,7 +182,7 @@ def get_precision(self):
         if self.store_precision:
             precision = self.precision_
         else:
-            precision = linalg.pinvh(self.covariance_)
+            precision = linalg.pinvh(self.covariance_, check_finite=False)
         return precision
 
     def fit(self, X, y=None):
diff --git a/sklearn/covariance/_robust_covariance.py b/sklearn/covariance/_robust_covariance.py
@@ -228,7 +228,7 @@ def select_candidates(X, n_support, n_trials, select=1, n_iter=30,
         (2 is enough to be close to the final solution. "Never" exceeds 20).
         This parameter must be a strictly positive integer.
 
-    verbose : bool, default False
+    verbose : bool, default=False
         Control the output verbosity.
 
     cov_computation_method : callable, \
diff --git a/sklearn/covariance/_shrunk_covariance.py b/sklearn/covariance/_shrunk_covariance.py
@@ -176,7 +176,7 @@ def ledoit_wolf_shrinkage(X, assume_centered=False, block_size=1000):
         If False, data will be centered before computation.
 
     block_size : int, default=1000
-        Size of the blocks into which the covariance matrix will be split.
+        Size of blocks into which the covariance matrix will be split.
 
     Returns
     -------
@@ -271,7 +271,7 @@ def ledoit_wolf(X, *, assume_centered=False, block_size=1000):
         If False, data will be centered before computation.
 
     block_size : int, default=1000
-        Size of the blocks into which the covariance matrix will be split.
+        Size of blocks into which the covariance matrix will be split.
         This is purely a memory optimization and does not affect results.
 
     Returns
@@ -339,7 +339,7 @@ class LedoitWolf(EmpiricalCovariance):
         If False (default), data will be centered before computation.
 
     block_size : int, default=1000
-        Size of the blocks into which the covariance matrix will be split
+        Size of blocks into which the covariance matrix will be split
         during its Ledoit-Wolf estimation. This is purely a memory
         optimization and does not affect results.
 
