Add deepcopy for lists to prevent hidden mutability of matrices

A bug that caused matrices saved in lists, i.e. using the multi-block
method, to be altered even though the copy parameter was set to true is
fixed. Now those lists are deepcopied.

Author: lvermue

mbpls/__init__.py

@@ -22,4 +22,4 @@
 
 __all__ = ["mbpls", "data"]
 
-__version__ = "1.0.3"
+__version__ = "1.0.4"

mbpls/mbpls.py

@@ -7,6 +7,7 @@
 # License: 3-clause BSD
 
 
+from copy import deepcopy
 from sklearn.base import BaseEstimator, RegressorMixin, TransformerMixin, MultiOutputMixin
 from sklearn.utils.validation import check_X_y, check_array, check_is_fitted, check_consistent_length
 from sklearn import metrics
@@ -298,6 +299,8 @@ def fit(self, X, Y):
         if self.standardize:
             self.x_scalers_ = []
             if isinstance(X, list) and not isinstance(X[0], list):
+                if self.copy:
+                    X = deepcopy(X)
                 if self.sparse_data is True:
                     self.sparse_X_info_ = {}
                 for block in range(len(X)):
@@ -323,6 +326,8 @@ def fit(self, X, Y):
             Y = self.y_scaler_.fit_transform(Y)
         else:
             if isinstance(X, list) and not isinstance(X[0], list):
+                if self.copy:
+                    X = deepcopy(X)
                 if self.sparse_data is True:
                     self.sparse_X_info_ = {}
                 for block in range(len(X)):
@@ -1082,6 +1087,8 @@ def transform(self, X, Y=None, return_block_scores=False, copy=True):
 
         if self.standardize:
             if isinstance(X, list) and not isinstance(X[0], list):
+                if copy:
+                    X = deepcopy(X)
                 for block in range(len(X)):
                     # Check dimensions
                     X[block] = check_array(X[block], dtype=np.float64, force_all_finite=not self.sparse_data, copy=copy)
@@ -1214,6 +1221,8 @@ def transform(self, X, Y=None, return_block_scores=False, copy=True):
 
         else:
             if isinstance(X, list) and not isinstance(X[0], list):
+                if copy:
+                    X = deepcopy(X)
                 for block in range(len(X)):
                     # Check dimensions
                     X[block] = check_array(X[block], dtype=np.float64, force_all_finite=not self.sparse_data, copy=copy)
@@ -1352,6 +1361,8 @@ def predict(self, X, copy=True):
 
         if self.standardize:
             if isinstance(X, list) and not isinstance(X[0], list):
+                if copy:
+                    X = deepcopy(X)
                 for block in range(len(X)):
                     # Check dimensions
                     X[block] = check_array(X[block], dtype=np.float64, force_all_finite=not self.sparse_data, copy=copy)
@@ -1375,6 +1386,8 @@ def predict(self, X, copy=True):
                 y_hat = self.y_scaler_.inverse_transform(X.dot(self.beta_))
         else:
             if isinstance(X, list) and not isinstance(X[0], list):
+                if copy:
+                    X = deepcopy(X)
                 for block in range(len(X)):
                     # Check dimensions
                     X[block] = check_array(X[block], dtype=np.float64, force_all_finite=not self.sparse_data, copy=copy)