Save images for split visualization

labdmitriy · Nov 20, 2021 · 06edb45 · 06edb45
1 parent 0e49ed9
commit 06edb45
Show file tree

Hide file tree

Showing 10 changed files with 82 additions and 61 deletions.
diff --git a/images/test_size_n_splits_cv.png b/images/test_size_n_splits_cv.png
diff --git a/images/test_size_n_splits_expanding_window_cv.png b/images/test_size_n_splits_expanding_window_cv.png
diff --git a/images/test_size_n_splits_gap_cv.png b/images/test_size_n_splits_gap_cv.png
diff --git a/images/test_size_n_splits_shift_size_cv.png b/images/test_size_n_splits_shift_size_cv.png
diff --git a/images/test_size_train_size_cv.png b/images/test_size_train_size_cv.png
diff --git a/images/test_size_train_size_n_splits_full_data_cv.png b/images/test_size_train_size_n_splits_full_data_cv.png
diff --git a/images/test_size_train_size_n_splits_partial_data_cv.png b/images/test_size_train_size_n_splits_partial_data_cv.png
diff --git a/images/train_test_split.png b/images/train_test_split.png
diff --git a/ml_lab/model_selection.py b/ml_lab/model_selection.py
@@ -170,7 +170,7 @@ def _calculate_split_params(self) -> None:
         )
 
         if (train_size is None) and (n_splits is not None):
-            train_size = n_groups - gap - (test_size + (n_splits - 1) * shift_size)
+            train_size = n_groups - gap - test_size - (n_splits - 1) * shift_size
             self.train_size = train_size
 
             if train_size <= 0:
@@ -181,7 +181,7 @@ def _calculate_split_params(self) -> None:
                 )
             train_start_idx = 0
         elif (n_splits is None) and (train_size is not None):
-            n_splits = ((n_groups - train_size - gap - test_size) // shift_size) + 1
+            n_splits = (n_groups - train_size - gap - test_size) // shift_size + 1
             self.n_splits = n_splits
 
             if self.n_splits <= 0:

diff --git a/notebooks/GroupTimeSeriesSplit.ipynb b/notebooks/GroupTimeSeriesSplit.ipynb