pybear 0.2.1__tar.gz → 0.2.2__tar.gz

{pybear-0.2.1 → pybear-0.2.2}/CHANGELOG.md

@@ -6,6 +6,33 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 
+## [0.2.2] - 2025-09-08
+
+### Added
+- None
+
+### Changed
+- model_selection.autogridsearch_wrapper:
+    Adjust how soft integer space calculates next grid's points when 
+    last best falls on a non-edge value.
+- utilities:
+    - nan_mask_string, nan_mask:
+        Add pandas.NaT to searched nan-like values.
+
+### Deprecated
+- None
+
+### Removed
+- None
+
+### Fixed
+- preprocessing.SlimPolyFeatures
+    Fix pickling error on "large" datasets. Set backend='loky' and 
+    max_nbytes="100M" for all usages of joblib.Parallel.
+
+### Security
+- None
+
 ## [0.2.1] - 2025-08-16
 
 ### Added

{pybear-0.2.1 → pybear-0.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pybear
-Version: 0.2.1
+Version: 0.2.2
 Summary: Python modules for miscellaneous data analytics applications
 License: BSD-3-Clause
 Keywords: python data analytics scikit-API text
@@ -87,6 +87,12 @@ pybear
    :target: https://doi.org/10.5281/zenodo.16547172
    :alt: DOI
 
+|BMC|
+
+.. |BMC| image:: https://www.buymeacoffee.com/assets/img/custom_images/orange_img.png
+   :target: https://www.buymeacoffee.com/pybear
+   :alt: Buy Me A Coffee
+
 .. _documentation: https://pybear.readthedocs.io/
 
 .. |PythonVersion| replace:: >=3.10
@@ -282,7 +288,7 @@ The modules include:
 
 Related Resources
 -----------------
-pybear has a sister package called pybear_dask. A few of the pybear modules
+pybear has a sister package called pybear-dask. A few of the pybear modules
 have a corresponding twin in pybear-dask. You can pip install pybear-dask
 from PyPI in the same way as pybear. There is no Read The Docs website for
 pybear-dask, but it does have a GitHub repo.

{pybear-0.2.1 → pybear-0.2.2}/README.rst

@@ -58,6 +58,12 @@ pybear
    :target: https://doi.org/10.5281/zenodo.16547172
    :alt: DOI
 
+|BMC|
+
+.. |BMC| image:: https://www.buymeacoffee.com/assets/img/custom_images/orange_img.png
+   :target: https://www.buymeacoffee.com/pybear
+   :alt: Buy Me A Coffee
+
 .. _documentation: https://pybear.readthedocs.io/
 
 .. |PythonVersion| replace:: >=3.10
@@ -253,7 +259,7 @@ The modules include:
 
 Related Resources
 -----------------
-pybear has a sister package called pybear_dask. A few of the pybear modules
+pybear has a sister package called pybear-dask. A few of the pybear modules
 have a corresponding twin in pybear-dask. You can pip install pybear-dask
 from PyPI in the same way as pybear. There is no Read The Docs website for
 pybear-dask, but it does have a GitHub repo.

{pybear-0.2.1 → pybear-0.2.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pybear"
-version = "0.2.1"
+version = "0.2.2"
 description = "Python modules for miscellaneous data analytics applications"
 license = "BSD-3-clause"
 authors = ["Bill Sousa <WKSJR99@msn.com>"]

{pybear-0.2.1 → pybear-0.2.2}/src/pybear/model_selection/autogridsearch/_autogridsearch_wrapper/_get_next_param_grid/_int/_int_linspace_gap_gt_1_soft.py

@@ -63,8 +63,16 @@ def _int_linspace_gap_gt_1_soft(
 
     else:
         _best = _SINGLE_GRID[_posn]
-        _left = min(_SINGLE_GRID[_posn - 1] + 1, _best - 1)
-        _right = max(_SINGLE_GRID[_posn + 1] - 1, _best + 1)
+        _left = min(
+            # _SINGLE_GRID[_posn - 1] + 1,
+            _SINGLE_GRID[_posn - 1] + (_best - _SINGLE_GRID[_posn - 1]) // 2,
+            _best - 1
+        )
+        _right = max(
+            # _SINGLE_GRID[_posn + 1] - 1,
+            _SINGLE_GRID[_posn + 1] - (_SINGLE_GRID[_posn + 1] - _best) // 2,
+            _best + 1
+        )
 
 
     # apply universal lower bound

{pybear-0.2.1 → pybear-0.2.2}/src/pybear/model_selection/autogridsearch/autogridsearch_wrapper.py

@@ -541,7 +541,7 @@ def autogridsearch_wrapper(GridSearchParent):
                 _best_params = self.best_params_
 
                 if self.agscv_verbose:
-                    print(f"Pass {_pass + 1} best params: ", end='')
+                    print(f"Pass {_pass + 1} best params: ")
                     for k,v in _best_params.items():
                         try:
                             print(f'{k[:15]}'.ljust(20), f'{v:,.4f}')

{pybear-0.2.1 → pybear-0.2.2}/src/pybear/preprocessing/_SlimPolyFeatures/_partial_fit/_get_dupls_for_combo_in_X_and_poly.py

@@ -160,7 +160,9 @@ def _get_dupls_for_combo_in_X_and_poly(
                     )
                 )
         else:
-            with joblib.parallel_config(prefer='processes', n_jobs=_n_jobs):
+            with joblib.parallel_config(
+                prefer='processes', n_jobs=_n_jobs, backend='loky', max_nbytes="100M"
+            ):
                 _X_dupls = joblib.Parallel(return_as='list')(
                     joblib.delayed(_parallel_chunk_comparer)(
                         _chunk1=_columns_getter(_X, _X_idxs),
@@ -226,7 +228,9 @@ def _get_dupls_for_combo_in_X_and_poly(
                 )
             )
     else:
-        with joblib.parallel_config(prefer='processes', n_jobs=_n_jobs):
+        with joblib.parallel_config(
+            prefer='processes', n_jobs=_n_jobs, backend='loky', max_nbytes="100M"
+        ):
             _poly_dupls = joblib.Parallel(return_as='list')(
                 joblib.delayed(_parallel_chunk_comparer)(
                     _chunk1=_columns_getter(_X, _poly1_idxs),

{pybear-0.2.1 → pybear-0.2.2}/src/pybear/preprocessing/_SlimPolyFeatures/_transform/_build_poly.py

@@ -79,7 +79,9 @@ def _build_poly(
     # @wrap_non_picklable_objects
     # def _poly_stacker(_columns):
     #     return ss.csc_array(_columns.prod(1).reshape((-1,1)))
-    # with joblib.parallel_config(prefer='processes', n_jobs=_n_jobs):
+    # with joblib.parallel_config(
+    #     prefer='processes', n_jobs=_n_jobs, backend='loky', max_nbytes="100M"
+    # ):
     #     POLY = joblib.Parallel(return_as='list')(
     #         joblib.delayed(_poly_stacker)(
     #             _columns_getter(_X, combo)

{pybear-0.2.1 → pybear-0.2.2}/src/pybear/utilities/_nan_masking.py

@@ -225,11 +225,11 @@ def nan_mask_string(
     vector indicating the locations of nan-like representations in the
     data.
 
-    "nan-like representations" include, at least, pandas.NA, None (of
-    type None, not string "None"), and string representations of "nan".
-    This function does not accept scipy sparse matrices or arrays, as
-    dok and lil formats are not handled globally in the nan_mask
-    functions, and the remaining sparse objects can only contain
+    "nan-like representations" include, at least, pandas.NA, pandas.NaT,
+    None (of type None, not string "None"), and string representations
+    of "nan". This function does not accept scipy sparse matrices or
+    arrays, as dok and lil formats are not handled globally in the
+    nan_mask functions, and the remaining sparse objects can only contain
     numeric data.
 
     This function accepts Python lists, tuples, and sets, numpy arrays,
@@ -372,6 +372,8 @@ def nan_mask_string(
 
     M = np.char.replace(M, '<NA>', 'nan')
 
+    M = np.char.replace(M, 'NaT', 'nan')
+
     M = np.char.upper(M)
 
     return (M == 'NAN').astype(bool)
@@ -396,10 +398,10 @@ def nan_mask(
     matrices / arrays of all formats except dok and lil. In that case,
     a numpy boolean vector of shape identical to that of the sparse
     object's 'data' attribute is returned. "nan-like representations"
-    include, at least, np.nan, pandas.NA, None (of type None, not string
-    "None"), and string representations of "nan". This function does not
-    accept any ragged Python built-ins, numpy recarrays, or numpy masked
-    arrays.
+    include, at least, np.nan, pandas.NA, pandas.NaT, None (of type
+    None, not string "None"), and string representations of "nan".
+    This function does not accept any ragged Python built-ins, numpy
+    recarrays, or numpy masked arrays.
 
     Parameters
     ----------

{pybear-0.2.1 → pybear-0.2.2}/tests/model_selection/autogridsearch/autogridsearch_wrapper/get_next_param_grid/drill/drill_test.py

@@ -414,13 +414,13 @@ class TestIntNonUnitGapGT3:
                     assert min(out_grid) == max(1, (__[0] - (__[1] - __[0])))
                     del __
                 else:
-                    assert min(out_grid) == good_grids[0][key][posn-1] + 1
+                    assert min(out_grid) >= good_grids[0][key][posn-1] + 1
             elif 'hard' in good_params[key][-1]:
                 if posn == 0:
                     assert min(out_grid) == good_grids[0][key][0]
                 else:
                     __ = good_grids[0][key]
-                    assert min(out_grid) == \
+                    assert min(out_grid) >= \
                         max(__[0], (__[posn] - (__[posn] - __[posn-1]))) + 1
                     del __
         # END min ** * ** * ** * ** * ** * ** * ** * ** * ** * ** * ** *

{pybear-0.2.1 → pybear-0.2.2}/tests/model_selection/autogridsearch/autogridsearch_wrapper/get_next_param_grid/int/int_linspace_gap_gt_1_hard_test.py

@@ -70,9 +70,9 @@ class TestIntLinspaceGapGT1Hard:
 
         elif posn == 'middle':
 
-            assert _left == max(hard_min, min(grid[0] + 1, grid[1] - 1))
+            assert _left >= max(hard_min, min(grid[0] + 1, grid[1] - 1))
 
-            assert _right == min(hard_max, max(grid[-1] - 1, grid[1] + 1))
+            assert _right <= min(hard_max, max(grid[-1] - 1, grid[1] + 1))
 
 
 

{pybear-0.2.1 → pybear-0.2.2}/tests/model_selection/autogridsearch/autogridsearch_wrapper/get_next_param_grid/int/int_linspace_gap_gt_1_soft_test.py

@@ -19,7 +19,7 @@ class TestIntLinspaceGapGT1Soft:
     # no _validation
 
 
-    # unit gap only to tests for robustness against span falling below 2
+    # unit gap only to test for robustness against span falling below 2
     @pytest.mark.parametrize('grid',
         ([1,3,5], [10,20,30], [10, 15, 25], [3,4,5])
     )
@@ -49,9 +49,9 @@ class TestIntLinspaceGapGT1Soft:
 
         elif posn == 'middle':
 
-            assert _left == min(grid[0] + 1, grid[1] - 1)
+            assert _left >= min(grid[0] + 1, grid[1] - 1)
 
-            assert _right == max(grid[-1] - 1, grid[1] + 1)
+            assert _right <= max(grid[-1] - 1, grid[1] + 1)
 
 
 

{pybear-0.2.1 → pybear-0.2.2}/tests/model_selection/autogridsearch/autogridsearch_wrapper/get_next_param_grid/int/int_linspace_gap_gt_1_test.py

@@ -134,8 +134,8 @@ class TestIntLinspaceGap_GT_1:
                 del _gap
 
             elif _posn == 'middle':
-                assert min(_grid_out) == _SINGLE_GRID[_posn_ - 1] + 1
-                assert max(_grid_out) == _SINGLE_GRID[_posn_ + 1] - 1
+                assert min(_grid_out) >= _SINGLE_GRID[_posn_ - 1] + 1
+                assert max(_grid_out) <= _SINGLE_GRID[_posn_ + 1] - 1
 
 
         elif not _is_hard:
@@ -153,8 +153,8 @@ class TestIntLinspaceGap_GT_1:
                 del _gap
 
             elif _posn == 'middle':
-                assert min(_grid_out) == _SINGLE_GRID[_posn_ - 1] + 1
-                assert max(_grid_out) == _SINGLE_GRID[_posn_ + 1] - 1
+                assert min(_grid_out) >= _SINGLE_GRID[_posn_ - 1] + 1
+                assert max(_grid_out) <= _SINGLE_GRID[_posn_ + 1] - 1