update version

Author: sdpython

HISTORY.rst

@@ -5,26 +5,34 @@
 History
 =======
 
-current - 2022-07-22 - 0.00Mb
+current - 2023-07-03 - 0.00Mb
+=============================
+
+* #115: Updates tree decision criterion for scikit-learn 1.2 (2023-07-02)
+* #113: Removes normalize attributes (deprecated) (2022-11-29)
+* #110: Fixes perplexity issue with PredictableTSNE (2022-08-06)
+* #109: Use f strings in more places (2022-07-22)
+
+0.3.649 - 2022-07-22 - 2.35Mb
 =============================
 
 * #105: Update for python 3.10 (2022-07-22)
 * #108: Uses f strings (2022-07-19)
 
-0.3.631 - 2022-05-19 - 0.52Mb
+0.3.631 - 2022-05-19 - 2.21Mb
 =============================
 
 * #107: Updates CI for scikit-learn==1.1 (2022-05-18)
 * #106: Fixes failing import _joblib_parallel_args (2022-02-18)
 * #99: LICENSE file missing in PyPI release (2021-11-20)
 
-0.3.614 - 2021-10-02 - 0.52Mb
+0.3.614 - 2021-10-02 - 1.73Mb
 =============================
 
 * #103: Updates for scikit-learn>=1.0 (2021-10-02)
 * #94: Fixed Numpy boolean array indexing issue for 2dim arrays. (2021-09-27)
 
-0.3.606 - 2021-08-22 - 0.52Mb
+0.3.606 - 2021-08-22 - 2.35Mb
 =============================
 
 * #102: Implements numpy.digitalize with a DecisionTreeRegressor (2021-08-22)
@@ -37,32 +45,16 @@ current - 2022-07-22 - 0.00Mb
 * #98: Fixes #97, fix issue with deepcopy and criterion (2021-05-03)
 * #97: piecewise_decision_tree does not compile with the latest version of scikit-learn (2021-05-03)
 * #85: Fixes #70, implements DecisionTreeLogisticRegression (2021-05-02)
-
-0.3.549 - 2021-01-09 - 0.67Mb
-=============================
-
 * #93: Include build wheel for all platforms in CI (2021-01-09)
-
-0.3.543 - 2021-01-03 - 0.67Mb
-=============================
-
 * #89: Install fails: ModuleNotFoundError: No module named 'sklearn' (2021-01-03)
 * #92: QuantileMLPRegressor does not work with scikit-learn 0.24 (2021-01-01)
 * #91: Fixes regression criterion for scikit-learn 0.24 (2021-01-01)
 * #90: Fixes PipelineCache for scikit-learn 0.24 (2021-01-01)
-
-0.2.508 - 2020-09-02 - 0.43Mb
-=============================
-
 * #88: Change for scikit-learn 0.24 (2020-09-02)
 * #87: Set up CI with Azure Pipelines (2020-09-02)
 * #86: Update CI, use python 3.8 (2020-09-02)
 * #71: update kmeans l1 to the latest kmeans (signatures changed) (2020-08-31)
 * #84: style (2020-08-30)
-
-0.2.491 - 2020-08-06 - 0.83Mb
-=============================
-
 * #83: Upgrade version (2020-08-06)
 * #82: Fixes #81, skl 0.22, 0.23 together (2020-08-06)
 * #81: Make mlinsights work with scikit-learn 0.22 and 0.23 (2020-08-06)

MANIFEST.in

@@ -1,7 +1,10 @@
+recursive-include onnx_extended *.c *.cpp *.h *.pyx *.pxd *.pxi *.py
+include pyproject.toml
+include MANIFEST.in
+include setup.cfg
 prune _doc
 prune _unittests
 prune bin
-prune .circleci
 exclude *.bat
 exclude *.yml
 exclude *.git*

_unittests/ut_mlmodel/test_categories_to_integers.py

@@ -105,7 +105,8 @@ def test_categories_to_integers_pickle(self):
         data = os.path.join(os.path.abspath(
             os.path.dirname(__file__)), "data", "adult_set.txt")
         df = pandas.read_csv(data, sep="\t")
-        run_test_sklearn_pickle(lambda: CategoriesToIntegers(skip_errors=True), df)
+        run_test_sklearn_pickle(
+            lambda: CategoriesToIntegers(skip_errors=True), df)
 
     @ignore_warnings(skipped_warnings)
     def test_categories_to_integers_clone(self):

mlinsights/__init__.py

@@ -4,7 +4,7 @@
 @brief Module *mlinsights*.
 Look for insights for machine learned models.
 """
-__version__ = "0.4.649"
+__version__ = "0.4.664"
 __author__ = "Xavier Dupré"
 __github__ = "https://github.com/sdpython/mlinsights"
 __url__ = "http://www.xavierdupre.fr/app/mlinsights/helpsphinx/index.html"

mlinsights/mlmodel/piecewise_tree_regression.py

@@ -132,7 +132,8 @@ def _fit_reglin(self, X, y, sample_weight):
             if len(ys.shape) == 1:
                 ys = ys[:, numpy.newaxis]
             ys = ys.copy()
-            ws = sample_weight[ind].copy() if sample_weight is not None else None
+            ws = sample_weight[ind].copy(
+            ) if sample_weight is not None else None
             dec = LinearRegressorCriterion.create(xs, ys, ws)
             dec.node_beta(self.betas_[i, :])
 

mlinsights/mltree/tree_digitize.py

@@ -84,7 +84,8 @@ def add_root(index):
         is_left = False
         is_leaf = False
         threshold = bins[index]
-        n = tree_add_node(tree, parent, is_left, is_leaf, 0, threshold, 0, 1, 1., 0)
+        n = tree_add_node(tree, parent, is_left, is_leaf,
+                          0, threshold, 0, 1, 1., 0)
         values.append(UNUSED)
         n_nodes.append(n)
         return n
@@ -95,15 +96,17 @@ def add_nodes(parent, i, j, is_left):
             # it means j is the parent split
             if i == j:
                 # leaf
-                n = tree_add_node(tree, parent, is_left, True, 0, 0, 0, 1, 1., 0)
+                n = tree_add_node(tree, parent, is_left,
+                                  True, 0, 0, 0, 1, 1., 0)
                 n_nodes.append(n)
                 values.append(i)
                 return n
             if i + 1 == j:
                 # split
                 values.append(UNUSED)
                 th = bins[i]
-                n = tree_add_node(tree, parent, is_left, False, 0, th, 0, 1, 1., 0)
+                n = tree_add_node(tree, parent, is_left,
+                                  False, 0, th, 0, 1, 1., 0)
                 n_nodes.append(n)
                 add_nodes(n, i, i, True)
                 add_nodes(n, i, j, False)
@@ -113,7 +116,8 @@ def add_nodes(parent, i, j, is_left):
                 values.append(UNUSED)
                 index = (i + j) // 2
                 th = bins[index]
-                n = tree_add_node(tree, parent, is_left, False, 0, th, 0, 1, 1., 0)
+                n = tree_add_node(tree, parent, is_left,
+                                  False, 0, th, 0, 1, 1., 0)
                 n_nodes.append(n)
                 add_nodes(n, i, index, True)
                 add_nodes(n, index, j, False)
@@ -123,15 +127,17 @@ def add_nodes(parent, i, j, is_left):
             if i + 1 == j:
                 # leaf
                 values.append(j)
-                n = tree_add_node(tree, parent, is_left, True, 0, 0, 0, 1, 1., 0)
+                n = tree_add_node(tree, parent, is_left,
+                                  True, 0, 0, 0, 1, 1., 0)
                 n_nodes.append(n)
                 return n
             if i + 1 < j:
                 # split
                 values.append(UNUSED)
                 index = (i + j) // 2
                 th = bins[index]
-                n = tree_add_node(tree, parent, is_left, False, 0, th, 0, 1, 1., 0)
+                n = tree_add_node(tree, parent, is_left,
+                                  False, 0, th, 0, 1, 1., 0)
                 n_nodes.append(n)
                 add_nodes(n, i, index, True)
                 add_nodes(n, index, j, False)

mlinsights/plotting/visualize.py

@@ -73,12 +73,12 @@ def _get_name_simple(name, data):
 
             info = _pipeline_info(
                 model, new_data, context, former_data=new_data)
-            #new_outputs = []
+            # new_outputs = []
             # for o in info[-1]['outputs']:
             #    add = _get_name(context, prefix=o, info=info)
             #    outputs.append(add)
             #    new_outputs.append(add)
-            #info[-1]['outputs'] = new_outputs
+            # info[-1]['outputs'] = new_outputs
             outputs.extend(info[-1]['outputs'])
             infos.extend(info)