Fix unit tests in all versions + update requirements

.github/workflows/auto-unit-test.yml

@@ -0,0 +1,54 @@
+name: Auto unit test
+
+on:
+  push:
+    branches: [master] # Only run push on default branch after merge
+  pull_request: # Run on all PRs
+
+# Cancel in-progress runs when a new commit is pushed
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  unit-tests:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - python-version: "3.6"
+            container: "python:3.6-buster"
+          - python-version: "3.7"
+            container: "python:3.7-buster"
+          - python-version: "3.8"
+          - python-version: "3.9"
+          - python-version: "3.10"
+          - python-version: "3.11"
+          - python-version: "3.12"
+          - python-version: "3.13"
+          - python-version: "3.14"
+
+    container: ${{ matrix.container }}
+
+    steps:
+      - uses: actions/checkout@v4
+      - name: Set up Python ${{ matrix.python-version }}
+        if: matrix.container == ''
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+          cache: 'pip'
+          cache-dependency-path: |
+            tests/python/unit/requirements.txt
+            code-env/python/spec/requirements.txt
+
+      - name: Install dependencies
+        run: |
+          pip install -r tests/python/unit/requirements.txt
+          pip install -r code-env/python/spec/requirements.txt
+
+      - name: Run tests
+        env:
+          PYTHONPATH: python-lib
+        run: pytest tests/python/unit -v

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Changelog
 
+## Version 1.3.5 (2026-01-29)
+- Add python 3.12, 3.13 and 3.14 official support
+
 ## Version 1.3.4 (2025-02-05)
 - Improve python 3.9 support (used to need local compilation of statsmodels, now uses a wheeled version)
 

Makefile

@@ -21,8 +21,8 @@ plugin:
 unit-tests:
 	@echo "Running unit tests..."
 	@( \
-		PYTHON_VERSION=`python3 -V 2>&1 | sed 's/[^0-9]*//g' | cut -c 1,2`; \
-		PYTHON_VERSION_IS_CORRECT=`cat code-env/python/desc.json | python3 -c "import sys, json; print(str($$PYTHON_VERSION) in [x[-2:] for x in json.load(sys.stdin)['acceptedPythonInterpreters']]);"`; \
+		PYTHON_VERSION=`python3 -c "import sys; print('%s%s' % (sys.version_info.major, sys.version_info.minor))"`; \
+		PYTHON_VERSION_IS_CORRECT=`cat code-env/python/desc.json | python3 -c "import sys, json; print('PYTHON' + str($$PYTHON_VERSION) in json.load(sys.stdin)['acceptedPythonInterpreters']);"`; \
 		if [ $$PYTHON_VERSION_IS_CORRECT == "False" ]; then echo "Python version $$PYTHON_VERSION is not in acceptedPythonInterpreters"; exit 1; else echo "Python version $$PYTHON_VERSION is in acceptedPythonInterpreters"; fi; \
 	)
 	@( \

code-env/python/desc.json

@@ -5,7 +5,10 @@
     "PYTHON38",
     "PYTHON39",
     "PYTHON310",
-    "PYTHON311"
+    "PYTHON311",
+    "PYTHON312",
+    "PYTHON313",
+    "PYTHON314"
   ],
   "corePackagesSet": "AUTO",
   "forceConda": false,

code-env/python/spec/requirements.txt

@@ -1,14 +1,27 @@
 graphviz==0.16
-matplotlib==3.3.4
+matplotlib==3.3.4; python_version <= '3.9'
+matplotlib>=3.5.0; python_version > '3.9'
+matplotlib-inline==0.1.6; python_version <= '3.9'
+matplotlib-inline>=0.1.7; python_version > '3.9'
 
 scikit-learn>=0.20,<1.1; python_version <= '3.10'
-scikit-learn==1.1.3; python_version >= '3.11'
+scikit-learn>=1.1.3,<=1.8.0; python_version == '3.11'
+scikit-learn==1.3.2; python_version == '3.12'
+scikit-learn>=1.5.0,<=1.8.0; python_version >= '3.13'
+
 scipy>=1.2,<1.3; python_version <= '3.7'
-scipy==1.10.1; python_version >= '3.8'
+scipy==1.10.1; python_version >= '3.8' and python_version < '3.12'
+scipy==1.11.3; python_version == '3.12'
+scipy>=1.13.0,<=1.17.0; python_version >= '3.13'
+
 xgboost==0.82
 lightgbm>=3.2,<3.3
+
 statsmodels>=0.10,<0.11; python_version < '3.9'
-statsmodels==0.13.5; python_version >= '3.9'
+statsmodels==0.13.5; python_version >= '3.9' and python_version < '3.12'
+statsmodels==0.14.0; python_version == '3.12'
+statsmodels>=0.14.2,<=0.14.6; python_version >= '3.13'
+
 jinja2>=2.10,<2.11
 flask>=1.0,<1.1
 cloudpickle>=1.3,<1.6

plugin.json

@@ -1,6 +1,6 @@
 {
   "id" : "model-error-analysis",
-  "version" : "1.3.4",
+  "version" : "1.3.5",
   "meta" : {
     "label" : "Model Error Analysis",
     "description" : "Debug model performance with error analysis. A code env is only required to use the Jupyter Notebook.",

python-lib/dku_error_analysis_decision_tree/tree.py

@@ -158,8 +158,10 @@ def get_stats_categorical_node(column, target_column, nr_bins, bins):
                 nr_bins = len(bins)
             target_grouped = target_column.groupby(column.fillna("No values").apply(safe_str))
             target_distrib = target_grouped.value_counts(dropna=False)
-            col_distrib = target_grouped.count().sort_values(ascending=False)
-            values = col_distrib.index if not bins else bins
+            col_distrib = target_grouped.count()
+            df = col_distrib.reset_index()
+            df.sort_values(by=[df.columns[1], df.columns[0]], ascending=[False, True], inplace=True)
+            values = df[df.columns[0]] if not bins else bins
 
             for value in values:
                 target_distrib_dict = target_distrib[value].to_dict()

python-lib/mealy_local/error_visualizer.py

@@ -45,7 +45,7 @@ def _plot_histograms(hist_data, label, **params):
     @staticmethod
     def _add_new_plot(figsize, bins, x_ticks, feature_name, suptitle):
         plt.figure(figsize=figsize)
-        plt.xticks(x_ticks, rotation="90")
+        plt.xticks(x_ticks, rotation=90)
         plt.gca().set_xticklabels(labels=bins)
         plt.ylabel('Proportion of samples')
         plt.title('Distribution of {}'.format(feature_name))

resource/py/test_tree_parser.py

@@ -53,7 +53,7 @@ def preproc_array():
 
 # PARSING METHODS
 @pytest.mark.parsing
-def test_create_tree(mocker, df, create_parser, caplog):
+def test_create_tree(mocker, df, create_parser, caplog, dss_target):
     error_model = mocker.Mock()
     error_model.feature_importances_ = np.array([1, 3, 0, 5, 2, 4])
     feature_names = [
@@ -106,7 +106,7 @@ def test_create_tree(mocker, df, create_parser, caplog):
     pd.testing.assert_series_equal(dataframe["vector [element #0]"],
         pd.Series(['e','a',np.nan,'e','i','e','e','i','i','i','i','i'], name="vector [element #0]"))
     pd.testing.assert_series_equal(dataframe["vector [element #1]"],
-        pd.Series([1, 0, np.nan, 0, 2, np.nan, np.nan, 1, 1, 2, 100, np.nan], name="vector [element #1]"))   
+        pd.Series([1, 0, np.nan, 0, 2, np.nan, np.nan, 1, 1, 2, 100, np.nan], name="vector [element #1]"))
     assert ranked_features[:3] == ["feat_c", "feat_a", "feat_b"]
     assert set(ranked_features) == {"feat_c", "feat_a", "feat_b", "vector [element #0]", "vector [element #1]", "cat_1", "num_1"}
     assert parser.num_features == {"num_1", "vector [element #1]"}
@@ -144,7 +144,7 @@ def mocked_get_split_param(feature):
         return TreeParser.SplitParameters(Node.TYPES.NUM, "foo", None, "num_1")
 
 @pytest.mark.parsing
-def test_build_tree(mocker, df, create_parser):
+def test_build_tree(mocker, df, create_parser, dss_target):
     mocker.patch("dku_error_analysis_tree_parsing.tree_parser.descale_numerical_thresholds",
                  return_value=[8, -2, .5, 3, -.5, -2, -2, -2, 1, -2, -2])
 
@@ -265,7 +265,7 @@ def check_dummy(split, name, value=None, others=False):
         assert split.invert_left_and_right and split.invert_left_and_right(0)
 
 @pytest.mark.categorical
-def test_dummy(create_parser, mocker):
+def test_dummy(create_parser, mocker, dss_target):
     parser = create_parser()
     step = mocker.Mock(values=["A", "B"], input_column_name="test", should_drop=True)
     parser._add_dummy_mapping(step)
@@ -298,7 +298,7 @@ def test_dummy(create_parser, mocker):
     check_dummy(others, "test", ["A", "B"], True)
 
 @pytest.mark.categorical
-def test_target_encoding(create_parser, mocker, preproc_array):
+def test_target_encoding(create_parser, mocker, preproc_array, dss_target):
     # Test classification
     parser = create_parser()
     step = mocker.Mock(column_name="test", encoding_name="enc_name")
@@ -339,7 +339,7 @@ def test_target_encoding(create_parser, mocker, preproc_array):
     assert (a.add_preprocessed_feature(preproc_array, 0) == [-1,0,0,0,1,0,4]).all()
 
 @pytest.mark.categorical
-def test_whole_cat_hashing(create_parser, mocker):
+def test_whole_cat_hashing(create_parser, mocker, dss_target):
     parser = create_parser()
     step = mocker.Mock(column_name="test", n_features=3)
     parser._add_cat_hashing_whole_mapping(step)
@@ -385,7 +385,7 @@ def test_whole_cat_hashing(create_parser, mocker):
     assert (third.add_preprocessed_feature(preproc_array, 2) == added_column).all()
 
 @pytest.mark.categorical
-def test_not_whole_cat_hashing(create_parser, mocker, preproc_array):
+def test_not_whole_cat_hashing(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(column_name="test", n_features=2)
     parser._add_cat_hashing_not_whole_mapping(step)
@@ -411,7 +411,7 @@ def test_not_whole_cat_hashing(create_parser, mocker, preproc_array):
         and not second.invert_left_and_right(.5)
     assert (second.add_preprocessed_feature(preproc_array, 1) == [0,2,3,0,1,0,0]).all()
 
-def test_frequency_encoding(create_parser, mocker, preproc_array):
+def test_frequency_encoding(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(column_name="test", suffix="suffix")
     parser._add_frequency_encoding_mapping(step)
@@ -427,7 +427,7 @@ def test_frequency_encoding(create_parser, mocker, preproc_array):
     assert (a.add_preprocessed_feature(preproc_array, 0) == [-1,0,0,0,1,0,4]).all()
 
 @pytest.mark.categorical
-def test_ordinal_encoding(create_parser, mocker, preproc_array):
+def test_ordinal_encoding(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(column_name="test", suffix="suffix")
     parser._add_ordinal_encoding_mapping(step)
@@ -444,7 +444,7 @@ def test_ordinal_encoding(create_parser, mocker, preproc_array):
 
 # VECTOR HANDLING
 @pytest.mark.vector
-def test_unfold(create_parser, mocker, preproc_array):
+def test_unfold(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(input_column_name="test", vector_length=2)
     parser._add_unfold_mapping(step)
@@ -473,7 +473,7 @@ def test_unfold(create_parser, mocker, preproc_array):
 
 # NUM HANDLINGS
 @pytest.mark.numerical
-def test_identity(create_parser, mocker, preproc_array):
+def test_identity(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     parser._add_identity_mapping("test")
     assert len(parser.preprocessed_feature_mapping) == 1
@@ -490,7 +490,7 @@ def test_identity(create_parser, mocker, preproc_array):
     assert (split.add_preprocessed_feature(preproc_array, 0) == [-1,0,0,0,1,0,4]).all()
 
 @pytest.mark.numerical
-def test_binarize(create_parser, mocker, preproc_array):
+def test_binarize(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(in_col="test", threshold=42)
     step._output_name.return_value = "output"
@@ -509,7 +509,7 @@ def test_binarize(create_parser, mocker, preproc_array):
     assert (split.add_preprocessed_feature(preproc_array, 0) == [-1,0,0,0,1,0,4]).all()
 
 @pytest.mark.numerical
-def test_quantize(create_parser, mocker, preproc_array):
+def test_quantize(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(in_col="test", nb_bins=42, r={"bounds": ["0.5", "1.6", "7.8"]})
     parser._add_quantize_mapping(step)
@@ -527,7 +527,7 @@ def test_quantize(create_parser, mocker, preproc_array):
     assert (split.add_preprocessed_feature(preproc_array, 0) == [-1,0,0,0,1,0,4]).all()
 
 @pytest.mark.numerical
-def test_flag_missing(create_parser, mocker):
+def test_flag_missing(create_parser, mocker, dss_target):
     # Flag on numerical feature
     parser = create_parser()
     step = mocker.Mock(feature="test", output_block_name="num_flagonly")
@@ -562,7 +562,7 @@ def test_flag_missing(create_parser, mocker):
         and not split.invert_left_and_right(.5)
 
 @pytest.mark.numerical
-def test_datetime_encoding(create_parser, mocker, preproc_array):
+def test_datetime_encoding(create_parser, mocker, preproc_array, dss_target):
     parser = create_parser()
     step = mocker.Mock(column_name="test", selected_periods=["p1", "p2"])
 
@@ -623,7 +623,7 @@ def check_text_features(preproc_array, split, name):
     assert (split.add_preprocessed_feature(preproc_array, 1) == [0,2,3,0,1,0,0]).all()
 
 @pytest.mark.text
-def test_vect_hashing(create_parser, mocker, caplog, preproc_array):
+def test_vect_hashing(create_parser, mocker, caplog, preproc_array, dss_target):
     caplog.set_level(logging.INFO)
     # Hash without SVD
     parser = create_parser()
@@ -653,12 +653,13 @@ def test_vect_hashing(create_parser, mocker, caplog, preproc_array):
     assert log.msg == "Feature test_bis is a text feature. Its distribution plot will not be available"
 
 @pytest.mark.text
-def test_count_vect(create_parser, mocker, caplog, preproc_array):
+def test_count_vect(create_parser, mocker, caplog, preproc_array, dss_target):
     caplog.set_level(logging.INFO)
     parser = create_parser()
     step = mocker.Mock(column_name="test", prefix="prefix")
     vectorizer = mocker.Mock()
     vectorizer.get_feature_names.return_value = ["word", "random"]
+    vectorizer.get_feature_names_out.return_value = ["word", "random"]
     step.resource = {"vectorizer": vectorizer}
     parser._add_text_count_vect_mapping(step)
     assert len(parser.preprocessed_feature_mapping) == 2
@@ -673,12 +674,13 @@ def test_count_vect(create_parser, mocker, caplog, preproc_array):
     assert log.msg == "Feature test is a text feature. Its distribution plot will not be available"
 
 @pytest.mark.text
-def test_tfidf_vect(create_parser, mocker, caplog, preproc_array):
+def test_tfidf_vect(create_parser, mocker, caplog, preproc_array, dss_target):
     caplog.set_level(logging.INFO)
     parser = create_parser()
     step = mocker.Mock(column_name="test")
     vectorizer = mocker.Mock(idf_=[42.4242])
     vectorizer.get_feature_names.return_value = ["word", "random"]
+    vectorizer.get_feature_names_out.return_value = ["word", "random"]
     step.resource = {"vectorizer": vectorizer}
     parser._add_tfidf_vect_mapping(step)
     assert len(parser.preprocessed_feature_mapping) == 1
@@ -693,6 +695,7 @@ def test_tfidf_vect(create_parser, mocker, caplog, preproc_array):
     step = mocker.Mock(column_name="test_bis")
     vectorizer = mocker.Mock(idf_=[42.4242, 1])
     vectorizer.get_feature_names.return_value = ["word", "random"]
+    vectorizer.get_feature_names_out.return_value = ["word", "random"]
     step.resource = {"vectorizer": vectorizer}
     parser._add_tfidf_vect_mapping(step)
     assert len(parser.preprocessed_feature_mapping) == 2

tests/python/integration/requirements.txt

@@ -1,3 +1,4 @@
-pytest~=6.2
+pytest~=6.2; python_version < '3.7'
+pytest>=7.4,<10.0; python_version >= '3.7'
 dataiku-api-client
-git+git://github.com/dataiku/dataiku-plugin-tests-utils.git@master#egg=dataiku-plugin-tests-utils
+git+https://github.com/dataiku/dataiku-plugin-tests-utils.git@master#egg=dataiku-plugin-tests-utils

tests/python/unit/requirements.txt

@@ -1,5 +1,6 @@
-pandas~=1.0
-pytest~=6.2
-allure-pytest==2.8.29
-numpy==1.16.6
-pytest-mock~=3.6
+pandas~=1.0; python_version < '3.12'
+pandas>=2.0,<3.0; python_version >= '3.12'
+pytest>=6.2,<10.0
+allure-pytest>=2.8.29,<3.0
+numpy>=1.16.6
+pytest-mock>=3.6,<4.0