Make EvalML compatible with Woodwork changes (#4066)

* update file * update release notes * lint * update latest dep * update min * update test * revert woodwork versions * set ww min to 0.22.0 * empty * merge target leakage update * update wodwork for meta.yaml * remove print * lint * Update timeseries.ipynb Reformat doc * update release notes * retrigger codecov * lint * trigger --------- Co-authored-by: bchen1116 <[email protected]>
alteryx · Mar 14, 2023 · 1412fc3 · 1412fc3
1 parent 2401d30
commit 1412fc3
Show file tree

Hide file tree

Showing 14 changed files with 70 additions and 37 deletions.
diff --git a/.github/meta.yaml b/.github/meta.yaml
@@ -36,7 +36,7 @@ outputs:
         - click>=8.0.0
         - shap >=0.40.0
         - texttable >=1.6.2
-        - woodwork >=0.21.1
+        - woodwork >=0.22.0
         - featuretools>=1.16.0
         - nlp-primitives>=2.9.0
         - python >=3.8.*

diff --git a/docs/source/release_notes.rst b/docs/source/release_notes.rst
@@ -23,6 +23,8 @@ Release Notes
     * Testing Changes
         * Use ``release.yaml`` for performance tests on merge to main :pr:`4007`
         * Pin ``github-action-check-linked-issues`` at v1.4.5 :pr:`4042`
+        * Updated tests to support Woodwork's object dtype inference for numeric columns :pr:`4066`
+        * Updated ``TargetLeakageDataCheck`` tests to handle boolean targets properly :pr:`4066`
 
 .. warning::
 

diff --git a/docs/source/user_guide/timeseries.ipynb b/docs/source/user_guide/timeseries.ipynb
@@ -131,6 +131,7 @@
    "outputs": [],
    "source": [
     "X[\"Categorical\"] = [str(i % 4) for i in range(len(X))]\n",
+    "X[\"Categorical\"] = X[\"Categorical\"].astype(\"category\")\n",
     "X[\"Numeric\"] = [i for i in range(len(X))]\n",
     "\n",
     "# Re-split the data since we modified X\n",

diff --git a/evalml/pipelines/classification_pipeline.py b/evalml/pipelines/classification_pipeline.py
@@ -1,7 +1,6 @@
 """Pipeline subclass for all classification pipelines."""
 import numpy as np
 import pandas as pd
-import woodwork as ww
 
 from evalml.pipelines import PipelineBase
 from evalml.problem_types import is_binary, is_multiclass
@@ -71,7 +70,7 @@ def fit(self, X, y):
 
         # TODO: Added this in because numpy's unique() does not support pandas.NA
         try:
-            self._classes_ = list(ww.init_series(np.unique(y)))
+            self._classes_ = list(np.unique(y))
         except TypeError as e:
             if "boolean value of NA is ambiguous" in str(e):
                 self._classes_ = y.unique()

diff --git a/evalml/tests/component_tests/test_column_selector_transformers.py b/evalml/tests/component_tests/test_column_selector_transformers.py
@@ -71,7 +71,7 @@ def test_select_by_type_empty_X():
             [
                 lambda X, X_t: X_t.empty,
                 lambda X, X_t: X_t.empty,
-                lambda X, X_t: X_t.equals(X[["three"]].astype("int64")),
+                lambda X, X_t: X_t.equals(X[["one", "three"]].astype("int64")),
                 lambda X, X_t: X_t.astype(str).equals(X.astype(str)),
             ],
         ),
@@ -135,7 +135,7 @@ def test_column_transformer_transform(class_to_test, checking_functions):
             SelectByType,
             [
                 lambda X, X_t: X_t.empty,
-                lambda X, X_t: X_t.equals(X[["three"]].astype("int64")),
+                lambda X, X_t: X_t.equals(X[["one", "three"]].astype("int64")),
                 lambda X, X_t: X_t.astype(str).equals(X.astype(str)),
             ],
         ),

diff --git a/evalml/tests/component_tests/test_target_encoder.py b/evalml/tests/component_tests/test_target_encoder.py
@@ -121,29 +121,32 @@ def test_cols():
         {
             "col_1": [1, 2, 1, 1, 2] * 2,
             "col_2": ["2", "1", "1", "1", "1"] * 2,
-            "col_3": ["a", "a", "a", "a", "a"] * 2,
+            "col_3": ["a", "a", "a", "a", "b"] * 2,
         },
     )
-    X_expected = X.astype({"col_1": "int64", "col_2": "category", "col_3": "category"})
+    X_expected = X.astype({"col_1": "int64", "col_2": "int64", "col_3": "category"})
     y = pd.Series([0, 1, 1, 1, 0] * 2)
     encoder = TargetEncoder(cols=[])
     encoder.fit(X, y)
     X_t = encoder.transform(X)
     assert_frame_equal(X_expected, X_t)
 
-    encoder = TargetEncoder(cols=["col_2"])
+    encoder = TargetEncoder(cols=["col_3"])
     encoder.fit(X, y)
     X_t = encoder.transform(X)
     X_expected = pd.DataFrame(
         {
             "col_1": pd.Series([1, 2, 1, 1, 2] * 2, dtype="int64"),
-            "col_2": [0.161365, 0.749863, 0.749863, 0.749863, 0.749863] * 2,
-            "col_3": pd.Series(["a", "a", "a", "a", "a"] * 2, dtype="category"),
+            "col_2": [2, 1, 1, 1, 1] * 2,
+            "col_3": pd.Series(
+                [0.749863, 0.749863, 0.749863, 0.749863, 0.161365] * 2,
+                dtype="float64",
+            ),
         },
     )
     assert_frame_equal(X_expected, X_t, check_less_precise=True)
 
-    encoder = TargetEncoder(cols=["col_2", "col_3"])
+    encoder = TargetEncoder(cols=["col_3"])
     encoder.fit(X, y)
     X_t = encoder.transform(X)
     encoder2 = TargetEncoder()

diff --git a/evalml/tests/data_checks_tests/test_id_columns_data_check.py b/evalml/tests/data_checks_tests/test_id_columns_data_check.py
@@ -127,15 +127,15 @@ def test_id_columns_strings():
     id_cols_check = IDColumnsDataCheck(id_threshold=1.0)
     assert id_cols_check.validate(X) == [
         DataCheckWarning(
-            message="Columns 'Id' are 100.0% or more likely to be an ID column",
+            message="Columns 'Id', 'col_3_id' are 100.0% or more likely to be an ID column",
             data_check_name=id_data_check_name,
             message_code=DataCheckMessageCode.HAS_ID_COLUMN,
-            details={"columns": ["Id"]},
+            details={"columns": ["Id", "col_3_id"]},
             action_options=[
                 DataCheckActionOption(
                     DataCheckActionCode.DROP_COL,
                     data_check_name=id_data_check_name,
-                    metadata={"columns": ["Id"]},
+                    metadata={"columns": ["Id", "col_3_id"]},
                 ),
             ],
         ).to_dict(),
@@ -293,17 +293,23 @@ def test_unidentified_first_col_primary_key(
     )
 
     id_cols_check = IDColumnsDataCheck(id_threshold=0.95)
+    if input_type == "string":
+        order = ["col_2", "col_3_id", "col_1_id"]
+    else:
+        order = ["col_2", "col_1_id", "col_3_id"]
+    order_msg = f"Columns '{order[0]}', '{order[1]}', '{order[2]}' are 95.0% or more likely to be an ID column"
+
     assert id_cols_check.validate(X) == [
         DataCheckWarning(
-            message="Columns 'col_2', 'col_1_id', 'col_3_id' are 95.0% or more likely to be an ID column",
+            message=order_msg,
             data_check_name=id_data_check_name,
             message_code=DataCheckMessageCode.HAS_ID_COLUMN,
-            details={"columns": ["col_2", "col_1_id", "col_3_id"]},
+            details={"columns": order},
             action_options=[
                 DataCheckActionOption(
                     DataCheckActionCode.DROP_COL,
                     data_check_name=id_data_check_name,
-                    metadata={"columns": ["col_2", "col_1_id", "col_3_id"]},
+                    metadata={"columns": order},
                 ),
             ],
         ).to_dict(),

diff --git a/evalml/tests/data_checks_tests/test_target_leakage_data_check.py b/evalml/tests/data_checks_tests/test_target_leakage_data_check.py
@@ -193,15 +193,15 @@ def test_target_leakage_types():
 
     expected = [
         DataCheckWarning(
-            message="Columns 'a', 'b' are 80.0% or more correlated with the target",
+            message="Columns 'a', 'b', 'c' are 80.0% or more correlated with the target",
             data_check_name=target_leakage_data_check_name,
             message_code=DataCheckMessageCode.TARGET_LEAKAGE,
-            details={"columns": ["a", "b"]},
+            details={"columns": ["a", "b", "c"]},
             action_options=[
                 DataCheckActionOption(
                     DataCheckActionCode.DROP_COL,
                     data_check_name=target_leakage_data_check_name,
-                    metadata={"columns": ["a", "b"]},
+                    metadata={"columns": ["a", "b", "c"]},
                 ),
             ],
         ).to_dict(),
@@ -356,8 +356,21 @@ def test_target_leakage_data_check_warnings_pearson():
     y = y.astype(bool)
 
     leakage_check = TargetLeakageDataCheck(pct_corr_threshold=0.5, method="pearson")
-    # pearsons does not support boolean columns
-    assert leakage_check.validate(X, y) == []
+    assert leakage_check.validate(X, y) == [
+        DataCheckWarning(
+            message="Columns 'a', 'b', 'c', 'd' are 50.0% or more correlated with the target",
+            data_check_name=target_leakage_data_check_name,
+            message_code=DataCheckMessageCode.TARGET_LEAKAGE,
+            details={"columns": ["a", "b", "c", "d"]},
+            action_options=[
+                DataCheckActionOption(
+                    DataCheckActionCode.DROP_COL,
+                    data_check_name=target_leakage_data_check_name,
+                    metadata={"columns": ["a", "b", "c", "d"]},
+                ),
+            ],
+        ).to_dict(),
+    ]
 
     y = y.astype(int)
     assert leakage_check.validate(X, y) == [
@@ -447,9 +460,6 @@ def test_target_leakage_none_measures(measures):
     X["b"] = y
     y = y.astype(bool)
 
-    if measures in ["pearson", "spearman"]:
-        assert leakage_check.validate(X, y) == []
-        return
     assert len(leakage_check.validate(X, y))
 
 

diff --git a/evalml/tests/dependency_update_check/latest_dependency_versions.txt b/evalml/tests/dependency_update_check/latest_dependency_versions.txt
@@ -33,5 +33,5 @@ statsmodels==0.13.5
 texttable==1.6.7
 tomli==2.0.1
 vowpalwabbit==9.7.0
-woodwork==0.21.2
+woodwork==0.22.0
 xgboost==1.7.4
diff --git a/evalml/tests/dependency_update_check/minimum_requirements.txt b/evalml/tests/dependency_update_check/minimum_requirements.txt
@@ -32,5 +32,5 @@ statsmodels==0.12.2
 texttable==1.6.2
 tomli==2.0.1
 vowpalwabbit==8.11.0
-woodwork==0.21.1
+woodwork==0.22.0
 xgboost==1.7.0
diff --git a/evalml/tests/dependency_update_check/minimum_test_requirements.txt b/evalml/tests/dependency_update_check/minimum_test_requirements.txt
@@ -41,5 +41,5 @@ statsmodels==0.12.2
 texttable==1.6.2
 tomli==2.0.1
 vowpalwabbit==8.11.0
-woodwork==0.21.1
+woodwork==0.22.0
 xgboost==1.7.0
diff --git a/evalml/tests/model_understanding_tests/prediction_explanations_tests/test_explainers.py b/evalml/tests/model_understanding_tests/prediction_explanations_tests/test_explainers.py
@@ -1132,11 +1132,11 @@ def test_json_serialization(
         pipeline = linear_regression_pipeline
     elif problem_type == problem_type.BINARY:
         X, y = X_y_binary
-        y = pd.Series(y).astype("str")
+        y = pd.Series(y).astype("string")
         pipeline = logistic_regression_binary_pipeline
     else:
         X, y = X_y_multi
-        y = pd.Series(y).astype("str")
+        y = pd.Series(y).astype("string")
         pipeline = logistic_regression_multiclass_pipeline
 
     pipeline.fit(X, y)
@@ -1148,6 +1148,7 @@ def test_json_serialization(
         num_to_explain=1,
         output_format="dict",
     )
+
     assert json.loads(json.dumps(best_worst)) == best_worst
 
     report = explain_predictions(

diff --git a/evalml/tests/model_understanding_tests/test_partial_dependence.py b/evalml/tests/model_understanding_tests/test_partial_dependence.py
@@ -1425,10 +1425,14 @@ def test_graph_partial_dependence_regression_and_binary_categorical(
 
     X = pd.DataFrame(X)
     X.columns = [str(i) for i in range(X.shape[1])]
-    X["categorical_column"] = pd.Series([i % 3 for i in range(X.shape[0])]).astype(
+    X["categorical_column"] = pd.Series(
+        [f"cat_{i % 3}" for i in range(X.shape[0])],
+    ).astype(
         "str",
     )
-    X["categorical_column_2"] = pd.Series([i % 6 for i in range(X.shape[0])]).astype(
+    X["categorical_column_2"] = pd.Series(
+        [f"cat_{i % 6}" for i in range(X.shape[0])],
+    ).astype(
         "str",
     )
 
@@ -1442,7 +1446,7 @@ def test_graph_partial_dependence_regression_and_binary_categorical(
     )
     plot_data = fig.to_dict()["data"][0]
     assert plot_data["type"] == "bar"
-    assert list(plot_data["x"]) == ["0", "1", "2"]
+    assert list(plot_data["x"]) == ["cat_0", "cat_1", "cat_2"]
 
     fig = graph_partial_dependence(
         pipeline,
@@ -1453,7 +1457,7 @@ def test_graph_partial_dependence_regression_and_binary_categorical(
     fig_dict = fig.to_dict()
     plot_data = fig_dict["data"][0]
     assert plot_data["type"] == "contour"
-    assert fig_dict["layout"]["yaxis"]["ticktext"] == ["0", "1", "2"]
+    assert fig_dict["layout"]["yaxis"]["ticktext"] == ["cat_0", "cat_1", "cat_2"]
     assert (
         fig_dict["layout"]["title"]["text"]
         == "Partial Dependence of 'categorical_column' vs. '0'"
@@ -1468,8 +1472,15 @@ def test_graph_partial_dependence_regression_and_binary_categorical(
     fig_dict = fig.to_dict()
     plot_data = fig_dict["data"][0]
     assert plot_data["type"] == "contour"
-    assert fig_dict["layout"]["xaxis"]["ticktext"] == ["0", "1", "2"]
-    assert fig_dict["layout"]["yaxis"]["ticktext"] == ["0", "1", "2", "3", "4", "5"]
+    assert fig_dict["layout"]["xaxis"]["ticktext"] == ["cat_0", "cat_1", "cat_2"]
+    assert fig_dict["layout"]["yaxis"]["ticktext"] == [
+        "cat_0",
+        "cat_1",
+        "cat_2",
+        "cat_3",
+        "cat_4",
+        "cat_5",
+    ]
     assert (
         fig_dict["layout"]["title"]["text"]
         == "Partial Dependence of 'categorical_column_2' vs. 'categorical_column'"

diff --git a/pyproject.toml b/pyproject.toml
@@ -41,7 +41,7 @@ dependencies = [
     "shap >= 0.40.0",
     "statsmodels >= 0.12.2",
     "texttable >= 1.6.2",
-    "woodwork >= 0.21.1",
+    "woodwork >= 0.22.0",
     "dask >= 2022.2.0, != 2022.10.1",
     "featuretools >= 1.16.0",
     "nlp-primitives >= 2.9.0",