Merge remote-tracking branch 'origin/main' into test-change-in-foreca…

…sting

* origin/main:
  [BUG] Fix pandas `FutureWarning` for silent upcasting (sktime#5395)
  [BUG] CLASP logic: remove indexes from exclusion zone that are out of range (sktime#5459)
  [MNT] Split CI jobs per components with specific soft-dependencies (sktime#5304)
  [MNT] new CI workflow to test extras (sktime#5375)
  [MNT] xfail remote data loaders to silence sporadic failures (sktime#5461)
  [BUG] [MNT] removed two recently added hooks (sktime#5453)
  [MNT] bound `pycatch22<0.4.4` due to breaking change in patch version (sktime#5434)

.all-contributorsrc

@@ -2423,7 +2423,7 @@
       "contributions": [
         "test"
       ]
-    }
+    },
     {
       "login": "xansh",
       "name": "Ansh Kumar",
@@ -2433,5 +2433,15 @@
         "doc"
       ]
     },
+    {
+      "login": "tpvasconcelos",
+      "name": "Tomas P. de Vasconcelos",
+      "avatar_url": "https://avatars.githubusercontent.com/u/17701527?v=4",
+      "profile": "https://github.com/tpvasconcelos",
+      "contributions": [
+        "bug",
+        "code"
+      ]
+    }
   ]
 }

.pre-commit-config.yaml

@@ -76,13 +76,3 @@ repos:
     hooks:
       - id: shellcheck
         name: shellcheck
-
-  - repo: https://github.com/tox-dev/pyproject-fmt
-    rev: 1.2.0
-    hooks:
-      - id: pyproject-fmt
-
-  - repo: https://github.com/abravalheri/validate-pyproject
-    rev: v0.15
-    hooks:
-      - id: validate-pyproject

pyproject.toml

@@ -78,7 +78,7 @@ all_extras = [
   'numba<0.59,>=0.53; python_version < "3.12"',
   'pmdarima!=1.8.1,<3.0.0,>=1.8; python_version < "3.12"',
   'prophet>=1.1; python_version < "3.12"',
-  "pycatch22",
+  "pycatch22<0.4.4",
   "pykalman-bardo<0.10,>=0.9.7",
   'pyod>=0.8; python_version < "3.11"',
   "scikit-optimize",
@@ -114,7 +114,7 @@ all_extras_pandas2 = [
   'numba<0.59,>=0.53; python_version < "3.12"',
   'pmdarima!=1.8.1,<3.0.0,>=1.8; python_version < "3.12"',
   'prophet>=1.1; python_version < "3.12"',
-  "pycatch22",
+  "pycatch22<0.4.4",
   "pykalman-bardo<0.10,>=0.9.7",
   'pyod>=0.8; python_version < "3.11"',
   "scikit_posthocs>=0.6.5",
@@ -228,7 +228,7 @@ transformations = [
   "holidays<0.35,>=0.29",
   "mne<1.6,>=1.5",
   'numba<0.59,>=0.53; python_version < "3.12"',
-  "pycatch22<0.5,>=0.4",
+  "pycatch22<0.4.4,>=0.4",
   "pykalman-bardo<0.10,>=0.9.7",
   "statsmodels<0.15,>=0.12.1",
   'stumpy<1.13,>=1.5.1; python_version < "3.12"',

sktime/datasets/tests/test_single_problem_loaders.py

@@ -81,6 +81,7 @@ def test_load_numpy2d_multivariate_raises(loader):
         X, y = loader(return_type="numpy2d")
 
 
+@pytest.mark.xfail(reason="known sporadic failure of unknown cause, see #5460")
 def test_load_UEA():
     """Test loading of a random subset of the UEA data, to check API."""
     from sktime.datasets.tsc_dataset_names import multivariate, univariate
@@ -114,6 +115,7 @@ def test_load_forecastingdata():
     assert metadata["contain_equal_length"] is False
 
 
+@pytest.mark.xfail(reason="known sporadic failure of unknown cause, see #5462")
 @pytest.mark.parametrize("name", TSF_SUBSAMPLE)
 def test_check_link_downloadable(name):
     """Test dataset URL from forecasting.org is downloadable and exits."""

sktime/transformations/series/_clasp_numba.py

@@ -161,7 +161,13 @@ def _calc_knn_labels(knn_mask, split_idx, m):
 
     # apply exclusion zone at split point
     exclusion_zone = np.arange(split_idx - m, split_idx)
-    y_pred[exclusion_zone] = np.ones(m, dtype=np.int64)
+
+    # Remove indexes outside the range of y_pred
+    exclusion_zone = exclusion_zone[
+        (exclusion_zone >= -len(y_pred)) & (exclusion_zone < len(y_pred))
+    ]
+
+    y_pred[exclusion_zone] = np.ones(len(exclusion_zone), dtype=np.int64)
 
     return y_true, y_pred
 

sktime/transformations/series/summarize.py

@@ -288,6 +288,7 @@ def _fit(self, X, y=None):
         lags = func_dict["summarizer"] == "lag"
         # Convert lags to default list notation with window_length 1
         boost_lag = func_dict.loc[lags, "window"].apply(lambda x: [int(x), 1])
+        func_dict.loc[:, "window"] = func_dict["window"].astype("object")
         func_dict.loc[lags, "window"] = boost_lag
         self.truncate_start = func_dict["window"].apply(lambda x: x[0] + x[1] - 1).max()
         self._func_dict = func_dict