Merge branch 'main' into warnings-config

sktime · May 18, 2023 · 932c1d6 · 932c1d6
2 parents bf0adbc + 192a0d8
commit 932c1d6
Show file tree

Hide file tree

Showing 58 changed files with 1,956 additions and 521 deletions.
diff --git a/.all-contributorsrc b/.all-contributorsrc
@@ -2156,6 +2156,15 @@
       "contributions": [
         "doc"
       ]
-    }
+    },
+    {
+      "login": "marrov",
+      "name": "Marc Rovira",
+      "avatar_url": "https://avatars.githubusercontent.com/u/54272586?v=4",
+      "profile": "https://github.com/marrov",
+      "contributions": [
+        "doc"
+      ]
+    }    
   ]
 }
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -44,6 +44,26 @@ jobs:
         run: build_tools/run_examples.sh
         shell: bash
 
+  test-nodevdeps:
+    needs: code-quality
+    runs-on: ubuntu-20.04
+    steps:
+      - uses: actions/checkout@v3
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: 3.9
+      - name: Display Python version
+        run: python -c "import sys; print(sys.version)"
+
+      - name: Install sktime and dependencies
+        run: |
+          python -m pip install .
+
+      - name: Run pytest-free tests
+        run: |
+          python sktime/_nopytest_tests.py
+
   test-nosoftdeps:
     needs: code-quality
     runs-on: ubuntu-20.04

diff --git a/CONTRIBUTORS.md b/CONTRIBUTORS.md
diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-<a href="https://www.sktime.net"><img src="https://github.com/sktime/sktime/blob/main/docs/source/images/sktime-logo.jpg?raw=true)" width="175" align="right" /></a>
+<a href="https://www.sktime.net"><img src="https://github.com/sktime/sktime/blob/main/docs/source/images/sktime-logo.svg" width="175" align="right" /></a>
 
 # Welcome to sktime
 

diff --git a/docs/source/about/artwork.rst b/docs/source/about/artwork.rst
@@ -8,5 +8,5 @@ High-quality logos are available in the `docs/source/images/ <https://github.com
 
 The logos were designed by `Miguel Garcia Cabrera <https://www.flickr.com/photos/miguelgarciacabrera/>`_.
 
-.. image:: ../images/sktime-logo-no-text.jpg
+.. image:: ../images/sktime-logo-no-text.png
     :align: center
diff --git a/docs/source/api_reference/classification.rst b/docs/source/api_reference/classification.rst
@@ -20,8 +20,19 @@ Composition
 
     ClassifierPipeline
     ColumnEnsembleClassifier
-    ComposableTimeSeriesForestClassifier
     SklearnClassifierPipeline
+
+Ensembles
+---------
+
+.. currentmodule:: sktime.classification.ensemble
+
+.. autosummary::
+    :toctree: auto_generated/
+    :template: class.rst
+
+    BaggingClassifier
+    ComposableTimeSeriesForestClassifier
     WeightedEnsembleClassifier
 
 Deep learning

diff --git a/docs/source/api_reference/dists_kernels.rst b/docs/source/api_reference/dists_kernels.rst
@@ -117,6 +117,8 @@ Dynamic Time Warping Distances
 Time warping distances can also be obtained by composing ``DistFromAligner`` with
 a time warping aligner, see docstring of ``DistFromAligner``:
 
+.. currentmodule:: sktime.dists_kernels.compose_from_align
+
 .. autosummary::
     :toctree: auto_generated/
     :template: class.rst

diff --git a/docs/source/api_reference/forecasting.rst b/docs/source/api_reference/forecasting.rst
@@ -77,6 +77,7 @@ Use ``make_reduction`` for easy specification.
     RecursiveTimeSeriesRegressionForecaster
     DirRecTabularRegressionForecaster
     DirRecTimeSeriesRegressionForecaster
+    YfromX
 
 Naive forecasters
 -----------------

diff --git a/docs/source/conf.py b/docs/source/conf.py
@@ -82,7 +82,7 @@
 #
 # This is also used if you do content translation via gettext catalogs.
 # Usually you set "language" from the command line for these cases.
-language = None
+language = "en"
 
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.

diff --git a/docs/source/images/sktime-logo.png b/docs/source/images/sktime-logo.png
diff --git a/docs/source/images/sktime-logo.svg b/docs/source/images/sktime-logo.svg
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -6,6 +6,12 @@ Welcome to sktime
 
 A unified framework for machine learning with time series.
 
+.. topic:: Applications for sktime internships 2023 open!
+
+    Application deadline is May 19.
+    `Apply here <https://github.com/sktime/mentoring/blob/main/internships/sktime2023.md>`_
+
+
 Mission
 =======
 

diff --git a/examples/minirocket.ipynb b/examples/minirocket.ipynb
@@ -69,7 +69,7 @@
    "source": [
     "### 1.2 Load the Training Data\n",
     "\n",
-    "For more details on the data set, see the [univariate time series classification notebook](https://github.com/sktime/sktime/blob/main/examples/02_classification_univariate.ipynb).\n",
+    "For more details on the data set, see the [univariate time series classification notebook](https://github.com/sktime/sktime/blob/main/examples/02_classification.ipynb).\n",
     "\n",
     "**Note**: Input time series must be *at least* of length 9.  Pad shorter time series using, e.g., `PaddingTransformer` (`sktime.transformers.panel.padder`)."
    ]

diff --git a/pyproject.toml b/pyproject.toml
@@ -47,6 +47,7 @@ dependencies = [
     "deprecated>=1.2.13",
     "numpy>=1.21.0,<1.25",
     "pandas>=1.1.0,<2.0.0",
+    "packaging",
     "scikit-base<0.5.0",
     "scikit-learn>=0.24.0,<1.3.0",
     "scipy<2.0.0,>=1.2.0",
@@ -79,9 +80,9 @@ all_extras = [
     "statsmodels>=0.12.1",
     "stumpy>=1.5.1; python_version < '3.11'",
     "tbats>=1.1.0",
-    "tensorflow-probability; python_version < '3.11'",
+    "tensorflow-probability<0.20.0; python_version < '3.11'",
     "tensorflow; python_version < '3.11'",
-    "tsfresh>=0.17.0; python_version < '3.10'",
+    "tsfresh>=0.17.0",
     "tslearn>=0.5.2; python_version < '3.11'",
     "xarray",
     "seasonal",

diff --git a/sktime/_nopytest_tests.py b/sktime/_nopytest_tests.py
@@ -0,0 +1,9 @@
+# -*- coding: utf-8 -*-
+"""Tests to run without pytest, to check pytest isolation."""
+# copyright: sktime developers, BSD-3-Clause License (see LICENSE file)
+
+from sktime.registry import all_estimators
+
+# all_estimators crawls all modules excepting pytest test files
+# if it encounters an unisolated import, it will throw an exception
+results = all_estimators()
diff --git a/sktime/benchmarking/forecasting.py b/sktime/benchmarking/forecasting.py
@@ -36,10 +36,9 @@ def forecasting_validation(
     """
     y = dataset_loader()
     results = {}
+    scores_df = evaluate(forecaster=estimator, y=y, cv=cv_splitter, scoring=scorers)
     for scorer in scorers:
         scorer_name = scorer.name
-        # TODO re-write evaluate to allow multiple scorers, to avoid recomputation
-        scores_df = evaluate(forecaster=estimator, y=y, cv=cv_splitter, scoring=scorer)
         for ix, row in scores_df.iterrows():
             results[f"{scorer_name}_fold_{ix}_test"] = row[f"test_{scorer_name}"]
         results[f"{scorer_name}_mean"] = scores_df[f"test_{scorer_name}"].mean()

diff --git a/sktime/benchmarking/tests/test_forecasting.py b/sktime/benchmarking/tests/test_forecasting.py
@@ -7,9 +7,43 @@
 from sktime.benchmarking import forecasting
 from sktime.forecasting.model_selection import ExpandingWindowSplitter
 from sktime.forecasting.naive import NaiveForecaster
-from sktime.performance_metrics.forecasting import MeanSquaredPercentageError
+from sktime.performance_metrics.forecasting import (
+    MeanAbsoluteError,
+    MeanAbsolutePercentageError,
+    MeanSquaredPercentageError,
+)
 from sktime.utils.validation._dependencies import _check_soft_dependencies
 
+expected_results_df_1 = pd.DataFrame(
+    data={
+        "validation_id": "[dataset=data_loader_simple]_"
+        + "[cv_splitter=ExpandingWindowSplitter]-v1",
+        "model_id": "NaiveForecaster-v1",
+        "MeanSquaredPercentageError_fold_0_test": 0.0,
+        "MeanSquaredPercentageError_fold_1_test": 0.111,
+        "MeanSquaredPercentageError_mean": 0.0555,
+        "MeanSquaredPercentageError_std": 0.0785,
+    },
+    index=[0],
+)
+
+expected_results_df_2 = pd.DataFrame(
+    data={
+        "validation_id": "[dataset=data_loader_simple]_"
+        + "[cv_splitter=ExpandingWindowSplitter]-v1",
+        "model_id": "NaiveForecaster-v1",
+        "MeanAbsolutePercentageError_fold_0_test": 0.0,
+        "MeanAbsolutePercentageError_fold_1_test": 0.333,
+        "MeanAbsolutePercentageError_mean": 0.1666,
+        "MeanAbsolutePercentageError_std": 0.2357,
+        "MeanAbsoluteError_fold_0_test": 0.0,
+        "MeanAbsoluteError_fold_1_test": 1.0,
+        "MeanAbsoluteError_mean": 0.5,
+        "MeanAbsoluteError_std": 0.7071,
+    },
+    index=[0],
+)
+
 
 def data_loader_simple() -> pd.DataFrame:
     """Return simple data for use in testing."""
@@ -20,7 +54,14 @@ def data_loader_simple() -> pd.DataFrame:
     not _check_soft_dependencies("kotsu", severity="none"),
     reason="skip test if required soft dependencies not available",
 )
-def test_forecastingbenchmark(tmp_path):
+@pytest.mark.parametrize(
+    "expected_results_df, scorers",
+    [
+        (expected_results_df_1, [MeanSquaredPercentageError()]),
+        (expected_results_df_2, [MeanAbsolutePercentageError(), MeanAbsoluteError()]),
+    ],
+)
+def test_forecastingbenchmark(tmp_path, expected_results_df, scorers):
     """Test benchmarking a forecaster estimator."""
     benchmark = forecasting.ForecastingBenchmark()
 
@@ -31,37 +72,13 @@ def test_forecastingbenchmark(tmp_path):
         step_length=1,
         fh=1,
     )
-    benchmark.add_task(data_loader_simple, cv_splitter, [MeanSquaredPercentageError()])
+    benchmark.add_task(data_loader_simple, cv_splitter, scorers)
 
     results_file = tmp_path / "results.csv"
     results_df = benchmark.run(results_file)
 
     results_df = results_df.drop(columns=["runtime_secs"])
 
-    expected_results_df = pd.DataFrame(
-        [
-            (
-                (
-                    "[dataset=data_loader_simple]_"
-                    "[cv_splitter=ExpandingWindowSplitter]-v1"
-                ),
-                "NaiveForecaster-v1",
-                0.0,
-                0.111,
-                0.0555,
-                0.0785,
-            )
-        ],
-        columns=[
-            "validation_id",
-            "model_id",
-            "MeanSquaredPercentageError_fold_0_test",
-            "MeanSquaredPercentageError_fold_1_test",
-            "MeanSquaredPercentageError_mean",
-            "MeanSquaredPercentageError_std",
-        ],
-    )
-
     pd.testing.assert_frame_equal(
         expected_results_df, results_df, check_exact=False, atol=0, rtol=0.001
     )
diff --git a/sktime/classification/compose/__init__.py b/sktime/classification/compose/__init__.py
@@ -12,11 +12,13 @@
 ]
 
 from sktime.classification.compose._column_ensemble import ColumnEnsembleClassifier
-from sktime.classification.compose._ensemble import (
-    ComposableTimeSeriesForestClassifier,
-    WeightedEnsembleClassifier,
-)
 from sktime.classification.compose._pipeline import (
     ClassifierPipeline,
     SklearnClassifierPipeline,
 )
+
+# 0.20.0 - remove this import
+from sktime.classification.ensemble import (
+    ComposableTimeSeriesForestClassifier,
+    WeightedEnsembleClassifier,
+)
diff --git a/sktime/classification/deep_learning/base.py b/sktime/classification/deep_learning/base.py
@@ -30,10 +30,9 @@ class BaseDeepClassifier(BaseClassifier, ABC):
     batch_size : int, default = 40
         training batch size for the model
 
-    Arguments
-    ---------
-    self.model = None
-
+    Attributes
+    ----------
+    self.model_ - the fitted DL model
     """
 
     _tags = {