[MNT] Simplify tests

.github/workflows/ci.yaml

@@ -10,7 +10,7 @@ on:
       - develop
       - release/*
   workflow_dispatch:
-    
+
 
 jobs:
   build:
@@ -34,13 +34,13 @@ jobs:
 
     - name: Install dependencies
       run: poetry install --no-interaction --no-root --all-extras
-    
+
     - name: Set PYTHONPATH
       run: echo "PYTHONPATH=$GITHUB_WORKSPACE/src" >> $GITHUB_ENV
 
     - name: Test with pytest
-      run: poetry run pytest --cov=prophetverse --cov-report=xml
-      
+      run: poetry run pytest --cov=prophetverse --cov-report=xml -m "not smoke"
+
     - name: Upload coverage reports to Codecov
       uses: codecov/codecov-action@v4.0.1
       with:

docs/development/development-guide.md

@@ -44,6 +44,16 @@ Community slack: None yet.
 ### Code standards
 Writing good code is not just about what you write. It is also about how you write it. During Continuous Integration testing, several tools will be run to check your code for stylistic errors. Generating any warnings will cause the test to fail. Thus, good style is a requirement for submitting code to Prophetverse.There are of tools in Prophetverse to help contributors verify their changes before contributing to the project
 
+#### [Pytest](https://docs.pytest.org/en/7.1.x/contents.html) 
+You can test your code with pytest integration with the poetry command
+<br> ```poetry run pytest```
+
+The CI tests are computationally intensive, so if you want to do a faster test you can run a [smoke test](https://en.wikipedia.org/wiki/Smoke_testing_(software)) with the command 
+<br> ```poetry run pytest -m "not ci"```
+
+If you also wanna run the tests even faster feel free to parallel processing the tests with [pytest-xdist](https://pytest-xdist.readthedocs.io/en/latest/how-to.html#making-session-scoped-fixtures-execute-only-once).
+
+
 #### [Pre-commit](https://pre-commit.com/)
 
 Additionally, Continuous Integration will run code formatting checks like black, isort, and mypy and more using pre-commit hooks. Any warnings from these checks will cause the Continuous Integration to fail; therefore, it is helpful to run the check yourself before submitting code. This can be done by installing pre-commit (which should already have happened if you followed the instructions in Setting up your development environment) and then running:

pyproject.toml

@@ -33,6 +33,11 @@ pydocstyle = "^6.3.0"
 mypy = "^1.10.0"
 pylint = "^3.2.2"
 
+[tool.pytest.ini_options]
+markers = [
+    "ci: marks tests for Continuous Integration",
+    "smoke: marks tests for smoke testing",
+    ]
 
 [build-system]
 requires = ["poetry-core"]

tests/conftest.py

@@ -1,10 +1,13 @@
+"""Configure tests and declare global fixtures."""
+
+import warnings
+
 import numpyro
 
+warnings.filterwarnings("ignore")
+
 
 def pytest_sessionstart(session):
-
-
+    """Avoid NaNs in tests."""
     print("Enabling x64")
-    # Avoid NaNs in tests
     numpyro.enable_x64()
-

tests/sktime/test_multivariate.py

@@ -5,38 +5,32 @@
 from sktime.forecasting.base import ForecastingHorizon
 from sktime.split import temporal_train_test_split
 from sktime.transformations.hierarchical.aggregate import Aggregator
-from sktime.utils._testing.hierarchical import (_bottom_hier_datagen,
-                                                _make_hierarchical)
+from sktime.utils._testing.hierarchical import _bottom_hier_datagen, _make_hierarchical
 
 from prophetverse.effects import LinearEffect
 from prophetverse.sktime.multivariate import HierarchicalProphet
 from prophetverse.sktime.seasonality import seasonal_transformer
 
-from ._utils import (execute_extra_predict_methods_tests,
-                     execute_fit_predict_test, make_empty_X, make_None_X,
-                     make_random_X, make_y)
+from ._utils import (
+    execute_extra_predict_methods_tests,
+    execute_fit_predict_test,
+    make_empty_X,
+    make_None_X,
+    make_random_X,
+    make_y,
+)
 
 HYPERPARAMS = [
+    dict(feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True)),
     dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        )
-    ),
-    dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        ),
+        feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True),
         default_effect=LinearEffect(effect_mode="multiplicative"),
     ),
     dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        ),
+        feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True),
         exogenous_effects=[
             LinearEffect(id="lineareffect1", regex=r"(x1).*"),
-            LinearEffect(
-                id="lineareffect2", regex=r"(x2).*", prior=dist.Laplace(0, 1)
-            ),
+            LinearEffect(id="lineareffect2", regex=r"(x2).*", prior=dist.Laplace(0, 1)),
         ],
     ),
     dict(
@@ -45,14 +39,34 @@
     dict(trend="logistic"),
     dict(inference_method="mcmc"),
     dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        ),
+        feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True),
         shared_features=["x1"],
     ),
 ]
 
 
+@pytest.mark.smoke
+@pytest.mark.parametrize("hierarchy_levels", [0, (1,), (2, 1), (1, 2), (3, 2, 2)])
+def test_hierarchy_levels(hierarchy_levels):
+    y = make_y(hierarchy_levels)
+    X = make_random_X(y)
+    forecaster = HierarchicalProphet(optimizer_steps=20, changepoint_interval=2, mcmc_samples=2, mcmc_warmup=2)
+    execute_fit_predict_test(forecaster, y, X)
+
+
+@pytest.mark.smoke
+@pytest.mark.parametrize("hyperparams", HYPERPARAMS)
+def test_hyperparams(hyperparams):
+    hierarchy_levels = (2, 1)
+    y = make_y(hierarchy_levels)
+    X = make_random_X(y)
+    forecaster = HierarchicalProphet(
+        **hyperparams, optimizer_steps=20, changepoint_interval=2, mcmc_samples=2, mcmc_warmup=2
+    )
+    execute_fit_predict_test(forecaster, y, X)
+
+
+@pytest.mark.ci
 @pytest.mark.parametrize("hierarchy_levels", [0, (1,), (2, 1), (1, 2), (3, 2, 2)])
 @pytest.mark.parametrize("make_X", [make_random_X, make_None_X, make_empty_X])
 @pytest.mark.parametrize("hyperparams", HYPERPARAMS)
@@ -62,11 +76,7 @@ def test_prophet2_fit_with_different_nlevels(hierarchy_levels, make_X, hyperpara
     X = make_X(y)
 
     forecaster = HierarchicalProphet(
-        **hyperparams,
-        optimizer_steps=20,
-        changepoint_interval=2,
-        mcmc_samples=2,
-        mcmc_warmup=2
+        **hyperparams, optimizer_steps=20, changepoint_interval=2, mcmc_samples=2, mcmc_warmup=2
     )
 
     execute_fit_predict_test(forecaster, y, X)
@@ -76,11 +86,6 @@ def test_prophet2_fit_with_different_nlevels(hierarchy_levels, make_X, hyperpara
 def test_extra_predict_methods(make_X):
     y = make_y((2, 1))
     X = make_X(y)
-    forecaster = HierarchicalProphet(
-        optimizer_steps=20,
-        changepoint_interval=2,
-        mcmc_samples=2,
-        mcmc_warmup=2
-    )
+    forecaster = HierarchicalProphet(optimizer_steps=20, changepoint_interval=2, mcmc_samples=2, mcmc_warmup=2)
 
     execute_extra_predict_methods_tests(forecaster=forecaster, X=X, y=y)

tests/sktime/test_univariate.py

@@ -4,18 +4,21 @@
 from numpyro import distributions as dist
 from sktime.forecasting.base import ForecastingHorizon
 from sktime.transformations.hierarchical.aggregate import Aggregator
-from sktime.utils._testing.hierarchical import (_bottom_hier_datagen,
-                                                _make_hierarchical)
+from sktime.utils._testing.hierarchical import _bottom_hier_datagen, _make_hierarchical
 
 from prophetverse.effects import LinearEffect
 from prophetverse.sktime.seasonality import seasonal_transformer
-from prophetverse.sktime.univariate import (Prophet, ProphetGamma,
-                                            ProphetNegBinomial)
+from prophetverse.sktime.univariate import Prophet, ProphetGamma, ProphetNegBinomial
 from prophetverse.trend.flat import FlatTrend
 
-from ._utils import (execute_extra_predict_methods_tests,
-                     execute_fit_predict_test, make_empty_X, make_None_X,
-                     make_random_X, make_y)
+from ._utils import (
+    execute_extra_predict_methods_tests,
+    execute_fit_predict_test,
+    make_empty_X,
+    make_None_X,
+    make_random_X,
+    make_y,
+)
 
 MODELS = [
     Prophet,
@@ -24,27 +27,16 @@
 ]
 
 HYPERPARAMS = [
+    dict(trend=FlatTrend(), feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True)),
     dict(
-        trend=FlatTrend(),
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        )
-    ),
-    dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        ),
+        feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True),
         default_effect=LinearEffect(effect_mode="multiplicative"),
     ),
     dict(
-        feature_transformer=seasonal_transformer(
-            yearly_seasonality=True, weekly_seasonality=True
-        ),
+        feature_transformer=seasonal_transformer(yearly_seasonality=True, weekly_seasonality=True),
         exogenous_effects=[
             LinearEffect(id="lineareffect1", regex=r"(x1).*"),
-            LinearEffect(
-                id="lineareffect2", regex=r"(x2).*", prior=dist.Laplace(0, 1)
-            ),
+            LinearEffect(id="lineareffect2", regex=r"(x2).*", prior=dist.Laplace(0, 1)),
         ],
     ),
     dict(
@@ -56,35 +48,70 @@
 ]
 
 
+@pytest.mark.smoke
 @pytest.mark.parametrize("model_class", MODELS)
+def test_model_class_fit(model_class):
+    hierarchy_levels = (1,)
+    make_X = make_random_X
+    hyperparams = HYPERPARAMS[0]
+
+    y = make_y(hierarchy_levels)
+    X = make_X(y)
+    forecaster = model_class(**hyperparams, optimizer_steps=10, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1)
+
+    execute_fit_predict_test(forecaster, y, X, test_size=4)
+
+
+@pytest.mark.smoke
 @pytest.mark.parametrize("hierarchy_levels", [(1,), (2,), (2, 1)])
-@pytest.mark.parametrize("make_X", [make_random_X, make_None_X, make_empty_X])
+def test_hierarchy_levels_fit(hierarchy_levels):
+    model_class = MODELS[0]
+    make_X = make_random_X
+    hyperparams = HYPERPARAMS[0]
+
+    y = make_y(hierarchy_levels)
+    X = make_X(y)
+    forecaster = model_class(**hyperparams, optimizer_steps=10, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1)
+
+    execute_fit_predict_test(forecaster, y, X, test_size=4)
+
+
+@pytest.mark.smoke
 @pytest.mark.parametrize("hyperparams", HYPERPARAMS)
-def test_prophet2_fit_with_different_nlevels(model_class, hierarchy_levels, make_X, hyperparams):
-
+def test_hyperparams_fit(hyperparams):
+    model_class = MODELS[1]
+    hierarchy_levels = (1,)
+    make_X = make_random_X
+
     y = make_y(hierarchy_levels)
-    X = make_X(
-        y
-    )
-    forecaster = model_class(
-        **hyperparams, optimizer_steps=100, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1
-    )
-
+    X = make_X(y)
+    forecaster = model_class(**hyperparams, optimizer_steps=10, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1)
+
     execute_fit_predict_test(forecaster, y, X, test_size=4)
 
+
 @pytest.mark.parametrize("make_X", [make_random_X, make_None_X, make_empty_X])
 def test_extra_predict_methods(make_X):
-    y = make_y((2,1))
-    X = make_X(
-        y
-    )
-    forecaster = Prophet(
-        optimizer_steps=100, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1
-    )
+    y = make_y((2, 1))
+    X = make_X(y)
+    forecaster = Prophet(optimizer_steps=10, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1)
     execute_extra_predict_methods_tests(forecaster=forecaster, X=X, y=y)
-
-
+
+
+@pytest.mark.ci
+@pytest.mark.parametrize("model_class", MODELS)
+@pytest.mark.parametrize("hierarchy_levels", [(1,), (2,), (2, 1)])
+@pytest.mark.parametrize("make_X", [make_random_X, make_None_X, make_empty_X])
+@pytest.mark.parametrize("hyperparams", HYPERPARAMS)
+def test_prophet2_fit_with_different_nlevels(model_class, hierarchy_levels, make_X, hyperparams):
+
+    y = make_y(hierarchy_levels)
+    X = make_X(y)
+    forecaster = model_class(**hyperparams, optimizer_steps=100, mcmc_samples=2, mcmc_warmup=2, mcmc_chains=1)
+
+    execute_fit_predict_test(forecaster, y, X, test_size=4)
+
+
 def test_raise_error_when_passing_bad_trend():
     with pytest.raises(ValueError):
         Prophet(trend="bad_trend")
-