Merge branch 'main' into deprecate-wops

.github/workflows/build-docs.yaml

@@ -11,14 +11,14 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
       - name: Clone docs repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
         with:
           repository: Nixtla/docs
           ref: scripts
           path: docs-scripts
-      - uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+      - uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: "3.10"
       - name: Build docs

.github/workflows/ci.yaml

@@ -24,10 +24,10 @@ jobs:
       AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY_NIXTLA_TMP }}
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
       - name: Set up environment
-        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: ${{ matrix.python-version }}
 
@@ -47,10 +47,10 @@ jobs:
         python-version: ["3.8", "3.9", "3.10", "3.11"]
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
       - name: Set up environment
-        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: ${{ matrix.python-version }}
 
@@ -68,10 +68,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
       - name: Set up python
-        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: "3.10"
 
@@ -84,27 +84,27 @@ jobs:
   efficiency-tests:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
-      - uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+      - uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: "3.12"
 
       - name: Install dependencies
         run: pip install uv && uv pip install --system . pytest-codspeed pytest-xdist
 
       - name: Run benchmarks
-        uses: CodSpeedHQ/action@ab07afd34cbbb7a1306e8d14b7cc44e029eee37a # 3.0.0
+        uses: CodSpeedHQ/action@fa1dcde8d58f2ab0b407a6a24d6cc5a8c1444a8c # 3.1.0
         with:
           token: ${{ secrets.CODESPEED_TOKEN }}
           run: pytest tests/test_pipeline.py --codspeed -n 2
 
   performance-tests:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
-      - uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+      - uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: "3.10"
 

.github/workflows/lint.yaml

@@ -11,10 +11,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
       - name: Set up python
-        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: '3.10'
           cache: 'pip'

.github/workflows/release.yml

@@ -17,10 +17,10 @@ jobs:
       id-token: write    
     steps:
       - name: Clone repo
-        uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
 
       - name: Set up python
-        uses: actions/setup-python@39cd14951b08e74b54015e9e001cdefcf80e669f # 5.1.1
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # 5.3.0
         with:
           python-version: '3.10' 
 
@@ -31,4 +31,4 @@ jobs:
         run: python -m build -sw
 
       - name: Publish package to PyPI
-        uses: pypa/gh-action-pypi-publish@ec4db0b4ddc65acdf4bff5fa45ac92d78b56bdf0 # v1.9.0
+        uses: pypa/gh-action-pypi-publish@15c56dba361d8335944d31a2ecd17d700fc7bcbc # v1.12.2

mlforecast/__init__.py

@@ -1,3 +1,3 @@
-__version__ = "0.13.4"
+__version__ = "0.15.0"
 __all__ = ['MLForecast']
 from mlforecast.forecast import MLForecast

mlforecast/_modidx.py

@@ -254,6 +254,8 @@
                                                                                             'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Combine.update': ( 'lag_transforms.html#combine.update',
                                                                                          'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms.Combine.update_samples': ( 'lag_transforms.html#combine.update_samples',
+                                                                                                 'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExpandingMax': ( 'lag_transforms.html#expandingmax',
                                                                                        'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExpandingMean': ( 'lag_transforms.html#expandingmean',
@@ -264,12 +266,16 @@
                                                                                             'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExpandingQuantile.__init__': ( 'lag_transforms.html#expandingquantile.__init__',
                                                                                                      'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms.ExpandingQuantile.update_samples': ( 'lag_transforms.html#expandingquantile.update_samples',
+                                                                                                           'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExpandingStd': ( 'lag_transforms.html#expandingstd',
                                                                                        'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExponentiallyWeightedMean': ( 'lag_transforms.html#exponentiallyweightedmean',
                                                                                                     'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.ExponentiallyWeightedMean.__init__': ( 'lag_transforms.html#exponentiallyweightedmean.__init__',
                                                                                                              'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms.ExponentiallyWeightedMean.update_samples': ( 'lag_transforms.html#exponentiallyweightedmean.update_samples',
+                                                                                                                   'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Lag': ('lag_transforms.html#lag', 'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Lag.__eq__': ( 'lag_transforms.html#lag.__eq__',
                                                                                      'mlforecast/lag_transforms.py'),
@@ -279,6 +285,8 @@
                                                                                         'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Lag._set_core_tfm': ( 'lag_transforms.html#lag._set_core_tfm',
                                                                                             'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms.Lag.update_samples': ( 'lag_transforms.html#lag.update_samples',
+                                                                                             'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Offset': ( 'lag_transforms.html#offset',
                                                                                  'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Offset.__init__': ( 'lag_transforms.html#offset.__init__',
@@ -287,6 +295,8 @@
                                                                                            'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.Offset._set_core_tfm': ( 'lag_transforms.html#offset._set_core_tfm',
                                                                                                'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms.Offset.update_samples': ( 'lag_transforms.html#offset.update_samples',
+                                                                                                'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.RollingMax': ( 'lag_transforms.html#rollingmax',
                                                                                      'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms.RollingMean': ( 'lag_transforms.html#rollingmean',
@@ -319,6 +329,8 @@
                                                                                                                 'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._BaseLagTransform._get_name': ( 'lag_transforms.html#_baselagtransform._get_name',
                                                                                                       'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms._BaseLagTransform._lag': ( 'lag_transforms.html#_baselagtransform._lag',
+                                                                                                 'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._BaseLagTransform._set_core_tfm': ( 'lag_transforms.html#_baselagtransform._set_core_tfm',
                                                                                                           'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._BaseLagTransform.stack': ( 'lag_transforms.html#_baselagtransform.stack',
@@ -329,18 +341,26 @@
                                                                                                       'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._BaseLagTransform.update': ( 'lag_transforms.html#_baselagtransform.update',
                                                                                                    'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms._BaseLagTransform.update_samples': ( 'lag_transforms.html#_baselagtransform.update_samples',
+                                                                                                           'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._ExpandingBase': ( 'lag_transforms.html#_expandingbase',
                                                                                          'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._ExpandingBase.__init__': ( 'lag_transforms.html#_expandingbase.__init__',
                                                                                                   'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms._ExpandingBase.update_samples': ( 'lag_transforms.html#_expandingbase.update_samples',
+                                                                                                        'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._RollingBase': ( 'lag_transforms.html#_rollingbase',
                                                                                        'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._RollingBase.__init__': ( 'lag_transforms.html#_rollingbase.__init__',
                                                                                                 'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms._RollingBase.update_samples': ( 'lag_transforms.html#_rollingbase.update_samples',
+                                                                                                      'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._Seasonal_RollingBase': ( 'lag_transforms.html#_seasonal_rollingbase',
                                                                                                 'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._Seasonal_RollingBase.__init__': ( 'lag_transforms.html#_seasonal_rollingbase.__init__',
                                                                                                          'mlforecast/lag_transforms.py'),
+                                           'mlforecast.lag_transforms._Seasonal_RollingBase.update_samples': ( 'lag_transforms.html#_seasonal_rollingbase.update_samples',
+                                                                                                               'mlforecast/lag_transforms.py'),
                                            'mlforecast.lag_transforms._pascal2camel': ( 'lag_transforms.html#_pascal2camel',
                                                                                         'mlforecast/lag_transforms.py')},
             'mlforecast.lgb_cv': { 'mlforecast.lgb_cv.LightGBMCV': ('lgb_cv.html#lightgbmcv', 'mlforecast/lgb_cv.py'),

mlforecast/auto.py

@@ -444,6 +444,8 @@ def fit(
         n_windows: int,
         h: int,
         num_samples: int,
+        step_size: Optional[int] = None,
+        input_size: Optional[int] = None,
         refit: Union[bool, int] = False,
         loss: Optional[Callable[[DataFrame, DataFrame], float]] = None,
         id_col: str = "unique_id",
@@ -467,6 +469,10 @@ def fit(
             Forecast horizon.
         num_samples : int
             Number of trials to run
+        step_size : int, optional (default=None)
+            Step size between each cross validation window. If None it will be equal to `h`.
+        input_size : int, optional (default=None)
+            Maximum training samples per serie in each window. If None, will use an expanding window.
         refit : bool or int (default=False)
             Retrain model for each cross validation window.
             If False, the models are trained at the beginning and then used to predict each window.
@@ -509,7 +515,12 @@ def fit(
         if loss is None:
 
             def loss(df, train_df):  # noqa: ARG001
-                return smape(df, models=["model"])["model"].mean()
+                return smape(
+                    df,
+                    models=["model"],
+                    id_col=id_col,
+                    target_col=target_col,
+                )["model"].mean()
 
         if study_kwargs is None:
             study_kwargs = {}
@@ -541,6 +552,8 @@ def config_fn(trial: optuna.Trial) -> Dict[str, Any]:
                 freq=self.freq,
                 n_windows=n_windows,
                 h=h,
+                step_size=step_size,
+                input_size=input_size,
                 refit=refit,
                 id_col=id_col,
                 time_col=time_col,
@@ -550,8 +563,14 @@ def config_fn(trial: optuna.Trial) -> Dict[str, Any]:
             study.optimize(objective, n_trials=num_samples, **optimize_kwargs)
             self.results_[name] = study
             best_config = study.best_trial.user_attrs["config"]
-            best_config["mlf_fit_params"].pop("fitted", None)
-            best_config["mlf_fit_params"].pop("prediction_intervals", None)
+            for arg in (
+                "fitted",
+                "prediction_intervals",
+                "id_col",
+                "time_col",
+                "target_col",
+            ):
+                best_config["mlf_fit_params"].pop(arg, None)
             best_model = clone(auto_model.model)
             best_model.set_params(**best_config["model_params"])
             self.models_[name] = MLForecast(
@@ -563,6 +582,9 @@ def config_fn(trial: optuna.Trial) -> Dict[str, Any]:
                 df,
                 fitted=fitted,
                 prediction_intervals=prediction_intervals,
+                id_col=id_col,
+                time_col=time_col,
+                target_col=target_col,
                 **best_config["mlf_fit_params"],
             )
         return self