autogluon.timeseries 1.0.0b20231122__tar.gz → 1.0.0b20231124__tar.gz

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.timeseries
-Version: 1.0.0b20231122
+Version: 1.0.0b20231124
 Summary: AutoML for Image, Text, and Tabular Data
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/setup.py

@@ -37,6 +37,8 @@ install_requires = [
     "utilsforecast>=0.0.10,<0.0.11",
     "tqdm",  # version range defined in `core/_setup_utils.py`
     "orjson~=3.9",  # use faster JSON implementation in GluonTS
+    # TODO v1.1: use lightning[pytorch-extra] instead of explicitly installing tensorboard
+    "tensorboard>=2.9,<3",  # fixes https://github.com/autogluon/autogluon/issues/3612
     f"autogluon.core[raytune]=={version}",
     f"autogluon.common=={version}",
     f"autogluon.tabular[catboost,lightgbm,xgboost]=={version}",

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/src/autogluon/timeseries/models/autogluon_tabular/mlforecast.py

@@ -296,10 +296,7 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
 
     def _get_scale_per_item(self, item_ids: pd.Index) -> pd.Series:
         """Extract the '_scale' values from the scaler object, if available."""
-        if self._scaler is not None:
-            return self._scaler.stats_["_scale"].copy().reindex(item_ids)
-        else:
-            return pd.Series(1.0, index=item_ids)
+        raise NotImplementedError
 
     def _remove_short_ts_and_generate_fallback_forecast(
         self,
@@ -469,7 +466,9 @@ class DirectTabularModel(AbstractMLForecastModel):
 
         raw_predictions = self._mlf.models_["mean"].predict(df)
         predictions = self._postprocess_predictions(raw_predictions, repeated_item_ids=df[MLF_ITEMID])
-        predictions[[MLF_ITEMID, MLF_TIMESTAMP]] = df[[MLF_ITEMID, MLF_TIMESTAMP]].values
+        # Paste columns one by one to preserve dtypes
+        predictions[MLF_ITEMID] = df[MLF_ITEMID].values
+        predictions[MLF_TIMESTAMP] = df[MLF_TIMESTAMP].values
 
         if hasattr(self._mlf.ts, "target_transforms"):
             # Ensure that transforms are fitted only on past data
@@ -498,6 +497,10 @@ class DirectTabularModel(AbstractMLForecastModel):
         column_order = ["mean"] + [col for col in predictions.columns if col != "mean"]
         return predictions[column_order]
 
+    def _get_scale_per_item(self, item_ids: pd.Index) -> pd.Series:
+        # Rescaling is applied in the inverse_transform step, no need to scale predictions
+        return pd.Series(1.0, index=item_ids)
+
     def _get_extra_tabular_init_kwargs(self) -> dict:
         if self.is_quantile_model:
             return {
@@ -603,3 +606,9 @@ class RecursiveTabularModel(AbstractMLForecastModel):
             "problem_type": ag.constants.REGRESSION,
             "eval_metric": self.eval_metric.equivalent_tabular_regression_metric or "mean_absolute_error",
         }
+
+    def _get_scale_per_item(self, item_ids: pd.Index) -> pd.Series:
+        if self._scaler is not None:
+            return self._scaler.stats_["_scale"].copy().reindex(item_ids)
+        else:
+            return pd.Series(1.0, index=item_ids)

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/src/autogluon/timeseries/version.py

@@ -1,3 +1,3 @@
 """This is the autogluon version file."""
-__version__ = '1.0.0b20231122'
+__version__ = '1.0.0b20231124'
 __lite__ = False

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/src/autogluon.timeseries.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.timeseries
-Version: 1.0.0b20231122
+Version: 1.0.0b20231124
 Summary: AutoML for Image, Text, and Tabular Data
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community

{autogluon.timeseries-1.0.0b20231122 → autogluon.timeseries-1.0.0b20231124}/src/autogluon.timeseries.egg-info/requires.txt

@@ -13,9 +13,10 @@ mlforecast<0.10.1,>=0.10.0
 utilsforecast<0.0.11,>=0.0.10
 tqdm<5,>=4.38
 orjson~=3.9
-autogluon.core[raytune]==1.0.0b20231122
-autogluon.common==1.0.0b20231122
-autogluon.tabular[catboost,lightgbm,xgboost]==1.0.0b20231122
+tensorboard<3,>=2.9
+autogluon.core[raytune]==1.0.0b20231124
+autogluon.common==1.0.0b20231124
+autogluon.tabular[catboost,lightgbm,xgboost]==1.0.0b20231124
 
 [all]