Feat/show_anomalies for multivariate (#2544)

* Add new feature to plot each series's component separately

* Update the docstring

* refactor the show_anomalies_from_scores()

* Update changelog

* Improve code in utils.py

* Improve code in utils.py

* make check_multivariate depend on multivariate_plot

* update utils

* improve the spacing between suptitle and axes

* update utils

* improve the spacing between suptitle and axes

* Fix the height_ratios of each subplot

* Change "multivariate_plot" parameter name to "component_wise"

* make title fit better

---------

Co-authored-by: madtoinou <[email protected]>
Co-authored-by: dennisbader <[email protected]>

CHANGELOG.md

@@ -12,6 +12,7 @@ but cannot always guarantee backwards compatibility. Changes that may **break co
 **Improved**
 
 - New model: `StatsForecastAutoTBATS`. This model offers the [AutoTBATS](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autotbats) model from Nixtla's `statsforecasts` library. [#2611](https://github.com/unit8co/darts/pull/2611) by [He Weilin](https://github.com/cnhwl).
+- Added parameter `component_wise` to `show_anomalies()` to separately plot each component in multivariate series. [#2544](https://github.com/unit8co/darts/pull/2544) by [He Weilin](https://github.com/cnhwl).
 
 **Fixed**
 - Fixed a bug when performing optimized historical forecasts with `stride=1` using a `RegressionModel` with `output_chunk_shift>=1` and `output_chunk_length=1`, where the forecast time index was not properly shifted. [#2634](https://github.com/unit8co/darts/pull/2634) by [Mattias De Charleroy](https://github.com/MattiasDC).

darts/ad/anomaly_model/anomaly_model.py

@@ -247,6 +247,7 @@ def show_anomalies(
         names_of_scorers: Union[str, Sequence[str]] = None,
         title: str = None,
         metric: Optional[Literal["AUC_ROC", "AUC_PR"]] = None,
+        component_wise: bool = False,
         **score_kwargs,
     ):
         """Plot the results of the anomaly model.
@@ -283,6 +284,8 @@ def show_anomalies(
             Default: "AUC_ROC".
         score_kwargs
             parameters for the `score()` method.
+        component_wise
+            If True, will separately plot each component in case of multivariate anomaly detection.
         """
         series = _check_input(series, name="series", num_series_expected=1)[0]
         predict_kwargs = predict_kwargs if predict_kwargs is not None else {}
@@ -310,6 +313,7 @@ def show_anomalies(
             names_of_scorers=names_of_scorers,
             title=title,
             metric=metric,
+            component_wise=component_wise,
         )
 
     @property

darts/ad/anomaly_model/forecasting_am.py

@@ -440,6 +440,7 @@ def show_anomalies(
         names_of_scorers: Union[str, Sequence[str]] = None,
         title: str = None,
         metric: Optional[Literal["AUC_ROC", "AUC_PR"]] = None,
+        component_wise: bool = False,
         **score_kwargs,
     ):
         """Plot the results of the anomaly model.
@@ -506,6 +507,8 @@ def show_anomalies(
             Optionally, the name of the metric function to use. Must be one of "AUC_ROC" (Area Under the
             Receiver Operating Characteristic Curve) and "AUC_PR" (Average Precision from scores).
             Default: "AUC_ROC".
+        component_wise
+            If True, will separately plot each component in case of multivariate anomaly detection.
         score_kwargs
             parameters for the `score()` method.
         """
@@ -527,6 +530,7 @@ def show_anomalies(
             names_of_scorers=names_of_scorers,
             title=title,
             metric=metric,
+            component_wise=component_wise,
             **score_kwargs,
         )
 

darts/ad/scorers/scorers.py

@@ -176,6 +176,7 @@ def show_anomalies_from_prediction(
         anomalies: TimeSeries = None,
         title: str = None,
         metric: Optional[Literal["AUC_ROC", "AUC_PR"]] = None,
+        component_wise: bool = False,
     ):
         """Plot the results of the scorer.
 
@@ -208,6 +209,8 @@ def show_anomalies_from_prediction(
             Optionally, the name of the metric function to use. Must be one of "AUC_ROC" (Area Under the
             Receiver Operating Characteristic Curve) and "AUC_PR" (Average Precision from scores).
             Default: "AUC_ROC".
+        component_wise
+            If True, will separately plot each component in case of multivariate anomaly detection.
         """
         series = _check_input(series, name="series", num_series_expected=1)[0]
         pred_series = _check_input(
@@ -230,6 +233,7 @@ def show_anomalies_from_prediction(
             names_of_scorers=scorer_name,
             title=title,
             metric=metric,
+            component_wise=component_wise,
         )
 
     @property
@@ -580,6 +584,7 @@ def show_anomalies(
         scorer_name: str = None,
         title: str = None,
         metric: Optional[Literal["AUC_ROC", "AUC_PR"]] = None,
+        component_wise: bool = False,
     ):
         """Plot the results of the scorer.
 
@@ -610,6 +615,8 @@ def show_anomalies(
             Optionally, the name of the metric function to use. Must be one of "AUC_ROC" (Area Under the
             Receiver Operating Characteristic Curve) and "AUC_PR" (Average Precision from scores).
             Default: "AUC_ROC".
+        component_wise
+            If True, will separately plot each component in case of multivariate anomaly detection.
         """
         series = _check_input(series, name="series", num_series_expected=1)[0]
         pred_scores = self.score(series)
@@ -633,6 +640,7 @@ def show_anomalies(
             names_of_scorers=scorer_name,
             title=title,
             metric=metric,
+            component_wise=component_wise,
         )
 
     @property