added-tbats-and-mfles

databricks-industry-solutions · Feb 5, 2025 · 8530757 · 8530757
1 parent ce7125a
commit 8530757
Show file tree

Hide file tree

Showing 11 changed files with 92 additions and 28 deletions.
diff --git a/README.md b/README.md
@@ -12,13 +12,14 @@ Get started now!
 
 ## What's New
 
-- Feb 2025: [Prophet](https://www.sktime.net/en/stable/api_reference/auto_generated/sktime.forecasting.fbprophet.Prophet.html) is available for univariate forecasting via `SKTimeProphet`. Try the [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/local_univariate_daily.py).
+- Feb 2025: [AutoTBATS](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autotbats) and [AutoMFLES](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#automfles) from statsforecast are available. Try the [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/local_univariate_daily.py).
+- Feb 2025: [Prophet](https://www.sktime.net/en/stable/api_reference/auto_generated/sktime.forecasting.fbprophet.Prophet.html) is available for univariate forecasting via sktime. Try the [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/local_univariate_daily.py).
 - Feb 2025: Added a post evaluation [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/post-evaluation-analysis.ipynb) for fine-grained model selection.
-- Feb 2025: Added [README](https://github.com/databricks-industry-solutions/many-model-forecasting/tree/main/mmf_sa/models) for a list of supported models.
-- Jan 2025: [TimesFM](https://github.com/google-research/timesfm) is available for univariate and covariate forecasting. Try the notebooks: [univariate](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/foundation_daily.py) and [covariate](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/external_regressors/foundation_external_regressors_daily.py).
+- Feb 2025: Added [README](https://github.com/databricks-industry-solutions/many-model-forecasting/tree/main/mmf_sa/models) for a comprehensive list of supported models.
+- Jan 2025: [TimesFM](https://github.com/google-research/timesfm) is available for univariate and covariate forecasting. Try the [univariate](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/foundation_daily.py) and [covariate](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/external_regressors/foundation_external_regressors_daily.py) notebooks.
 - Jan 2025: [Chronos Bolt](https://github.com/amazon-science/chronos-forecasting) models are available for univariate forecasting. Try the [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/foundation_daily.py).
 - Jan 2025: [Moirai MoE](https://github.com/SalesforceAIResearch/uni2ts) models are available for univariate forecasting. Try the [notebook](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/daily/foundation_daily.py).
-- Jan 2025: Added support for hourly (`freq="H"`) and weekly (`freq="W"`) time series . Try the notebooks: [hourly](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/hourly) and [weekly](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/weekly).
+- Jan 2025: Added support for hourly (`freq="H"`) and weekly (`freq="W"`) time series . Try the [hourly](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/hourly) and [weekly](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/examples/weekly) notebooks.
 
 ## Getting started
 
@@ -42,6 +43,8 @@ active_models = [
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/daily/local_univariate_daily.py b/examples/daily/local_univariate_daily.py
@@ -140,6 +140,8 @@ def transform_group(df):
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/external_regressors/local_univariate_external_regressors_daily.py b/examples/external_regressors/local_univariate_external_regressors_daily.py
@@ -115,6 +115,8 @@
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/hourly/local_univariate_hourly.py b/examples/hourly/local_univariate_hourly.py
@@ -134,6 +134,8 @@ def transform_group(df):
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/m5/local_univariate_daily_m5.py b/examples/m5/local_univariate_daily_m5.py
@@ -45,6 +45,8 @@
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/monthly/local_univariate_monthly.py b/examples/monthly/local_univariate_monthly.py
@@ -149,6 +149,8 @@ def transform_group(df):
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/examples/weekly/local_univariate_weekly.py b/examples/weekly/local_univariate_weekly.py
@@ -138,6 +138,8 @@ def transform_group(df):
     "StatsForecastAutoETS",
     "StatsForecastAutoCES",
     "StatsForecastAutoTheta",
+    "StatsForecastAutoTbats",
+    "StatsForecastAutoMfles",
     "StatsForecastTSB",
     "StatsForecastADIDA",
     "StatsForecastIMAPA",

diff --git a/mmf_sa/models/README.md b/mmf_sa/models/README.md
@@ -3,30 +3,32 @@
 Model hyperparameters can be modified under [mmf_sa/models/models_conf.yaml](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/models_conf.yaml).
 
 ## Local
-| model | source | covariate support |
-|----------------------------------------|-------------------------|------------|
-| StatsForecastBaselineWindowAverage | [Statsforecast Window Average](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#windowaverage) |  | 
-| StatsForecastBaselineSeasonalWindowAverage | [Statsforecast Seasonal Window Average](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#seasonalwindowaverage) |  | 
-| StatsForecastBaselineNaive | [Statsforecast Naive](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#naive) |  | 
-| StatsForecastBaselineSeasonalNaive | [Statsforecast Seasonal Naive](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#seasonalnaive) |  | 
-| StatsForecastAutoArima | [Statsforecast AutoARIMA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoarima) | ✅ | 
-| StatsForecastAutoETS | [Statsforecast AutoETS](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoets) |  | 
-| StatsForecastAutoCES | [Statsforecast AutoCES](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoces) |  | 
-| StatsForecastAutoTheta | [Statsforecast AutoTheta](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autotheta) |  | 
-| StatsForecastTSB | [Statsforecast TSB](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#tsb) |  | 
-| StatsForecastADIDA | [Statsforecast ADIDA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#adida) |  | 
-| StatsForecastIMAPA | [Statsforecast IMAPA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#imapa) |  | 
-| StatsForecastCrostonClassic | [Statsforecast Croston Classic](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonclassic) |  | 
-| StatsForecastCrostonOptimized | [Statsforecast Croston Optimized](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonoptimized) |  | 
-| StatsForecastCrostonSBA | [Statsforecast Croston SBA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonsba) |  | 
-| RFableArima | [fable ARIMA](https://fable.tidyverts.org/reference/ARIMA.html) |  | 
-| RFableETS | [fable ETS](https://fable.tidyverts.org/reference/ETS.html) |  | 
-| RFableNNETAR | [fable NNETAR](https://fable.tidyverts.org/reference/NNETAR.html) |  | 
-| RFableEnsemble | [RFableEnsemble ](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/r_fable/RFableForecastingPipeline.py) |  | 
+| model | source                                                                                                                                                              | covariate support |
+|----------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------|------------|
+| StatsForecastBaselineWindowAverage | [Statsforecast Window Average](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#windowaverage)                                                      |  | 
+| StatsForecastBaselineSeasonalWindowAverage | [Statsforecast Seasonal Window Average](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#seasonalwindowaverage)                                     |  | 
+| StatsForecastBaselineNaive | [Statsforecast Naive](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#naive)                                                                       |  | 
+| StatsForecastBaselineSeasonalNaive | [Statsforecast Seasonal Naive](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#seasonalnaive)                                                      |  | 
+| StatsForecastAutoArima | [Statsforecast AutoARIMA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoarima)                                                               | ✅ | 
+| StatsForecastAutoETS | [Statsforecast AutoETS](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoets)                                                                   |  | 
+| StatsForecastAutoCES | [Statsforecast AutoCES](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autoces)                                                                   |  | 
+| StatsForecastAutoTheta | [Statsforecast AutoTheta](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autotheta)                                                               |  | 
+| StatsForecastAutoTbats | [Statsforecast AutoTBATS](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#autotbats) |  | 
+| StatsForecastAutoMfles | [Statsforecast AutoMFLES](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#automfles) | ✅ | 
+| StatsForecastTSB | [Statsforecast TSB](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#tsb)                                                                           |  | 
+| StatsForecastADIDA | [Statsforecast ADIDA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#adida)                                                                       |  | 
+| StatsForecastIMAPA | [Statsforecast IMAPA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#imapa)                                                                       |  | 
+| StatsForecastCrostonClassic | [Statsforecast Croston Classic](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonclassic)                                                    |  | 
+| StatsForecastCrostonOptimized | [Statsforecast Croston Optimized](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonoptimized)                                                |  | 
+| StatsForecastCrostonSBA | [Statsforecast Croston SBA](https://nixtlaverse.nixtla.io/statsforecast/src/core/models.html#crostonsba)                                                            |  | 
+| RFableArima | [fable ARIMA](https://fable.tidyverts.org/reference/ARIMA.html)                                                                                                     |  | 
+| RFableETS | [fable ETS](https://fable.tidyverts.org/reference/ETS.html)                                                                                                         |  | 
+| RFableNNETAR | [fable NNETAR](https://fable.tidyverts.org/reference/NNETAR.html)                                                                                                   |  | 
+| RFableEnsemble | [RFableEnsemble ](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/r_fable/RFableForecastingPipeline.py)             |  | 
 | RDynamicHarmonicRegression | [RDynamicHarmonicRegression ](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/r_fable/RFableForecastingPipeline.py) |  | 
-| SKTimeTBats | [sktime TBATS](https://www.sktime.net/en/latest/api_reference/auto_generated/sktime.forecasting.tbats.TBATS.html) |  | 
-| SKTimeProphet | [sktime Prophet](https://www.sktime.net/en/latest/api_reference/auto_generated/sktime.forecasting.fbprophet.Prophet.html) |  | 
-| SKTimeLgbmDsDt | [SKTimeLgbmDsDt](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/sktime/SKTimeForecastingPipeline.py) |  | 
+| SKTimeTBats | [sktime TBATS](https://www.sktime.net/en/latest/api_reference/auto_generated/sktime.forecasting.tbats.TBATS.html)                                                   |  | 
+| SKTimeProphet | [sktime Prophet](https://www.sktime.net/en/latest/api_reference/auto_generated/sktime.forecasting.fbprophet.Prophet.html)                                           |  | 
+| SKTimeLgbmDsDt | [SKTimeLgbmDsDt](https://github.com/databricks-industry-solutions/many-model-forecasting/blob/main/mmf_sa/models/sktime/SKTimeForecastingPipeline.py)               |  | 
 
 ## Global
 | model | source | covariate support |

diff --git a/mmf_sa/models/models_conf.yaml b/mmf_sa/models/models_conf.yaml
@@ -84,6 +84,28 @@ models:
       season_length: 7
       decomposition_type: "multiplicative"
 
+  StatsForecastAutoTbats:
+    module: mmf_sa.models.statsforecast.StatsFcForecastingPipeline
+    model_class: StatsFcAutoTbats
+    framework: StatsForecast
+    model_type: local
+    model_spec:
+      season_length: 7
+      use_boxcox: true
+      bc_lower_bound: 0.0
+      bc_upper_bound: 1.0
+      use_trend: true
+      use_damped_trend: true
+      use_arma_errors: true
+
+  StatsForecastAutoMfles:
+    module: mmf_sa.models.statsforecast.StatsFcForecastingPipeline
+    model_class: StatsFcAutoMfles
+    framework: StatsForecast
+    model_type: local
+    model_spec:
+      season_length: 7
+
   StatsForecastTSB:
     module: mmf_sa.models.statsforecast.StatsFcForecastingPipeline
     model_class: StatsFcTSB

diff --git a/mmf_sa/models/statsforecast/StatsFcForecastingPipeline.py b/mmf_sa/models/statsforecast/StatsFcForecastingPipeline.py
@@ -9,6 +9,8 @@
     TSB,
     AutoCES,
     AutoTheta,
+    AutoTBATS,
+    AutoMFLES,
     CrostonClassic,
     CrostonOptimized,
     CrostonSBA,
@@ -206,6 +208,29 @@ def __init__(self, params):
         )
 
 
+class StatsFcAutoTbats(StatsFcForecaster):
+    def __init__(self, params):
+        super().__init__(params)
+        self.model_spec = AutoTBATS(
+            season_length=self.params.model_spec.season_length,
+            use_boxcox=self.params.model_spec.use_boxcox,
+            bc_lower_bound=self.params.model_spec.bc_lower_bound,
+            bc_upper_bound=self.params.model_spec.bc_upper_bound,
+            use_trend=self.params.model_spec.use_trend,
+            use_damped_trend=self.params.model_spec.use_damped_trend,
+            use_arma_errors=self.params.model_spec.use_arma_errors,
+        )
+
+
+class StatsFcAutoMfles(StatsFcForecaster):
+    def __init__(self, params):
+        super().__init__(params)
+        self.model_spec = AutoMFLES(
+            test_size=self.params.prediction_length,
+            season_length=self.params.model_spec.season_length,
+        )
+
+
 class StatsFcTSB(StatsFcForecaster):
     def __init__(self, params):
         super().__init__(params)

diff --git a/requirements.txt b/requirements.txt
@@ -3,7 +3,7 @@ kaleido==0.2.1
 Jinja2
 omegaconf==2.3.0
 numba==0.60.0
-statsforecast==1.7.4
+statsforecast==2.0.0
 missingno==0.5.2
 tbats==1.1.3
 sktime==0.29.0