Initial implementation of the SsaForecaster entry point.

src/DotNetBridge/Bridge.cs

@@ -17,7 +17,7 @@
 using Microsoft.ML.Trainers.FastTree;
 using Microsoft.ML.Trainers.LightGbm;
 using Microsoft.ML.Transforms;
-using Microsoft.ML.TimeSeries;
+using Microsoft.ML.Transforms.TimeSeries;
 
 namespace Microsoft.MachineLearning.DotNetBridge
 {
@@ -329,7 +329,7 @@ private static unsafe int GenericExec(EnvironmentBlock* penv, sbyte* psz, int cd
             //env.ComponentCatalog.RegisterAssembly(typeof(SaveOnnxCommand).Assembly);
             //env.ComponentCatalog.RegisterAssembly(typeof(TimeSeriesProcessingEntryPoints).Assembly);
             //env.ComponentCatalog.RegisterAssembly(typeof(ParquetLoader).Assembly);
-            env.ComponentCatalog.RegisterAssembly(typeof(ForecastExtensions).Assembly);
+            env.ComponentCatalog.RegisterAssembly(typeof(SsaChangePointDetector).Assembly);
 
             using (var ch = host.Start("Executing"))
             {

src/python/nimbusml.pyproj

@@ -89,6 +89,7 @@
     <Compile Include="nimbusml\examples\examples_from_dataframe\GlobalContrastRowScaler_df.py" />
     <Compile Include="nimbusml\examples\examples_from_dataframe\Handler_df.py" />
     <Compile Include="nimbusml\examples\examples_from_dataframe\IidChangePointDetector_df.py" />
+    <Compile Include="nimbusml\examples\examples_from_dataframe\SsaForecaster_df.py" />
     <Compile Include="nimbusml\examples\examples_from_dataframe\SsaChangePointDetector_df.py" />
     <Compile Include="nimbusml\examples\examples_from_dataframe\SsaSpikeDetector_df.py" />
     <Compile Include="nimbusml\examples\examples_from_dataframe\Image_df.py" />
@@ -139,6 +140,7 @@
     <Compile Include="nimbusml\examples\Handler.py" />
     <Compile Include="nimbusml\examples\Hinge.py" />
     <Compile Include="nimbusml\examples\IidChangePointDetector.py" />
+    <Compile Include="nimbusml\examples\SsaForecaster.py" />
     <Compile Include="nimbusml\examples\SsaChangePointDetector.py" />
     <Compile Include="nimbusml\examples\SsaSpikeDetector.py" />
     <Compile Include="nimbusml\examples\Image.py" />
@@ -235,6 +237,7 @@
     <Compile Include="nimbusml\internal\core\timeseries\iidchangepointdetector.py" />
     <Compile Include="nimbusml\internal\core\timeseries\iidspikedetector.py" />
     <Compile Include="nimbusml\internal\core\timeseries\ssachangepointdetector.py" />
+    <Compile Include="nimbusml\internal\core\timeseries\ssaforecaster.py" />
     <Compile Include="nimbusml\internal\core\timeseries\ssaspikedetector.py" />
     <Compile Include="nimbusml\internal\core\timeseries\__init__.py" />
     <Compile Include="nimbusml\internal\entrypoints\data_customtextloader.py" />
@@ -277,6 +280,7 @@
     <Compile Include="nimbusml\internal\entrypoints\timeseriesprocessingentrypoints_pvaluetransform.py" />
     <Compile Include="nimbusml\internal\entrypoints\timeseriesprocessingentrypoints_slidingwindowtransform.py" />
     <Compile Include="nimbusml\internal\entrypoints\timeseriesprocessingentrypoints_ssachangepointdetector.py" />
+    <Compile Include="nimbusml\internal\entrypoints\timeseriesprocessingentrypoints_ssaforecasting.py" />
     <Compile Include="nimbusml\internal\entrypoints\timeseriesprocessingentrypoints_ssaspikedetector.py" />
     <Compile Include="nimbusml\internal\entrypoints\trainers_averagedperceptronbinaryclassifier.py" />
     <Compile Include="nimbusml\internal\entrypoints\trainers_fastforestbinaryclassifier.py" />
@@ -586,13 +590,15 @@
     <Compile Include="nimbusml\tests\pipeline\test_pipeline_subclassing.py" />
     <Compile Include="nimbusml\tests\preprocessing\normalization\test_meanvariancescaler.py" />
     <Compile Include="nimbusml\tests\timeseries\test_iidchangepointdetector.py" />
+    <Compile Include="nimbusml\tests\timeseries\test_ssaforecaster.py" />
     <Compile Include="nimbusml\tests\timeseries\test_ssachangepointdetector.py" />
     <Compile Include="nimbusml\tests\timeseries\test_ssaspikedetector.py" />
     <Compile Include="nimbusml\tests\timeseries\test_iidspikedetector.py" />
     <Compile Include="nimbusml\tests\timeseries\__init__.py" />
     <Compile Include="nimbusml\timeseries\iidchangepointdetector.py" />
     <Compile Include="nimbusml\timeseries\iidspikedetector.py" />
     <Compile Include="nimbusml\timeseries\ssachangepointdetector.py" />
+    <Compile Include="nimbusml\timeseries\ssaforecaster.py" />
     <Compile Include="nimbusml\timeseries\ssaspikedetector.py" />
     <Compile Include="nimbusml\timeseries\__init__.py" />
     <Compile Include="tests\test_estimator_checks.py" />

src/python/nimbusml/examples/SsaForecaster.py

@@ -0,0 +1,47 @@
+###############################################################################
+# SsaForecaster
+import pandas as pd
+from nimbusml import Pipeline, FileDataStream
+from nimbusml.datasets import get_dataset
+from nimbusml.timeseries import SsaForecaster
+
+# data input (as a FileDataStream)
+path = get_dataset('timeseries').as_filepath()
+
+data = FileDataStream.read_csv(path)
+print(data.head())
+#      t1    t2      t3
+# 0  0.01  0.01  0.0100
+# 1  0.02  0.02  0.0200
+# 2  0.03  0.03  0.0200
+# 3  0.03  0.03  0.0250
+# 4  0.03  0.03  0.0005
+
+# define the training pipeline
+pipeline = Pipeline([
+    SsaForecaster(series_length=6,
+                  train_size=8,
+                  window_size=3,
+                  horizon=2,
+                  columns={'t2_fc': 't2'})
+])
+
+result = pipeline.fit_transform(data)
+
+pd.set_option('display.float_format', lambda x: '%.2f' % x)
+print(result)
+
+# Output
+#
+#     t1    t2    t3  t2_fc.0  t2_fc.1
+# 0 0.01  0.01  0.01     0.10     0.12
+# 1 0.02  0.02  0.02     0.06     0.08
+# 2 0.03  0.03  0.02     0.04     0.05
+# 3 0.03  0.03  0.03     0.05     0.06
+# 4 0.03  0.03  0.00     0.05     0.07
+# 5 0.03  0.05  0.01     0.06     0.08
+# 6 0.05  0.07  0.05     0.09     0.12
+# 7 0.07  0.09  0.09     0.12     0.16
+# 8 0.09 99.00 99.00    57.92    82.88
+# 9 1.10  0.10  0.10    60.50    77.18
+

src/python/nimbusml/examples/examples_from_dataframe/SsaForecaster_df.py

@@ -0,0 +1,86 @@
+###############################################################################
+# SsaForecaster
+import numpy as np
+import pandas as pd
+from nimbusml.timeseries import SsaForecaster
+
+# This example creates a time series (list of data with the
+# i-th element corresponding to the i-th time slot).
+
+# Generate sample series data with a recurring pattern
+seasonality_size = 5
+seasonal_data = np.arange(seasonality_size)
+
+data = np.tile(seasonal_data, 3)
+X_train = pd.Series(data, name="ts")
+
+# X_train looks like this
+# 0       0
+# 1       1
+# 2       2
+# 3       3
+# 4       4
+# 5       0
+# 6       1
+# 7       2
+# 8       3
+# 9       4
+# 10      0
+# 11      1
+# 12      2
+# 13      3
+# 14      4
+
+x_test = X_train.copy()
+x_test[-3:] = [100, 110, 120]
+
+# x_test looks like this
+# 0       0
+# 1       1
+# 2       2
+# 3       3
+# 4       4
+# 5       0
+# 6       1
+# 7       2
+# 8       3
+# 9       4
+# 10      0
+# 11      1
+# 12    100
+# 13    110
+# 14    120
+
+training_seasons = 3
+training_size = seasonality_size * training_seasons
+
+forecaster = SsaForecaster(series_length=8,
+                           train_size=training_size,
+                           window_size=seasonality_size + 1,
+                           horizon=4) <<   {'fc': 'ts'}
+
+forecaster.fit(X_train, verbose=1)
+data = forecaster.transform(x_test)
+
+pd.set_option('display.float_format', lambda x: '%.2f' % x)
+print(data)
+
+# The fc.x columns are the forecasts
+# given the input in the ts column.
+#
+#      ts  fc.0  fc.1  fc.2  fc.3
+# 0     0  1.00  2.00  3.00  4.00
+# 1     1  2.00  3.00  4.00 -0.00
+# 2     2  3.00  4.00 -0.00  1.00
+# 3     3  4.00 -0.00  1.00  2.00
+# 4     4 -0.00  1.00  2.00  3.00
+# 5     0  1.00  2.00  3.00  4.00
+# 6     1  2.00  3.00  4.00 -0.00
+# 7     2  3.00  4.00 -0.00  1.00
+# 8     3  4.00 -0.00  1.00  2.00
+# 9     4 -0.00  1.00  2.00  3.00
+# 10    0  1.00  2.00  3.00  4.00
+# 11    1  2.00  3.00  4.00 -0.00
+# 12  100  3.00  4.00  0.00  1.00
+# 13  110  4.00 -0.00  1.00 75.50
+# 14  120 -0.00  1.00 83.67 83.25

src/python/nimbusml/internal/core/timeseries/ssaforecaster.py

@@ -0,0 +1,141 @@
+# --------------------------------------------------------------------------------------------
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License.
+# --------------------------------------------------------------------------------------------
+# - Generated by tools/entrypoint_compiler.py: do not edit by hand
+"""
+SsaForecaster
+"""
+
+__all__ = ["SsaForecaster"]
+
+
+from ...entrypoints.timeseriesprocessingentrypoints_ssaforecasting import \
+    timeseriesprocessingentrypoints_ssaforecasting
+from ...utils.utils import trace
+from ..base_pipeline_item import BasePipelineItem, DefaultSignature
+
+
+class SsaForecaster(BasePipelineItem, DefaultSignature):
+    """
+    **Description**
+        This transform forecasts using Singular Spectrum Analysis (SSA).
+
+    :param window_size: The length of the window on the series for building the
+        trajectory matrix (parameter L).
+
+    :param series_length: The length of series that is kept in buffer for
+        modeling (parameter N).
+
+    :param train_size: The length of series from the begining used for
+        training.
+
+    :param horizon: The number of values to forecast.
+
+    :param confidence_level: The confidence level in [0, 1) for forecasting.
+
+    :param variable_horizon: Set this to true horizon will change at prediction
+        time.
+
+    :param confidence_lower_bound_column: The name of the confidence interval
+        lower bound column.
+
+    :param confidence_upper_bound_column: The name of the confidence interval
+        upper bound column.
+
+    :param rank_selection_method: The rank selection method.
+
+    :param rank: The desired rank of the subspace used for SSA projection
+        (parameter r). This parameter should be in the range in [1,
+        windowSize]. If set to null, the rank is automatically determined based
+        on prediction error minimization.
+
+    :param max_rank: The maximum rank considered during the rank selection
+        process. If not provided (i.e. set to null), it is set to windowSize -
+        1.
+
+    :param should_stabilize: The flag determining whether the model should be
+        stabilized.
+
+    :param should_maintain_info: The flag determining whether the meta
+        information for the model needs to be maintained.
+
+    :param max_growth: The maximum growth on the exponential trend.
+
+    :param discount_factor: The discount factor in [0,1] used for online
+        updates.
+
+    :param is_adaptive: The flag determing whether the model is adaptive.
+
+    :param params: Additional arguments sent to compute engine.
+
+    """
+
+    @trace
+    def __init__(
+            self,
+            window_size=0,
+            series_length=0,
+            train_size=0,
+            horizon=0,
+            confidence_level=0.95,
+            variable_horizon=False,
+            confidence_lower_bound_column=None,
+            confidence_upper_bound_column=None,
+            rank_selection_method='Exact',
+            rank=None,
+            max_rank=None,
+            should_stabilize=True,
+            should_maintain_info=False,
+            max_growth=None,
+            discount_factor=1.0,
+            is_adaptive=False,
+            **params):
+        BasePipelineItem.__init__(
+            self, type='transform', **params)
+
+        self.window_size = window_size
+        self.series_length = series_length
+        self.train_size = train_size
+        self.horizon = horizon
+        self.confidence_level = confidence_level
+        self.variable_horizon = variable_horizon
+        self.confidence_lower_bound_column = confidence_lower_bound_column
+        self.confidence_upper_bound_column = confidence_upper_bound_column
+        self.rank_selection_method = rank_selection_method
+        self.rank = rank
+        self.max_rank = max_rank
+        self.should_stabilize = should_stabilize
+        self.should_maintain_info = should_maintain_info
+        self.max_growth = max_growth
+        self.discount_factor = discount_factor
+        self.is_adaptive = is_adaptive
+
+    @property
+    def _entrypoint(self):
+        return timeseriesprocessingentrypoints_ssaforecasting
+
+    @trace
+    def _get_node(self, **all_args):
+        algo_args = dict(
+            source=self.source,
+            name=self._name_or_source,
+            window_size=self.window_size,
+            series_length=self.series_length,
+            train_size=self.train_size,
+            horizon=self.horizon,
+            confidence_level=self.confidence_level,
+            variable_horizon=self.variable_horizon,
+            confidence_lower_bound_column=self.confidence_lower_bound_column,
+            confidence_upper_bound_column=self.confidence_upper_bound_column,
+            rank_selection_method=self.rank_selection_method,
+            rank=self.rank,
+            max_rank=self.max_rank,
+            should_stabilize=self.should_stabilize,
+            should_maintain_info=self.should_maintain_info,
+            max_growth=self.max_growth,
+            discount_factor=self.discount_factor,
+            is_adaptive=self.is_adaptive)
+
+        all_args.update(algo_args)
+        return self._entrypoint(**all_args)