0.3.7

Author: winedarksea

TODO.md

@@ -15,12 +15,17 @@
 * Forecasts are desired for the future immediately following the most recent data.
 
 # Latest
-* back_forecast for forecast on training data
-* Mosaic ensembles can now be used beyond training forecast_length and for shorter lengths too
-* best_model_name, best_model_params, and best_model_transformation_params AutoTS attributes now available
-* mean, median, and ffill NaN now handle fully NaN series by returning 0.
-* fixed bug that was causing mosaic generalization to fail if ffill/bfill handled all missing values
-* STLFilter and HPFilter and convolution_filter Transformers added
+* bug fix in fake_date imputation
+* bug fix in Round
+* make SinTrend fail if it fails on all series (may revert this)
+* load_linear and load_sine artificial datasets
+* new NVAR model based on https://github.com/quantinfo/ng-rc-paper-code/
+* tuning retrieve_regressor to allow it to better work with multioutput and univariate
+* expand GluonTS models included
+* GluonTS now works on univariate inputs
+* GluonTS now works with regressors
+* fixed bug where model_count wrong for mosaic ensembles
+* fixed bug in VECM that meant it didn't couldn't utilize future_regressor
 
 # Errors: 
 DynamicFactor holidays 	Exceptions 'numpy.ndarray' object has no attribute 'values'

autots/__init__.py

@@ -11,6 +11,7 @@
     load_weekly,
     load_weekdays,
     load_live_daily,
+    load_linear,
 )
 
 from autots.evaluator.auto_ts import AutoTS
@@ -19,7 +20,7 @@
 from autots.tools.regressor import create_lagged_regressor
 from autots.evaluator.auto_model import model_forecast
 
-__version__ = '0.3.6'
+__version__ = '0.3.7'
 
 TransformTS = GeneralTransformer
 
@@ -31,6 +32,7 @@
     'load_weekly',
     'load_weekdays',
     'load_live_daily',
+    'load_linear',
     'AutoTS',
     'TransformTS',
     'GeneralTransformer',

autots/datasets/__init__.py

@@ -10,6 +10,8 @@
     load_weekly,
     load_weekdays,
     load_zeroes,
+    load_linear,
+    load_sine,
 )
 
 __all__ = [
@@ -21,4 +23,6 @@
     'load_weekdays',
     'load_live_daily',
     'load_zeroes',
+    'load_linear',
+    'load_sine',
 ]

autots/datasets/_base.py

@@ -391,3 +391,47 @@ def load_zeroes(long=False, shape=None, start_date: str = "2021-01-01"):
             id_vars=['datetime'], var_name='series_id', value_name='value'
         )
         return df_long
+
+
+def load_linear(long=False, shape=None, start_date: str = "2021-01-01"):
+    """Create a dataset of just zeroes for testing edge case."""
+    if shape is None:
+        shape = (500, 5)
+    df_wide = pd.DataFrame(
+        np.ones(shape), index=pd.date_range(start_date, periods=shape[0], freq="D")
+    )
+    df_wide = (df_wide * list(range(0, shape[1]))).cumsum()
+    if not long:
+        return df_wide
+    else:
+        df_wide.index.name = "datetime"
+        df_long = df_wide.reset_index(drop=False).melt(
+            id_vars=['datetime'], var_name='series_id', value_name='value'
+        )
+        return df_long
+
+
+def load_sine(long=False, shape=None, start_date: str = "2021-01-01"):
+    """Create a dataset of just zeroes for testing edge case."""
+    if shape is None:
+        shape = (500, 5)
+    df_wide = pd.DataFrame(
+        np.ones(shape),
+        index=pd.date_range(start_date, periods=shape[0], freq="D"),
+        columns=range(shape[1]),
+    )
+    X = pd.to_numeric(df_wide.index, errors='coerce', downcast='integer').values
+
+    def sin_func(a, X):
+        return a * np.sin(1 * X) + a
+
+    for column in df_wide.columns:
+        df_wide[column] = sin_func(column, X)
+    if not long:
+        return df_wide
+    else:
+        df_wide.index.name = "datetime"
+        df_long = df_wide.reset_index(drop=False).melt(
+            id_vars=['datetime'], var_name='series_id', value_name='value'
+        )
+        return df_long

autots/evaluator/auto_model.py

@@ -30,6 +30,7 @@
     SeasonalNaive,
     ZeroesNaive,
     Motif,
+    NVAR,
 )
 from autots.models.statsmodels import (
     GLS,
@@ -425,6 +426,15 @@ def ModelMonster(
             **parameters,
         )
         return model
+    elif model == 'NVAR':
+        model = NVAR(
+            frequency=frequency,
+            prediction_interval=prediction_interval,
+            random_seed=random_seed,
+            verbose=verbose,
+            **parameters,
+        )
+        return model
     else:
         raise AttributeError(
             ("Model String '{}' not a recognized model type").format(model)
@@ -987,6 +997,7 @@ def TemplateWizard(
     # template = unpack_ensemble_models(template, template_cols, keep_ensemble = False)
 
     for index, row in template.iterrows():
+        template_start_time = datetime.datetime.now()
         try:
             model_str = row['Model']
             parameter_dict = json.loads(row['ModelParameters'])
@@ -1065,11 +1076,6 @@ def TemplateWizard(
                 df_forecast.model_parameters,
                 df_forecast.transformation_parameters,
             )
-            total_runtime = (
-                df_forecast.fit_runtime
-                + df_forecast.predict_runtime  # noqa W503
-                + df_forecast.transformation_runtime  # noqa W503
-            )
             result = pd.DataFrame(
                 {
                     'ID': model_id,
@@ -1081,7 +1087,7 @@ def TemplateWizard(
                     'TransformationRuntime': df_forecast.transformation_runtime,
                     'FitRuntime': df_forecast.fit_runtime,
                     'PredictRuntime': df_forecast.predict_runtime,
-                    'TotalRuntime': total_runtime,
+                    'TotalRuntime': datetime.datetime.now() - template_start_time,
                     'Ensemble': ensemble_input,
                     'Exceptions': np.nan,
                     'Runs': 1,
@@ -1147,6 +1153,7 @@ def TemplateWizard(
 
         except KeyboardInterrupt:
             if model_interrupt:
+                fit_runtime = datetime.datetime.now() - template_start_time
                 result = pd.DataFrame(
                     {
                         'ID': create_model_id(
@@ -1157,9 +1164,9 @@ def TemplateWizard(
                         'TransformationParameters': json.dumps(transformation_dict),
                         'Ensemble': ensemble_input,
                         'TransformationRuntime': datetime.timedelta(0),
-                        'FitRuntime': datetime.timedelta(0),
+                        'FitRuntime': fit_runtime,
                         'PredictRuntime': datetime.timedelta(0),
-                        'TotalRuntime': datetime.timedelta(0),
+                        'TotalRuntime': fit_runtime,
                         'Exceptions': "KeyboardInterrupt by user",
                         'Runs': 1,
                         'Generation': current_generation,
@@ -1194,7 +1201,7 @@ def TemplateWizard(
                             (repr(e)), template_result.model_count, model_str
                         )
                     )
-
+            fit_runtime = datetime.datetime.now() - template_start_time
             result = pd.DataFrame(
                 {
                     'ID': create_model_id(
@@ -1205,9 +1212,9 @@ def TemplateWizard(
                     'TransformationParameters': json.dumps(transformation_dict),
                     'Ensemble': ensemble_input,
                     'TransformationRuntime': datetime.timedelta(0),
-                    'FitRuntime': datetime.timedelta(0),
+                    'FitRuntime': fit_runtime,
                     'PredictRuntime': datetime.timedelta(0),
-                    'TotalRuntime': datetime.timedelta(0),
+                    'TotalRuntime': fit_runtime,
                     'Exceptions': repr(e),
                     'Runs': 1,
                     'Generation': current_generation,

autots/evaluator/auto_ts.py

@@ -880,8 +880,8 @@ def fit(
                     min_allowed_train_percent=self.min_allowed_train_percent,
                     verbose=self.verbose,
                 )
-                self.validation_train_indexes.append(val_df_train)
-                self.validation_test_indexes.append(val_df_test)
+                self.validation_train_indexes.append(val_df_train.index)
+                self.validation_test_indexes.append(val_df_test.index)
                 if self.verbose >= 2:
                     print(f'Validation index is {val_df_train.index}')
 
@@ -901,7 +901,7 @@ def fit(
                 if self.introduce_na or (self.introduce_na is None and self._nan_tail):
                     nan_frac = val_df_train.shape[1] / num_validations
                     val_df_train.iloc[
-                        -1, int(nan_frac * y) : int(nan_frac * (y + 1))
+                        -2:, int(nan_frac * y) : int(nan_frac * (y + 1))
                     ] = np.nan
 
                 # run validation template on current slice

autots/evaluator/metrics.py

@@ -56,7 +56,9 @@ def SPL(A, F, df_train, quantile):
     """Scaled pinball loss."""
     # scaler = df_train.tail(1000).diff().abs().mean(axis=0)
     # scaler = np.abs(np.diff(df_train[-1000:], axis=0)).mean(axis=0)
-    scaler = np.nanmean(np.abs(np.diff(df_train[-1000:], axis=0)), axis=0)
+    with warnings.catch_warnings():
+        warnings.simplefilter("ignore", category=RuntimeWarning)
+        scaler = np.nanmean(np.abs(np.diff(df_train[-1000:], axis=0)), axis=0)
     # need to handle zeroes to prevent div 0 errors.
     # this will tend to make that series irrelevant to the overall evaluation
     fill_val = np.nanmax(scaler)