Merge remote-tracking branch 'origin/master' into issue-106

CHANGELOG.md

@@ -12,6 +12,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Examples to TSDataset methods with doctest ([#92](https://github.com/tinkoff-ai/etna-ts/pull/92))
 - WandbLogger ([#71](https://github.com/tinkoff-ai/etna-ts/pull/71))
 - Pipeline ([#78](https://github.com/tinkoff-ai/etna-ts/pull/78))
+- 'is_weekend' feature in DateFlagsTransform ([#101](https://github.com/tinkoff-ai/etna-ts/pull/101))
 
 ### Changed
 - SklearnTransform out column names ([#99](https://github.com/tinkoff-ai/etna-ts/pull/99))

etna/transforms/datetime_flags.py

@@ -25,6 +25,7 @@ def __init__(
         week_number_in_year: Optional[bool] = False,
         month_number_in_year: Optional[bool] = False,
         year_number: Optional[bool] = False,
+        is_weekend: Optional[bool] = True,
         special_days_in_week: Sequence[int] = (),
         special_days_in_month: Sequence[int] = (),
     ):
@@ -46,6 +47,8 @@ def __init__(
             if True, add column "month_number_in_year" with month info to feature dataframe in transform
         year_number:
             if True, add column "year_number" with year info to feature dataframe in transform
+        is_weekend:
+            if True: add column "regressor_is_weekend" with weekends flags to feature dataframe in transform
         special_days_in_week:
             list of weekdays number (from [0, 6]) that should be interpreted as special ones, if given add column
             "special_days_in_week" with flag that shows given date is a special day
@@ -77,14 +80,15 @@ def __init__(
                 week_number_in_year,
                 month_number_in_year,
                 year_number,
+                is_weekend,
                 special_days_in_week,
                 special_days_in_month,
             ]
         ):
             raise ValueError(
                 f"{type(self).__name__} feature does nothing with given init args configuration, "
                 f"at least one of day_number_in_week, day_number_in_month, week_number_in_month, "
-                f"week_number_in_year, month_number_in_year, year_number should be True or any of "
+                f"week_number_in_year, month_number_in_year, year_number, is_weekend should be True or any of "
                 f"specyal_days_in_week, special_days_in_month should be not empty."
             )
 
@@ -94,6 +98,7 @@ def __init__(
         self.week_number_in_year = week_number_in_year
         self.month_number_in_year = month_number_in_year
         self.year_number = year_number
+        self.is_weekend = is_weekend
 
         self.special_days_in_week = special_days_in_week
         self.special_days_in_month = special_days_in_month
@@ -137,6 +142,9 @@ def transform(self, df: pd.DataFrame) -> pd.DataFrame:
         if self.year_number:
             features["year_number"] = self._get_year(timestamp_series=timestamp_series)
 
+        if self.is_weekend:
+            features["is_weekend"] = self._get_weekends(timestamp_series=timestamp_series)
+
         if self.special_days_in_week:
             features["special_days_in_week"] = self._get_special_day_in_week(
                 special_days=self.special_days_in_week, timestamp_series=timestamp_series
@@ -233,6 +241,12 @@ def _get_year(timestamp_series: pd.Series) -> np.array:
         """Generate an array with the week number in the year."""
         return timestamp_series.apply(lambda x: x.year).values
 
+    @staticmethod
+    def _get_weekends(timestamp_series: pd.Series) -> np.array:
+        """Generate an array with the weekends flags."""
+        weekend_days = (5, 6)
+        return timestamp_series.apply(lambda x: x.weekday() in weekend_days).values
+
 
 class TimeFlagsTransform(Transform):
     """Class for holding time transform."""

tests/test_model_selection/test_backtest.py

@@ -258,7 +258,7 @@ def test_get_metrics_interface(aggregate_metrics: bool, expected_columns: List[s
 
 def test_get_forecasts_interface_daily(big_daily_example_tsdf: TSDataset):
     """Test interface of TimeSeriesCrossValidation.get_forecasts"""
-    date_flags = DateFlagsTransform(day_number_in_week=True, day_number_in_month=True)
+    date_flags = DateFlagsTransform(day_number_in_week=True, day_number_in_month=True, is_weekend=False)
     tsvc = _fit_backtest_pipeline(
         model=CatBoostModelMultiSegment(), horizon=24, ts=big_daily_example_tsdf, transforms=[date_flags]
     )
@@ -271,7 +271,7 @@ def test_get_forecasts_interface_daily(big_daily_example_tsdf: TSDataset):
 
 def test_get_forecasts_interface_hours(example_tsdf: TSDataset):
     """Test interface of TimeSeriesCrossValidation.get_forecasts"""
-    date_flags = DateFlagsTransform(day_number_in_week=True, day_number_in_month=True)
+    date_flags = DateFlagsTransform(day_number_in_week=True, day_number_in_month=True, is_weekend=False)
     tsvc = _fit_backtest_pipeline(
         model=CatBoostModelMultiSegment(), horizon=24, ts=example_tsdf, transforms=[date_flags]
     )

tests/test_transforms/test_dateflags_transform.py

@@ -10,16 +10,18 @@
 
 from etna.transforms.datetime_flags import DateFlagsTransform
 
+WEEKEND_DAYS = (5, 6)
 SPECIAL_DAYS = [1, 4]
 INIT_PARAMS_TEMPLATE = {
     "day_number_in_week": False,
     "day_number_in_month": False,
     "week_number_in_year": False,
     "week_number_in_month": False,
     "month_number_in_year": False,
+    "year_number": False,
+    "is_weekend": False,
     "special_days_in_week": (),
     "special_days_in_month": (),
-    "year_number": False,
 }
 
 
@@ -46,6 +48,7 @@ def dateflags_true_df() -> pd.DataFrame:
         df["week_number_in_month"] = df["timestamp"].apply(
             lambda x: int(x.weekday() < (x - timedelta(days=x.day - 1)).weekday()) + (x.day - 1) // 7 + 1
         )
+        df["is_weekend"] = df["timestamp"].apply(lambda x: x.weekday() in WEEKEND_DAYS)
         df["special_days_in_week"] = df["day_number_in_week"].apply(lambda x: x in SPECIAL_DAYS)
         df["special_days_in_month"] = df["day_number_in_month"].apply(lambda x: x in SPECIAL_DAYS)
 
@@ -93,6 +96,7 @@ def test_invalid_arguments_configuration():
             week_number_in_year=False,
             month_number_in_year=False,
             year_number=False,
+            is_weekend=False,
             special_days_in_week=(),
             special_days_in_month=(),
         )
@@ -108,13 +112,14 @@ def test_repr():
         week_number_in_year=False,
         month_number_in_year=True,
         year_number=True,
+        is_weekend=True,
         special_days_in_week=(1, 2),
         special_days_in_month=(12,),
     )
     transform_repr = transform.__repr__()
     true_repr = (
         f"{transform_class_repr}(day_number_in_week = True, day_number_in_month = True, week_number_in_month = False, "
-        f"week_number_in_year = False, month_number_in_year = True, year_number = True, special_days_in_week = (1, 2), "
+        f"week_number_in_year = False, month_number_in_year = True, year_number = True, is_weekend = True, special_days_in_week = (1, 2), "
         f"special_days_in_month = (12,), )"
     )
     assert transform_repr == true_repr
@@ -129,13 +134,15 @@ def test_repr():
         ["week_number_in_month"],
         ["month_number_in_year"],
         ["year_number"],
+        ["is_weekend"],
         [
             "day_number_in_week",
             "day_number_in_month",
             "week_number_in_year",
             "week_number_in_month",
             "month_number_in_year",
             "year_number",
+            "is_weekend",
         ],
     ),
 )
@@ -192,6 +199,7 @@ def test_interface_correct_tuple_args(true_params: List[str], train_df: pd.DataF
         {"week_number_in_month": True},
         {"month_number_in_year": True},
         {"year_number": True},
+        {"is_weekend": True},
         {"special_days_in_week": SPECIAL_DAYS},
         {"special_days_in_month": SPECIAL_DAYS},
     ),