[ENH] Add hour_of_week option to DateTimeFeatures transformer

sktime/transformations/series/date.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 # copyright: sktime developers, BSD-3-Clause License (see LICENSE file)
 """Extract calendar features from datetimeindex."""
-__author__ = ["danbartl", "KishManani"]
+__author__ = ["danbartl", "KishManani", "VyomkeshVyas"]
 __all__ = ["DateTimeFeatures"]
 
 import warnings
@@ -26,6 +26,7 @@
     ["day", "month", "day", "efficient"],
     ["day", "week", "weekday", "minimal"],
     ["hour", "day", "hour", "minimal"],
+    ["hour", "week", "hour_of_week", "comprehensive"],
     ["minute", "hour", "minute", "minimal"],
     ["second", "minute", "second", "minimal"],
     ["millisecond", "second", "millisecond", "minimal"],
@@ -299,6 +300,8 @@ def _calendar_dummies(x, funcs):
             (x["date_sequence"] - quarter_start) / pd.to_timedelta("1D") + 1
         ).astype(int)
         cd = values
+    elif funcs == "hour_of_week":
+        cd = date_sequence.day_of_week * 24 + date_sequence.hour
     elif funcs == "is_weekend":
         cd = date_sequence.day_of_week > 4
     else:

sktime/transformations/series/tests/test_date.py

@@ -122,6 +122,7 @@ def df_panel():
     "day_of_month",
     "day_of_week",
     "hour_of_day",
+    "hour_of_week",
     "minute_of_hour",
     "second_of_minute",
     "millisecond_of_second",
@@ -286,3 +287,25 @@ def test_month_of_quarter(df_panel):
     yt = t.fit_transform(y)[:13]
     expected = np.array([1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1])
     assert (expected == yt.values).all()
+
+
+@pytest.mark.skipif(
+    not _check_estimator_deps(DateTimeFeatures, severity="none"),
+    reason="skip test if required soft dependencies not available",
+)
+def test_manual_selection_hour_of_week(df_panel):
+    """Tests that "hour_of_week" returns correct result in `manual_selection`."""
+    y = pd.DataFrame(
+        data={"y": range(6)},
+        index=pd.date_range(start="2023-01-01", freq="H", periods=6),
+    )
+    transformer = DateTimeFeatures(
+        manual_selection=["hour_of_week"], keep_original_columns=True
+    )
+
+    yt = transformer.fit_transform(y)
+    expected = pd.DataFrame(
+        data={"y": range(6), "hour_of_week": [144, 145, 146, 147, 148, 149]},
+        index=y.index,
+    )
+    assert_frame_equal(yt, expected)