fix(Events): added FixedForwardWindowIndexerNoTruncation class, mak…

…e sure `truncate_if_timeframe_is_smaller==True` outputs numbers instead of nans for the right tail of the time series (#526)

* fix(Events): added `FixedForwardWindowIndexerNoTruncation` class, make sure `truncate_if_timeframe_is_smaller==True` outputs numbers instead of nans for the right tail of the time series

* fix: types

src/fold/events/labeling/fixed.py

@@ -63,6 +63,7 @@ def label_events(
             series=y,
             period=self.time_horizon,
             shift_by=self.shift_by,
+            truncate_end=self.truncate_if_timeframe_is_smaller,
         )
         if self.truncate_if_timeframe_is_smaller:
             cutoff_point = y.index[-self.time_horizon]

src/fold/utils/forward.py

@@ -1,6 +1,56 @@
-from typing import Callable, Optional
+from typing import Callable, Optional, Tuple
 
+import numpy as np
 import pandas as pd
+from pandas.core.indexers.objects import BaseIndexer
+
+
+class FixedForwardWindowIndexerNoTruncation(BaseIndexer):
+    """
+    Creates window boundaries for fixed-length windows that include the current row.
+
+    Examples
+    --------
+    >>> df = pd.DataFrame({'B': [0, 1, 2, np.nan, 4]})
+    >>> df
+         B
+    0  0.0
+    1  1.0
+    2  2.0
+    3  NaN
+    4  4.0
+
+    >>> indexer = pd.api.indexers.FixedForwardWindowIndexer(window_size=2)
+    >>> df.rolling(window=indexer, min_periods=1).sum()
+         B
+    0  1.0
+    1  3.0
+    2  2.0
+    3  4.0
+    4  4.0
+    """
+
+    def get_window_bounds(
+        self,
+        num_values: int = 0,
+        min_periods: Optional[int] = None,
+        center: Optional[bool] = None,
+        closed: Optional[str] = None,
+        step: Optional[int] = None,
+    ) -> Tuple[np.ndarray, np.ndarray]:
+        if center:
+            raise ValueError("Forward-looking windows can't have center=True")
+        if closed is not None:
+            raise ValueError(
+                "Forward-looking windows don't support setting the closed argument"
+            )
+        if step is None:
+            step = 1
+
+        start = np.arange(0, num_values, step, dtype="int64")
+        end = start + self.window_size
+
+        return start, end
 
 
 def create_forward_rolling(
@@ -9,10 +59,17 @@ def create_forward_rolling(
     series: pd.Series,
     period: int,
     shift_by: Optional[int],
+    truncate_end: bool,
 ) -> pd.Series:
     assert period > 0
-    indexer = pd.api.indexers.FixedForwardWindowIndexer(window_size=period)
+    indexer = (
+        pd.api.indexers.FixedForwardWindowIndexer(window_size=period)
+        if truncate_end
+        else FixedForwardWindowIndexerNoTruncation(window_size=period)
+    )
     shift_by = shift_by if shift_by is not None else -1
     assert shift_by < 0
     transformation_func = transformation_func if transformation_func else lambda x: x
-    return agg_func(transformation_func(series).rolling(window=indexer)).shift(shift_by)
+    return agg_func(
+        transformation_func(series).rolling(window=indexer, min_periods=1)
+    ).shift(shift_by)