RollingMean: index values must not have NaT

[alanzhanglei]

[RollingMean: index values must not have NaT]

Code Sample, a copy-pastable example to reproduce your bug.

def rolling_mean(datetime, numeric):
    x = pd.Series(numeric.values, index=datetime.values)
    return apply_rolling_agg_to_series(
        x,
        np.mean,
        self.window_length,
        self.gap,
        self.min_periods,
     )

======datetime======
0       2023-06-18
1       2023-06-18
2       2023-06-18
3       2023-06-18
4              NaT     --> the value [2023-06-19] was set to NaT when the rolling date is 2023-06-18
           ...    
77246          NaT
77247          NaT
77248          NaT
77249          NaT
77250          NaT
Name: wh_item_day_train.day_dt, Length: 77251, dtype: datetime64[ns]

Output of featuretools.show_info()

File "..\lib\site-packages\featuretools\computational_backends\feature_set_calculator.py", line 511, in _calculate_transform_features values = feature_func(*column_data) File "..\lib\site-packages\featuretools\primitives\standard\transform\time_series\rolling_mean.py", line 97, in rolling_mean return apply_rolling_agg_to_series( File "..\lib\site-packages\featuretools\primitives\standard\transform\time_series\utils.py", line 224, in apply_rolling_agg_to_series rolled_series = roll_series_with_gap(series, window_length, gap, min_periods) File "..\lib\site-packages\featuretools\primitives\standard\transform\time_series\utils.py", line 58, in roll_series_with_gap return series.rolling(functional_window_length, min_periods)

ValueError: index values must not have NaT