Fitted values in sktime

[saumya-verma-DS]

My code is
from sklearn.neighbors import KNeighborsRegressor
from sktime.forecasting.compose import make_reduction
regressor = KNeighborsRegressor(n_neighbors=5)
forecaster = make_reduction(regressor, strategy="direct", window_length=6)
forecaster.fit(train_targets,fh =[-1])
predictions_train = forecaster.predict(fh = [-1])

and the error I am getting NotImplementedError: In-sample predictions are not implemented. But there is no other way to get fitted values in sktime.
My main problem is that I would need fitted values for my models. How can I get those?

[fkiraly]

You could try the alternative implementatoin, DirectReductionForecaster - it is slower, but can return in-sample forecasts.

[saumya-verma-DS]

Thankyou for the help. I will try this out