kumar781

1961-01 451.578434 1961-02 427.259888 1961-03 464.704207 1961-04 500.104578 1961-05 508.663801 1961-06 567.717163 1961-07 647.899372 1961-08 632.464688 1961-09 534.377506 Using sktime auto-ets

> ```python > from sktime.datasets import load_airline > y = load_airline() > > from statsmodels.tsa.exponential_smoothing.ets import ETSModel > forecaster = ETSModel(y, trend="add", seasonal="add", seasonal_periods=12) > > fitted_forecaster = forecaster.fit() >...

However when I run on Linux machine, i get different results: 1961-01 451.578741 1961-02 427.260109 1961-03 464.704148 1961-04 500.104779 1961-05 508.664083 1961-06 567.717524 1961-07 647.899930 1961-08 632.465243 1961-09 534.378124

I tried introducing a random_state param in statsmodel adapter get_predictions(), still the forecast values fluctuate across environments. Any other workarounds you have in mind?