kumar781

I tried 'pip3 install --upgrade setuptools' and then 'pip install supplychainpy' on jupyter notebook (windows 10 64 bit). The notebook basically freezes. Could anyone help me on this?

Just checked, with latest sktime==0.28.0. I see fluctuations in forecasts on different machines (windows,linux,etc.) I used the same input data across different machines.

I have tried with random_state=None and random_state=42. But still get variations in results. Ideally this is my model with hyper-params: model = AutoETS(auto=False, additive_only=True, sp=12, trend='add', damped_trend=True, random_state=42, allow_multiplicative_trend=False, bounds={"damping_trend":...

pandas version: 1.1.5 statsmodels version: 0.13.2 I have maintained the above versions in all of the machines

 This screenshot shows differences in forecast values in two machines

Sorry for the confusion, let me make it simple for you. Here is the reference code: %python from sktime.forecasting.ets import AutoETS from sktime.utils.plotting import plot_series import pandas as pd import...

Ideally i would want the same forecast values to be reproducible across environments

> I have two questions for the script you provided as reproducible example. > > 1. The script does not run for me. It fails with `ValueError: Cannot compute initial...

@fkiraly @yarnabrina i think i addressed the issues you pointed out previously. With the new code there are still forecast fluctuations.

Tried with sktime 0.28.0, still see forecast value differences there.