Time Series Forecast MLP

Hi KNIME community,

I followed an example and created this workflow to do a Sales forecast.

The data I used train the model are the Sales from the last years. As you can see in this line plot there are seasonal fluctuations in the last years but in the predicted data (50-62 on the x axis) no seasonality is taken into account.

Can somebody please help me to fix this. Do I have to prepare the data differently?

Thanks for your responses and kind regards!

hi,
we have some components on the hub that can deal with seasonality. You can first remove the seasonality, train your model, and then add it later again. Here are the nodes on the hub: https://hub.knime.com/search?q=seasonality.
Kind regards,
Alexander

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.