What type of AI model do you run?
We use multiple open source libraries, mainly we use Generalized additive model, we train the AI with the provided data history and generate the next N time series datapoints.
Every forecast instance is unique and trained once (per forecast).
Model settings are automatically calculated based on the type of history data (length, type of fields (integer, float etc.)) with the aim to have better results.
Please note that our AI doesn't correlate multiple fields, because every field is trained "alone".
If you have a specific use case or you need some help you can contact us firstname.lastname@example.org