In case of models created with the one-step-fitting approach it seams that the Model Writer node save / attach only one data record (several times) to the model instead of saving / attaching all different data records used for model fitting.
See attached workflow.
the workflow has only one single estimated primary model named "TEST-Model". But different secondary models and timeseries data sets.
Firstly: the database structure is not designed for "one primary" vs. "many time series"
Secondly: Even the estimated secondary models do not correspond logically with the time series (i.e. estimated secondary model ids point to different time series)
That's the structure for models created by "One-Step-Fit". There is a global model (primary + its secondary models), that is combined with each time series.
I know, that the DB is not designed for "One-Step-Fit" models. So we have to decide, if that feature is important enough to implement it in the DB.
the database is not designed for one-step-fitting models
We should consider to make the necessary adjustments to the DB somewhen.