Can we run model deviation task or other tasks other then forecasting task on time series models in Predictive Factory 3.0

[sandeep_mekala]

Hi

I am using Predict Factory 3.0 tool to automate my model management. I have imported time series model into my project and I am able to run forecasting tasks. But when I try to configure other tasks like model application task or model deviation task. It is not detecting time series models. So, is it the case that I can run only forecasting task on time series models not other tasks like model application, deviation and retrianing tasks on time series models?

What I want to do is: I want to apply forecasting only if the model is confident enough having less MAPE. How can I achieve this?

Thanks

[achab · Accepted solution]

Yes for time series models, only time series forecasting task.

The rationale is below:

- Each time you use the TSF task, a new training is performed.

- No need then to check for the deviations, as training is always performed

- You can output the forecasting results to a new table which meets the same purpose as the model application task.

In a nutshell, the TSF task contains all that you need 🙂

I'll raise to our doc team that we should do a better job explaining which kind of tasks apply to which models and what is the typical matching or task flow users have to do.

Best regards,

Antoine

PS: we have internal communities setup on SAP JAM to help SAP employees, if interested I'll invite you.

[sandeep_mekala · Accepted solution]

Thank you for the reply. Sure, I am interested.

[achab · Accepted solution]

Done! Also sent the email to the doc team.