data collection duration for a new user

[nstsj]

Hi there! I am currently working on reproducing the experiments from the original GLOBEM paper and i came up with a follow-up question. For introducing a new user who wasn't the part of the training set, how many months of data would you suggest gathering to ensure reliable predictions from the Reorder and Chikersal algorithms? I'm considering a range of 3-6 months based on data variability and would greatly appreciate your expert opinion.

Kind regards, Ana