search

sambofra / bnstruct

R package for Bayesian Network Structure Learning
GNU General Public License v3.0
17 stars 9 forks source link

Variable length sequence #27

Closed ae-droid closed 10 months ago

ae-droid commented 2 years ago

Hi, thank you for the package, it is brilliant. I have a question about learning the structures for the sequences with varying length. For example, assume that we are recording 4 outcomes for a group of students and student A has observations for 5 consecutive time steps, student B has 8 consecutive time steps etc. How the package handles those cases for discovering DBN? I am not sure if i am missing something you have already shared in the manual. If this is the case apologies, otherwise any comment of yours qould be extremely helpful. Thank you.

albertofranzin commented 2 years ago

Hi, and thank you.

The input dataset is essentially assumed to be a table, so the number of time steps considered is the same for all the observations, and it corresponds to the maximum value available. The missing time steps, e.g. time steps 6, 7 and 8 for student A, are considered missing values (NAs), and you have to indicate them as such.

Hope this helps.

ae-droid commented 2 years ago

Thank you very much for your respond, best.