Tutorial for multitask learning workflow

[daria-dc]

Hi,

I would like to try the multitask learning workflow that was used in the publication "Multi-study inference of regulatory networks for more accurate models of gene regulation"... however, I can not find any instructions or tutorial how to run it. Could you provide a tutorial similar to the one for single task learning?

Furthermore, do you know if very different sample sizes between data sets affect the outcome?

Thanks in advance!

[asistradition]

I'm still working up the tutorial and documentation for multitask learning. It should be available with the next minor version release (if not this week, than the first week of January).

My general rule of thumb is that I have to have enough data in each task to make reasonable inference from a regression model built from just that task; if the data is noisy, the number of observations will need to be increased, of course. After that the size difference between tasks is not usually a big concern.

[asistradition]

There is now a tutorial (https://inferelator.readthedocs.io/en/latest/tutorial.html) and several example scripts for multitask as of v0.3.2 (PR #19)