Stanford Center for Biomedical Informatics Research Colloquia - Shared screen with speaker view
What about bagging? That requires no weights.
Post questions into the Q&A box and we'll relay to our speaker at the breaks
I retract my question. The speaker doesn’t seem to be talking about ensembles after all.
How do you begin exploration in an emergency situation [phase 3 + 1] on a global scale? At your first shot, just how many data sets are you required to acknowledge, accept and analyze? What range of numbers makes is scientifically and academicallyT good(C) to begin analysis.
how we do decide the rsquared to input in the ronde. can we assume or from previous literature
Does the functional form of predictors have any effect on sample size requirement?
Post questions to Q&A
I'm not really clear about your response to the use of domain knowledge in sample size- would the analysis still be considered data-driven if domain knowledge is applied?
What's your view about data augmentation ?
Is an analysis of purely synthetic data a publishable project- in the medical domain particularly?
Any threshold for imputation?
thank you for that perspective!