i have a vector of time series data (1020x30; an index of neural activation measured at 1020 timepoints in 30 participants). I also have assessed 5 covariates of interest at a group level (eg age). I want to estimate how much those factors contribute to neural activation across time. I could estimate an ols at each time point, of course. However, what would be the best way to correct for multiple comparisons? Make a null distribution at each time point by randomly shuffling matrices? Every help would be very much appreciated!


1 Answer 1


I read a paper a while back that used an L2-norm to do sort of a "temporal smoothing" process across the predictions, which I thought was an interesting way to deal with time-varying components. I unfortunately can't find it but here's a similar paper I found

Learning Time-Varying Graphs using Temporally Smoothed L1-Regularized Logistic Regression http://citeseerx.ist.psu.edu/viewdoc/download?doi=

I don't think this answers the question directly, but perhaps it can give you ideas of how to play around with penalty functions across the time points, to get a "stable" estimate of each factor's contribution to the neural activation?

On the other hand, going back to your proposed idea of shuffling and permuting the values/matrices to determine each factor's estimate, then here's a page that talks about how this same procedure is done in random forests.

Hope this helps!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.