You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The problem: often, when we calculate confidence intervals on tract profiles we separately calculate the 95% CI resampling values at each node. I believe that inflates the size of the CIs, because nodes are not independent. Instead, we should sample entire tract profiles with replacement and come up with a way to calculate the percentile score of entire curves, finding the extreme 5% in that multi-variate space. A simple initial way to do this would be to calculate the mean value for each curve in bootstrapped sampling and assigning percentiles based on that. Another approach would be to calculate z scores in a multivariate Gaussian (but then we'd lose the non-parametric charm of the bootstrap!).
The text was updated successfully, but these errors were encountered:
The problem: often, when we calculate confidence intervals on tract profiles we separately calculate the 95% CI resampling values at each node. I believe that inflates the size of the CIs, because nodes are not independent. Instead, we should sample entire tract profiles with replacement and come up with a way to calculate the percentile score of entire curves, finding the extreme 5% in that multi-variate space. A simple initial way to do this would be to calculate the mean value for each curve in bootstrapped sampling and assigning percentiles based on that. Another approach would be to calculate z scores in a multivariate Gaussian (but then we'd lose the non-parametric charm of the bootstrap!).
The text was updated successfully, but these errors were encountered: