[ncl-talk] A question on bootstrap functions

[Dennis Shea]

The "method" used is brute force. Nothing fancy! See:

*https://www.ncl.ucar.edu/Applications/bootstrap.shtml*
<https://www.ncl.ucar.edu/Applications/bootstrap.shtml>

See the text section beginning with:

*A simple example of using sampling with replacement* t*o derive a
bootstrapped mean and 95% confidence limits. Here x(N):*

This shows that the statistic is calculated. The N statistics from
resampled data are sorted into ascending order.
The lower (say) 2.5% [*toint*(0.025*N) ]and upper [*toint*(0.975*N)] index
[subscripts] are used to extract the appropriate values.




On Tue, Apr 30, 2019 at 8:23 PM 宣守丽 <15295538792 at 163.com> wrote:

> Dear NCL experts,
>
>
>
> There are several bootstrap approaches to significance testing:
>
> 1.Permutation tests (without replacement)
>
> 2.Bootstrap confidence intervals (with replacement, including bootstrap t
> confidence interval, bootstrap percentile confidence interval, bootstrap
> bias-corrected accelerated (BCa) interval and bootstrap tilting interval
> etc).
>
> According to the description of the bootstrap function “Bootstrapping
> <https://en.wikipedia.org/wiki/Bootstrapping_%28statistics%29> is a
> statistical method that uses data resampling with replacement”, it seems
> that NCL chooses the bootstrap confidence intervals method. Could anyone
> tell me which one of the several bootstrap confidence intervals is used to
> get the confidence intervals, BCa or tilting, or other methods?
>
>
>
> Sincerely,
> Shouli Xuan
> Email:15295538792 at 163.com
>
>
>
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20190430/e01b907b/attachment.html>