[ncl-talk] generate indices for resampling

[Dennis Shea]

Not sure I understand

http://www.ncl.ucar.edu/Applications/bootstrap.shtml

http://www.ncl.ucar.edu/Document/Functions/Contributed/generate_sample_indices.shtml

====
ntim = dimsizes(time)
iwo = generate_sample_indices(ntim,0)   ; *entire* period

ntim80 = toint(0.80*ntim)
iwo80  = iwo(0:ntim80-1)  ; extract the 1st 80% of the indices

====




On Fri, Jun 1, 2018 at 10:06 AM, Andrew Kren - NOAA Affiliate <
andrew.kren at noaa.gov> wrote:

> Dear ncl-talk,
>
> This is more of a science question than a coding issue. I noticed ncl has
> the function generate_sample_indices to resample with and without
> replacement. My data is temperature and moisture retrievals as function of
> (time,lev). The issue I am tackling is that my data has a high temporal
> frequency. I want to thin the data prior to assimilating it in the HWRF.
> The data is reported at approximately every 2-3 seconds for a full 24
> hours. So I have a lot of measurements and want to thin it.
>
> I did a test where I retain 80% of the data randomly:
>
> ; generate indices for resampling without replacement to sub-sample the
> HAMSR data
>
> n_retain = toint(ceil(0.8*dimsizes(time)))
>
> print(dimsizes(time))
>
> iwo = generate_sample_indices(n_retain,0)
>
> print(dimsizes(iwo))
>
>
> 14398 - dimsizes(time)
>
> 11519 - dimsizes(iwo)
>
>
> However, I noticed that maybe this may not be correct since it will only
> select n_retain values, and thus may not randomly get indices from the full
> dataset to subset. Is there a way to do this with the function without
> having to do replacement?
>
>
> Thanks,
>
>
> --
> Andrew Kren
> Assistant Scientist
> CIMAS & NOAA/AOML
> 325 Broadway, Boulder, CO 80305
> (303) 497-5418
>
> _______________________________________________
> 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/20180601/efbc6521/attachment.html>