Re: significance test between two correlation dataset

From: EUNSIL JUNG <eunsil.jung_at_nyahnyahspammersnyahnyah>
Date: Sat Jul 19 2014 - 06:18:20 MDT

Thank you so much, and am sorry that I was not clear for the question.

I'm looking for something like Z-test.

By using, "escort", I got correlation between two variables (time, lat, =
lon) for one particular period
(Say this correlation dataset is A(lat,lon))

Then, I got second correlation data by using the same two variables but =
for different time period
(Say this correlation dataset is B(lat,lon))

I'd like to know whether the correlation differences between two time =
periods (A and B) are significant or not.

Thanks in advance.
Eunsil


On Jul 18, 2014, at 5:40 PM, Dennis Shea <shea@ucar.edu> wrote:

> The question is not clear.
>
> Did you calculate the pattern correlation between A and B?
>
> =
https://www.ncl.ucar.edu/Document/Functions/Contributed/pattern_cor.shtml
>
> If so, there is only one correlation. If you make an astute estimate =
of the number of independent samples, it is likely that NCL's 'rtest' =
can be used.
>
> Please 'google' topics like 'pattern correlation testing, or =
'bivariate correlation testing' and examine some of those sites.
>
> Also, look at 'R' documentation for function testing.
>
>
>
>
>
>
>
> On Fri, Jul 18, 2014 at 7:38 AM, Eunsil Jung <eunsil.jung@gmail.com> =
wrote:
> Hello NCL users
>
> Are there any ways to calculate the significance of “correlation
> difference” in NCL?
> I'd like to test whether correlation A (lat,lon) is significantly
> different from correlation B(lat,lon).
>
> Thanks in advance.
>
> Eunsil
> =
--------------------------------------------------------------------------=
----------------------
>
>
> RSMAS/MPO
> 4600 Rickenbacker cswy
> Miami FL 33149
>
>
> Eunsil Jung
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>



_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk


Received on Sat Jul 19 12:18:24 2014

This archive was generated by hypermail 2.1.8 : Fri Aug 01 2014 - 15:10:55 MDT