Re: Statistical Significance Test in NCL

From: Soumik Basu <sbasu_at_nyahnyahspammersnyahnyah>
Date: Mon Mar 24 2014 - 15:08:13 MDT

Thanks Dennis for the reply. I wanted to determine if two distributions
are same but as you said the function for it isn't released yet. So I am
now using ttest but I am getting an error :
warning:ttest: encountered 1 cases where var1 and/or var2 were less than
or equal to 0. Output set to missing values in these cases

I am not sure why I am getting this since I have printed out the
variance and there is no zero or negative number.

Variable: var_p
Type: float
Total Size: 20 bytes
             5 values
Number of Dimensions: 1
Dimensions and sizes: [5]
Coordinates:
Number Of Attributes: 2
   _FillValue : 9.96921e+36
   variance_op_ncl : dim_variance_n function was applied
(0) 7.990942
(1) 9.834744
(2) 3.267889
(3) 5.992806
(4) 5.159097

Variable: var_n
Type: float
Total Size: 20 bytes
             5 values
Number of Dimensions: 1
Dimensions and sizes: [5]
Coordinates:
Number Of Attributes: 2
   _FillValue : 9.96921e+36
   variance_op_ncl : dim_variance_n function was applied
(0) 4.094823
(1) 8.155126
(2) 8.030828
(3) 2.706451
(4) 8.858122

Part of the code doing ttest is:

;************************************************************
; Get average and variance and do ttest
;***********************************************************
   av_p = dim_avg_n_Wrap(cai_p(0,:,:),1)
   var_p = dim_variance_n_Wrap(cai_p(0,:,:),1)
   av_n = dim_avg_n_Wrap(cai_n(0,:,:),1)
   var_n = dim_variance_n_Wrap(cai_n(0,:,:),1)

   print(av_p)
   print(var_p)
   print(av_n)
   print(var_n)

    s1 = (/5,5,5,5,5/)
; s1 = dimsizes(cai_p(0,:,:))
    s2 = s1
   s2 = s1
   printVarSummary(s1)
   siglvl = 0.05

   iflag = False

   sig = ttest(av_p,av_n,var_p,var_n,s1,s2,iflag,True)

   print(sig)

Thanks,
Soumik

-- 
"Numbing the pain for a while will make it worse when you finally feel 
it.
  ~J.K. Rowling"
*****************************************
Soumik Basu
Graduate Student, Research Assistant
International Arctic Research Center
University of Alaska Fairbanks
Fairbanks,Alaska,USA
*****************************************
On 2014-03-24 11:16, Dennis Shea wrote:
> On 3/24/14, 12:22 PM, Soumik Basu wrote:
>> Hello,
>> 
>> Is there any function in NCL that I can use to perform a statistical
>> significance test of PDF distribution?
> 
> What are you testing? What is your null hypothesis?
> 
> If you want to determine if two distributions are the same,
> The in the soon-to-be-released 6.2.0 there is a
> Kolmogorov-Smirnov two sample test. See "kolsm2_n"
> 
>    http://www.ncl.ucar.edu/future_release.shtml
> 
>> Also, is there any example showing how to mark the confidence 
>> intervals
>> on a PDF plot?
> 
> No.
> 
>> 
>> Thanks,
>> Soumik
>> 
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Mon Mar 24 15:08:24 2014

This archive was generated by hypermail 2.1.8 : Mon Mar 31 2014 - 11:47:09 MDT