Das GEO-Forum - The theory and practice in JAG3D

The theory and practice in JAG3D (Antwort)

MichaeL — Fri, 26 Jun 2015 20:20:21 +0000

Hi Alexander,

was the example of a network simulation traceable?

I mean, am (m> 1) for the local t-test. How is it calculated?

You have to use the non-central F-distribution. An algorithm is given by Hahn, Heck, Jaeger, Scheuring (1989/99), Ein Verfahren zur Abstimmung der Signifikanzniveaus für allgemeine Fm,n-verteilte Teststatistiken−Teil I/II, ZfV. As far as I know, you can get the source code (in Fortran), too.

in the latest versions of your program in the protocol computations were logarithms.

It is the so-called p-value (to avoid zero numbers, it is given in logarithmic representation).

have a nice weekend
Micha

The theory and practice in JAG3D (Antwort)

Alexander — Fri, 26 Jun 2015 18:22:01 +0000

Sorry, no time to rest, we have to work.
But Aydin aT is calculated for a global test and m> 1, as you have in the online calculator on the site. For her, I did the procedure in Matlab. it gives the same results as you. I mean, am (m> 1) for the local t-test. How is it calculated?
And more. in the latest versions of your program in the protocol computations were logarithms. What is it and why. never before seen.
Thank you and good luck.

The theory and practice in JAG3D (Antwort)

MichaeL — Fri, 26 Jun 2015 17:05:57 +0000

Hi,

If it is possible from the viewpoint of Matlab.

Again, an iterative algorithm (in Matlab) is given by Aydin, Demirel (2004). Please contact the authors to get access to their m-Files. I don't have Matlab Code, that I can share.

2. Is there any example file to adjustment the planned, not large network where measure the angles and lines? The site in the examples I have not found. I would love to see.

You can download an example for some days here: example.zip

Have a nice weekend
Micha

The theory and practice in JAG3D (Antwort)

Alexander — Fri, 26 Jun 2015 13:35:39 +0000

Hi Michael
Sorry to trouble you, but again there were questions:
1. I can not calculate the alpha(m) for m > 1. I know that for alpha = 0.05, beta = 0.2 and m = 2, alpha (m = 2) = 0.089, and the corresponding quantile k = 2.44. How were they calculated? If it is possible from the viewpoint of Matlab.
2. Is there any example file to adjustment the planned, not large network where measure the angles and lines? The site in the examples I have not found. I would love to see.
3. Thank you for your past support. Written in Matlab itself is not a big procedure for calculating the aG, works well.
Thank you for any help.
Sincerely, Alexander

The theory and practice in JAG3D (Antwort)

MichaeL — Wed, 20 May 2015 14:04:56 +0000

Hi,

as already mentioned, I use the idea proposed by Baarda. An iterative algorithm (in Matlab) is given by Aydin, Demirel (2004), Computation of Baarda’s lower bound of the non-centrality parameter.

regards
Micha