No Model Averaging with Presence 7.6

announcements (new versions, changes, bugs, installation problems...) related to program PRESENCE

No Model Averaging with Presence 7.6

Postby Anamaria » Mon Dec 08, 2014 9:25 am

Hello,

I was trying to use the 'Model-averaged estimates' tool in Presence 6.2 and it did't work. Then, although in the forum I had already read that some people had done it with even Presence 3.1, I downloaded the last version (7.6) to see if it was a problem of updating. But it wasn't. The tool name is actually 'Model-averaged estimates (experimental)'. Don't know if experimental means that it doesn't work.

Everytime I try I receive the following messege : " Model averaging has been revised. Currently it only works with the first set of estimates in the output file. You can have estimates for each model printed or just the model averages for each site". Then, there is an OK option, and a click on it. Afterwards it asks for printing model estimates and averages for each site, and I say YES.

Then a completely white file called 'pres_modavg' is opened in editor. With no estimates and no averages.

Am I doing something wrong? Is the tool not working?

Thanks for the answer!

Best,

Ana.
Anamaria
 
Posts: 3
Joined: Tue Nov 18, 2014 12:44 pm

Re: No Model Averaging with Presence 7.6

Postby jhines » Mon Dec 08, 2014 9:44 am

Hi Ana,

It says 'experimental' because I've programmed it to work on the datasets that I've used, but there a lot of things that can go wrong if you've done something differently. In order to fix it for your case, would you mind sending me your most recent backup zipfile in your project folder, off list?

Thanks,

Jim (jhines@usgs.gov)
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: No Model Averaging with Presence 7.6

Postby jhines » Wed Dec 10, 2014 2:04 pm

Hi Ana,

Thanks for sending me your data. I was unable to replicate the error you're having. I suspect the error was due to a problem with the directory structure on your computer. Sometimes foreign or special characters in the PATH will cause PRESENCE to be unable to find the model averaging code.

I have updated PRESENCE such that it now computes the model-averaged estimates within the user-interface (without calling an external program to do the work). This should solve your problem. Please download the latest version of PRESENCE from the website and let me know if you have any more trouble.

Cheers,

Jim
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: No Model Averaging with Presence 7.6

Postby Anamaria » Thu Dec 11, 2014 6:16 am

Hello James,

it works!! Thank you very much!!
Best,

Ana.
Anamaria
 
Posts: 3
Joined: Tue Nov 18, 2014 12:44 pm

Re: No Model Averaging with Presence 7.6

Postby Anamaria » Fri Dec 12, 2014 8:13 am

Hello,

I have still a question related with the calculation of the confidence interval. Although in the forums this topic has been already widely discussed, I couldn't find any reference related with Model Average. I have clear that the CI for the average is calculated with the resulted estimate and SE. However, in order to compare the averaged models with the resulting average, I am not sure if calculating the CI of each model with the Beta's or with the estimates. I've read that the correct way to do it usually is with Beta's, but I am not sure if this is the same way in order to compare the CI's between models and average.


Other point is, neither the Beta's SE nor the estimated ones are calculated with overdispersion. One has to do it manually multiplying them by root(c-hat), right?

Thanks for the help in advance.

Best,

Ana
Anamaria
 
Posts: 3
Joined: Tue Nov 18, 2014 12:44 pm

Re: No Model Averaging with Presence 7.6

Postby cooch » Fri Dec 12, 2014 9:34 am

Anamaria wrote:Hello,

I have still a question related with the calculation of the confidence interval. Although in the forums this topic has been already widely discussed, I couldn't find any reference related with Model Average. I have clear that the CI for the average is calculated with the resulted estimate and SE. However, in order to compare the averaged models with the resulting average, I am not sure if calculating the CI of each model with the Beta's or with the estimates. I've read that the correct way to do it usually is with Beta's, but I am not sure if this is the same way in order to compare the CI's between models and average.


Other point is, neither the Beta's SE nor the estimated ones are calculated with overdispersion. One has to do it manually multiplying them by root(c-hat), right?

Thanks for the help in advance.

Best,

Ana


The issue of CI for model averaged estimates is discussed in the occupancy book (as I recall). But, the basic ideas for the SE and CI of model averaged estimates apply generally (not simply to 'occupancy' parameters), so you might want to have a look at section 4.5.1 in the 'MARK book':

http://www.phidot.org/software/mark/doc ... df#page=48
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: No Model Averaging with Presence 8.4

Postby sprihharsh » Mon Mar 30, 2015 3:23 am

I am facing problem while running model averaging with Presence 8.4. Whenever I click on this "Model Averaged estimates(experimental)" option, a pop-up comes saying "Model averaging has been revised. Currently it only works with the first set of estimates in the output files. Would you like to have estimates from each model printed? (Default is to just the model averages for each site.)

When I say Yes, the whole program gets shut down and if I say No, nothing happens.

Please let me know where I am going wrong.


Thanks,
Sprih
sprihharsh
 
Posts: 4
Joined: Sun Aug 24, 2014 12:43 pm


Return to software problems/news

Who is online

Users browsing this forum: No registered users and 10 guests