Missing parameters - 'Reduced Set of Parameters'

questions concerning analysis/theory using programs M-SURGE, E-SURGE and U-CARE

Missing parameters - 'Reduced Set of Parameters'

Postby jaimie_cleeland » Wed Jul 06, 2016 1:18 am

Hi Remi and forum participants,

I am currently running a multi-event model of survival, return, breeding probability and breeding success with unobservable states and uncertainty applied to breeding outcome for 3 populations of albatrosses (20-year dataset).

I have incoporated individual covariate data in the Gemaco to test the effects of different environmental covariates on these vital rates. For several of the covariates tested the model seems to be estimating well. Some time steps the parameters appear to be non-estimatable and showing confidence intervals from 0 to 1.

The problem I am having is that with two of eight covariates tested on breeding success of one species I am not seeing any estimates of breeding success on several time steps. For example, where I expect to see estimates from time step 1 to 20 in the "Reduced Set of Parameters" output, I am only seeing estimates for time steps 2, 4, 5, 6, 8, 10, 11, 12, 13, 15, 17, 20. If these were not estimable than wouldn't I see an estimate with broad confidence intervals? I suspected an error in the covariate file and have subsequently checked this and it appears to be the same as previous working covariates. Happy to provide Gepat matrices and Gemaco sentences if you think it is required.

I am a novice with E-Surge, therefore apologise in advance for lack of detail. Any help is greatly apprecitated.

Thank-you.

Jaimie Cleeland

Quantitative Antarctic Science PhD Program
University of Tasmania/Australian Antarctic Division
Private Bag 129, Hobart 7001
jaimie.cleeland@utas.edu.au
https://www.researchgate.net/profile/Jaimie_Cleeland
jaimie_cleeland
 
Posts: 3
Joined: Tue Jul 05, 2016 10:23 pm

Re: Missing parameters - 'Reduced Set of Parameters'

Postby Guillaume Souchay » Thu Jul 07, 2016 5:51 am

Dear Jaimie,

Could you post the complete GEMACO sentence you are using?
this sentence is a bit confusing:
I have incoporated individual covariate data in the Gemaco to test the effects of different environmental covariates on these vital rates.
So, are you using individual covariates varying over time (e.g. weight of the bird known for each capture occasion) or environmental covariates?

What is the structure of your model? Is there an age-effect incorporated in your model?

best wishes,
Guillaume
Guillaume Souchay
 
Posts: 63
Joined: Wed Feb 27, 2008 5:20 am
Location: Nantes, FRANCE

Re: Missing parameters - 'Reduced Set of Parameters'

Postby jaimie_cleeland » Fri Jul 08, 2016 12:55 am

Dear Guillaume,

Thank you for taking the time to have a look at this.
Apologies,for the lack of clarity - I am incorporating environmental covariates that change over time. There is no age effect in the model.

I have seven states (three observable, four unobservable):
- successful breeder (SB), failed breeder (FB), non-breeder (NB), post successful breeder (PSB), post failed breeder (PFB), post non-breeder (PNB) and dead (+).
And five events in my encounter histories:
- SB (coded 1), FB (coded 2), NB (coded 3), breeder with unknown breeding outcome (UB, coded 4) and not seen (coded 0).

Gemaco sentences:

Initial state - i

Transitions
Survival probability (s) - t(1:42)+t(43_63)
Return probability (r) - t(1:42)+t(43_63)
Breeding probability (b) - t(1:42)+t(43_63)
Breeding success (k) - t(1:42)+[i+t(43_63)*x(1)]

Event
Resighting probability including unobservable states (c) - firste+nexte.[t(43_63)+f(1 2,3)] + others
Breeding outcome uncertainty (a) - i

I have included a link with some images of my problem with missing parameters for both the time-dependent and covariate model.
https://www.dropbox.com/sh/si0z9r4tgqsa ... vbRqa?dl=0

Kind regards,

Jaimie
jaimie_cleeland
 
Posts: 3
Joined: Tue Jul 05, 2016 10:23 pm

Re: Missing parameters - 'Reduced Set of Parameters'

Postby jaimie_cleeland » Sun Jul 17, 2016 9:31 pm

Thank you Guilluame for your help.

For others that may have the same problem (missing parameters), Guillaume provided me with this insight:

E-SURGE backtransformed the estimated value for each year and put it in both “Parameter” and the “Reduced set of parameters” sheets. I think this is because there are similar values of breeding success and once the value has been reported, it is not shown again. Remember that you are estimating the relationship between rainfall and breeding success and that the “reduced set of parameters” sheet provides the value of unique estimated rates. Thus, when estimating breeding success for each occasion, you may estimate similar values, so the repeated occasion are discarded from the “reduced set of parameters”.

Let’s illustrated this with your case: occasions 43, 49, 51 and 58 have the same estimated breeding success, and only the value for occasion 43 is shown. The same for occasion 44 and 45, only values for 44 is shown, for occasions 47, 54 and 60, only values for 47 is shown and finally 48 and 56 and only 48 is shown. Thus, 49,51,58,45,54,60 and 56 are not shown in the reduced set of parameters because they are already given, but you can find the value for these occasions in the “parameters” sheet.


Thanks again,

Jaimie
jaimie_cleeland
 
Posts: 3
Joined: Tue Jul 05, 2016 10:23 pm


Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest