starting value issues in secr 4.2.3

questions concerning anlysis/theory using program DENSITY and R package secr. Focus on spatially-explicit analysis.

starting value issues in secr 4.2.3

Postby KerryK » Tue Jun 23, 2020 4:55 am

Hello again

I've run into an issue with running some of the secr models. Some of my data is a bit sparse so secr was struggling to run the models;

Error in secr.fit(HYBRIDCH, model = list(D ~ 1, g0 ~ 1), mask = HYBRIDhabitatmask, :
too few values session 1 to determine start;
set manually or select different session with details autoini

I tried the autoini:

autoini(HYBRIDCH, HYBRIDhabitatmask)
Error in autoini(HYBRIDCH, HYBRIDhabitatmask) :
too few values for autoini

So then I tried setting the start values e.g.:

HYBRIDsecrBASE<- secr.fit(HYBRIDCH, model=list(D=0.01,g0=0.01, sigma=1000), mask=HYBRIDhabitatmask, method = "Nelder-Mead",CL=TRUE, detectfn = 'HN',trace = FALSE)

This worked fine, I ran the other models, they seemed to run fine as well, but when I looked at the AIC output all the models came out exactly the same:

AIC(fits)
model detectfn npar logLik AIC AICc dAICc AICcwt
BASE g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
TRP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
TRN g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
SRSP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
STRP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
STTR g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667

So I then re-ran the models using a previous model that had worked e.g:

HYBRIDsecrTRP <- secr.fit(HYBRIDCH, start=ALLsecrBASE, mask=HYBRIDhabitatmask, method =
+ "Nelder-Mead",CL=TRUE, detectfn = 'HN', trace=FALSE)

AIC(fits)
model detectfn npar logLik AIC AICc dAICc AICcwt
BASE g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
TRP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
TRN g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
SRSP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
STRP g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667
STTR g0~1 sigma~1 halfnormal 2 -190.1319 384.264 396.264 0 0.1667

Which gave the same problem....I'm obviously doing something wrong but I can't see what.

Any help would be much appreciated!

Thanks

Kerry
KerryK
 
Posts: 17
Joined: Sun Jul 20, 2014 10:56 am

Re: starting value issues in secr 4.2.3

Postby murray.efford » Wed Jul 29, 2020 5:09 am

Hi Kerry
That does look odd, but we can't see how you generated the models in the 'fits' object. Are you sure the models were different and fitted successfully? I presume you meant
Code: Select all
HYBRIDsecrBASE<- secr.fit(HYBRIDCH, start=list(D=0.01,g0=0.01, sigma=1000), mask=HYBRIDhabitatmask, method = "Nelder-Mead",CL=TRUE, detectfn = 'HN',trace = FALSE)
(and if CL = TRUE then D=0.01 is not used).
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand


Return to analysis help

Who is online

Users browsing this forum: No registered users and 17 guests