Model average using 2-class models

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

Model average using 2-class models

Postby anzuck » Thu Jan 23, 2020 7:10 am

Hello everyone,

I would like to experiment with my data a model averaging approach to see the differences between using a half-normal model with only one class or with two classes or a average of both. Does anyone have an advice for me on how to do that when averaging over a model with 1 class and another with 2 classes? As the 2-class model comprises 2 sessions, respectively 2 parameter estimates, the model.averaging gives me 2 sigma estimates that are quite different. How is the model average calculated in this case?

Thanks a lot for any help on this topic!
Chiara

Using the model.average on 1-class and 2-class model:

> model.average(m3,m9)
, , D

estimate SE.estimate lcl ucl
session=1,h2=1 0.0003192451 8.793928e-05 0.0001878997 0.0005424034
session=1,h2=2 0.0003192451 8.793928e-05 0.0001878997 0.0005424034

, , g0

estimate SE.estimate lcl ucl
session=1,h2=1 0.05542333 0.0104384 0.03817932 0.0798095
session=1,h2=2 0.05542333 0.0104384 0.03817932 0.0798095

, , sigma

estimate SE.estimate lcl ucl
session=1,h2=1 11480.169 2926.565 7020.425 18772.98
session=1,h2=2 8240.307 2018.690 5133.919 13226.28


These are the best fitting models (m3=1-class model and m9=2-class model):

model detectfn npar logLik AIC AICc dAICc AICcwt
m3 D~1 g0~1 sigma~1 halfnormal 3 -519.9562 1045.912 1046.293 0.000 0.2481
m9 D~1 g0~1 sigma~h2 pmix~h2 halfnormal 5 -517.8296 1045.659 1046.643 0.350 0.2083
m10 D~1 g0~1 sigma~h2 z~1 pmix~h2 hazard rate 6 -516.8758 1045.752 1047.152 0.859 0.1615
m12 D~1 g0~h2 sigma~h2 pmix~h2 halfnormal 6 -517.1730 1046.346 1047.746 1.453 0.1200
m13 D~1 g0~h2 sigma~h2 z~1 pmix~h2 hazard rate 7 -516.3398 1046.680 1048.578 2.285 0.0792
m5 D~1 g0~1 sigma~1 exponential 3 -521.3158 1048.632 1049.013 2.720 0.0637
m4 D~1 g0~1 sigma~1 z~1 hazard rate 4 -520.5903 1049.181 1049.826 3.533 0.0424
m6 D~1 g0~h2 sigma~1 pmix~h2 halfnormal 5 -519.7568 1049.514 1050.497 4.204 0.0303
m11 D~1 g0~1 sigma~h2 pmix~h2 exponential 5 -520.0138 1050.028 1051.011 4.718 0.0235
m14 D~1 g0~h2 sigma~h2 pmix~h2 exponential 6 -519.6203 1051.241 1052.641 6.348 0.0104
m8 D~1 g0~h2 sigma~1 pmix~h2 exponential 5 -521.1352 1052.270 1053.254 6.961 0.0076
m7 D~1 g0~h2 sigma~1 z~1 pmix~h2 hazard rate 6 -520.3534 1052.707 1054.107 7.814 0.0050
m2 D~1 g0~1 sigma~1 exponential 3 -605.5663 1217.133 1217.514 171.221 0.0000
m0 D~1 g0~1 sigma~1 halfnormal 3 -605.6553 1217.311 1217.692 171.399 0.0000
m1 D~1 g0~1 sigma~1 z~1 hazard rate 4 -605.9889 1219.978 1220.623 174.330 0.0000
anzuck
 
Posts: 1
Joined: Mon Jan 13, 2020 6:57 am

Re: Model average using 2-class models

Postby murray.efford » Mon Jan 27, 2020 4:57 pm

see same question on secrgroup
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 15 guests