[ADMB Users] Retrospective analysis

Mark Maunder mmaunder at iattc.org
Mon Nov 1 11:24:23 PDT 2010

Depending on how you code the model, you can sometimes still run the
model through to the end year (i.e. if catch is input and the only year
specific parameters are rec devs that have a penalty), but just don't
fit to the data that is greater than retrospective year. 




Mark Maunder 

Head of the Stock Assessment Program

Inter-American  Tropical Tuna Commission
8604 La Jolla Shores Drive
La Jolla, CA, 92037-1508, USA
Tel: (858) 546-7027
Fax: (858) 546-7133
mmaunder at iattc.org


Visit the AD Model Builder project at
 http://admb-project.org/ <http://admb-project.org/> 


See the following website for information on fisheries stock assessment




From: users-bounces at admb-project.org
[mailto:users-bounces at admb-project.org] On Behalf Of Jim Ianelli
Sent: Monday, November 01, 2010 9:57 AM
To: 'Luis Ridao'; users at admb-project.org
Subject: Re: [ADMB Users] Retrospective analysis


My approach has been to not change the data file but fake it out.


For example, if the full model runs from styr to endyr I read all the
data in that way then have a variable (integer) also read in that sets
the number of years to peel off (zero for full model).


After reading in the data, I use a surrogate endyr to dimension all the
other variables.  E.g., endyr_r (_r for retrospective) and ensure that
all predicted values are dimensioned accordingly.


Care must be taken however to test for cases where peeling off data
exceeds the number of observations for some data types.  E.g., if you
have a survey time series that starts in 2002 and you peel of 10 years
from a model that ends in 2010...


It takes a little extra coding but has paid off.  


Be sure to compile in safe mode!






James Ianelli

REFM Division, Alaska Fisheries Science Center

National Marine Fisheries Service, National Oceanic and Atmospheric

7600 Sand Point Way NE

Seattle WA 98115


Ph. 206 526 6510


From: users-bounces at admb-project.org
[mailto:users-bounces at admb-project.org] On Behalf Of Luis Ridao
Sent: Monday, November 01, 2010 9:46 AM
To: users at admb-project.org
Subject: [ADMB Users] Retrospective analysis



I was wondering how to implement a retrospective analysis of a given
catch-at-age model.
The straightforward but also the slowest approach might be to shorten
the data file and adjust
the parameter section accordingly but I guess there must be a way to get
around this
(maybe some shell scripting?)

Is there anyone out there who has faced a similar issue?

Thanks in advance,

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20101101/daab9fed/attachment.html>

More information about the Users mailing list