[FLR-list] Reproducibility and Crossvalidation

[laurie]

http://www.sciencedirect.com/science/article/pii/S0165783616301540 shows 
the need for evaluation of prediction skill.

The hindcast paper is now published, see 
http://www.sciencedirect.com/science/article/pii/S0165783616301540
A simple but powerful method and easy to do in FLR as all you need to do 
is a retrospective then a projection e.g.

xval=mlply(year=2016:2010,function(year) {
          tmp=window(ple4,end=year)
          tmp=fit(tmp)
          fwd(tmp,catch=start(catch(ple4),start=year)})

Laurie