\usepackage{lscape} \newcommand{\blandscape}{\begin{landscape}} \newcommand{\elandscape}{\end{landscape}}
aucTss<-function(om,mp){
rtn=mydas:::roc(om,mp)
data.frame("tss"=data.frame(with(rtn[abs(rtn$reference-1)==min(abs(rtn$reference-1)),],TPR-FPR))[1,],
data.frame("auc"=pROC:::auc(as.character(rtn$label),rtn$reference)))}
dat=data.frame(om=rep(c(1,2),10),mp=rlnorm(20))
aucTss(dat$om>1,dat$mp)
## Setting levels: control = FALSE, case = TRUE
## Setting direction: controls < cases
## tss auc
## 1 -0.3 0.64
Figure 1 Life history parameters and the correlations between them.
Figure 2 Operating Models for base case.
Figure 3 Length data
Figure 4. Example of length based indicators and estimators for Base Case by species.
Figure 5. Example of length based indicators and estimators for Base Case by species.
Figure 6. Example of length based indicators and estimators for pollack across scenarios.
Figure 7. Example of length based indicators and estimators by scenarios contrasted with base case.
Figure 8. True Skill Score, analytical
Figure 9. True Skill Score, relative
Figure 10. True Skill Score, relative
Figure 11. ROC curves
Figure 12. Area under the ROC curves
Figure 13. Estimates of slope.
Figure 14 One and three step ahead fits.
Are system dynamics the result of deterministic or stochastic processes [@pennekamp2019intrinsic]? I.e. can we predict the results of management actions based on a production function, as used to estimate reference points [@sissenwine1987alternative], or are the consequences depend on process error? i.e. processes not included in the assessment model.
Permutation entropy is a measure of the complexity of a time series [@bandt2002permutation] that is negatively correlated with a system’s predictability [@garland2018anomaly]. Permutation entropy quantifies the combined new and lost information, and is scaled to range between a minimum of 0 and a maximum of 1.
Figure 15. Entropy for base case
Figure 16. Entropy
Figure 17. Entropy by scenarios
Figure 18. Entropy by enytropy
Figure 19. Entropy v AUC
Figure 20. Status 100
Figure 21. Status 118
Figure 22. Comparison of Operating Model run for F with HCR run with perfect CPUE and with \(L_{Mean}\)
t looks like LBSPR can get you a reasonable estimate of F:FMSY from 5 years ago, if you collect data today I think what we can for jabba-z is look at what happens if i) you can go back and get data from past years ii) if you start collecting data tofday how long before you get a good estimate i.e. go from a single F/.FMSY point to multiple years this would be good for real data poor situations i.e. where the world bankl throws some money at a country and then goes away this would allow value to be obtained from past projects, or to design new ones