search

MaC-Perez / Hydra-self-testing

0 stars 0 forks source link

issues with sim data #6

Open MaC-Perez opened 11 months ago

MaC-Perez commented 11 months ago

When I plot the 100 simulations I get very similar values in the 100 sims, maybe I am using the worng values here....

for example to simulate catch data https://github.com/MaC-Perez/Hydra-self-testing/blob/344e2ae7f74439c2349910728e0696b72494981b/hydra_sim.R#L46

and survey abundance https://github.com/MaC-Perez/Hydra-self-testing/blob/344e2ae7f74439c2349910728e0696b72494981b/hydra_sim.R#L53

I'm pretty sure there's everything you need in this repo to run the code

MaC-Perez commented 11 months ago

I forgot to tag you @gavinfay

gavinfay commented 11 months ago

Same problem with both so here for biomass: You are assuming normal random variables with mean indexfits[[1]]$pred_bio which appears to be in real space (has range in the 100s to hundreds of trillions, and standard deviation = indexfits[[2]]$cv, which is 0.1. So the simulated values will be practically identical. The likelihood component (and hence your simulated observations) for these data is lognormal not normal.

(Definitely check your units for your biomass and catch, these numbers seem implausibly huge - these seem really big even if you have not got the weight conversion right - check this by looking how big your individuals are))