6 Output and Plots

6.1 Summarizing output

Output is summarized using SSMSE_summary_all():

summary_list <- SSMSE_summary_all(dir = "path/to/scenario/dir",
                  scenarios = c("sample_low", "sample_high"),
                  run_parallel = TRUE)

Relying on ss3sim::get_results_all(), this function creates:

For each scenario, 3 scenario level .csv files
For all scenarios, 2 cross-scenario .csv files named by default to SSMSE_tsand SSMSE_scalar.
For all scenarios, the function returns a list object containing data frames of timeseries (ts), scalar, and derived quantities (dq) summaries.

Note that run_parallel = TRUE is only faster than run_parallel FALSE when there is more than once scenario and none of the scenario-level .csv files have been created yet.

By default, if a user doesn’t specify scenarios, all scenarios in dir will be summarized.

6.2 Checking estimation model convergence

One of the first checks to do if using an estimation model after running an MSE analysis is to check that the estimation model has converged. A number of checks could be done, but a basic one is checking the gradients for the estimation model, which are added to the SSMSE_scalar summary sheet.

6.3 Calculating performance metrics

Typically, a suite of performance metrics are used in MSE. Punt et al. (2016) recommends at least using metrics related to:

average catch
variation in catch over time
population size