IMAREScourse

Course outline

• How the estimation of N-at-age and F-at-age works
  • The random walk
  • The statistical properties of the model
  • What is fitted and how
  • The 'mixed-model' aspects
• Introduction to the FLSAM & FLSAMs class
  • Updating an FLStock with FLSAM object
  • Parameter values, residuals and uncertainties
    • nopar, nlogl, vcov, params, residuals
• Introduction to the control object
  • Setup of control object
  • Binding of parameters
  • Elements of control object and how to set these
    • fleets
    • plus.group
    • states
    • logN.vars
    • catchabilities
    • power.law.exps
    • f.vars
    • obs.vars
    • srr
    • cor.F
    • hessian, running time and executable to use
• Introduction to the diagnostics
  • Plotting an FLSAM or FLSAMs object
  • plotting pre-defined residual diagnostics
  • Plotting other useful diagnostics
    • Observation variance
    • CI of variance estimates
    • The 'banana' / otolith plot
    • Correlation plot
  • Reporting a SAM run
• Introduction to benchmarking
  • Leave-one-in (loi)
  • Leave-one-out (loo)
  • Retrospectives
  • AIC & Likelihood ratio tests
  • Adjusting the executable
• Introduction to MSE
  • MonteCarlo simulation
  • The pin file
  • Simulating retrospective error / batch FLSAM
  • Predicting selection at age
• Do's and don'ts
  • Recruitment patterns
  • Over-parameterisation
  • Correlated random walks
  • Hitting the bounds of a parameter