Tim Cline's original code to Fit a DFA model with TMB.

Source: R/dfaTMB.R

This can be called to fit a DFA with his syntax, but generally no work should be done here.

dfaTMB(
  y,
  model = list(m = 1, R = "diagonal and equal"),
  inits = list(R = 0.05),
  EstCovar = FALSE,
  Covars = NULL,
  indivCovar = FALSE,
  Dmat = NULL,
  Dfac = NULL,
  EstSE = FALSE,
  silent = TRUE,
  fun.opt = c("nlminb", "optim", "nlminb+optim"),
  method = c("Nelder-Mead", "BFGS", "CG", "L-BFGS-B", "SANN", "Brent"),
  control = NULL,
  form = c("dfa", "marxss")
)

Arguments

y

Vector of observations n x T.

model

list with

  • R "diagonal and equal", "unconstrained", "diagonal and unequal"

  • m number of states (x)

inits

list of initial conditions

EstCovar

TRUE/FALSE

Covars

An optional matrix, dimensioned nD x T, where nD is the number of covariates

indivCovar

flag for structure of covariates

Dmat

D initial matrix

Dfac

What elements of D are fixed

EstSE

TRUE / FALSE, whether to return the Hessian from the TMB sdreport

silent

Show TMB output when fitting, defaults to TRUE

fun.opt

function to use for optimization: stats::nlminb() or stats::optim()

method

to pass to optim call; ignored for fun="nlminb"

control

a list with the control settings for the optimatization function. See details for the defaults.

form

The equation form used in the marssTMB() call. The default is "dfa".

Value

A list with Optimization, Estimates, Fits, and AIC

Details

The control defaults for stats::nlminb() are iter.max = 2000 and eval.max = 2000. For stats::optim(), the defaults are reltol = 1e-12 and maxit = 2000.

Author

Tim Cline wrote most of this while a graduate student in the Fish 507 Time Series Analysis course. Eli Holmes later modified it to replicate the MARSS(x, form="dfa") model.

Examples

library(MARSS)
data(lakeWAplankton, package = "MARSS")
phytoplankton <- c("Cryptomonas", "Diatoms", "Greens", "Unicells", "Other.algae")
dat <- as.data.frame(lakeWAplanktonTrans) |>
  subset(Year >= 1980 & Year <= 1989) |>
  subset(select=phytoplankton) |>
  t() |>
  MARSS::zscore()

#fit with MARSS
m1.em <- MARSS(dat, model=list(R='unconstrained', m=1, tinitx=1), form='dfa', z.score=FALSE)
#> Success! abstol and log-log tests passed at 449 iterations.
#> Alert: conv.test.slope.tol is 0.5.
#> Test with smaller values (<0.1) to ensure convergence.
#> 
#> MARSS fit is
#> Estimation method: kem 
#> Convergence test: conv.test.slope.tol = 0.5, abstol = 0.001
#> Estimation converged in 449 iterations. 
#> Log-likelihood: -772.4017 
#> AIC: 1584.803   AICc: 1586.264   
#>  
#>         Estimate
#> Z.11      0.1158
#> Z.21      0.1239
#> Z.31      0.0898
#> Z.41      0.3919
#> Z.51     -0.1652
#> R.(1,1)   0.9364
#> R.(2,1)   0.3157
#> R.(3,1)   0.1475
#> R.(4,1)   0.2863
#> R.(5,1)   0.2006
#> R.(2,2)   0.9284
#> R.(3,2)  -0.0404
#> R.(4,2)   0.0636
#> R.(5,2)  -0.0233
#> R.(3,3)   0.9641
#> R.(4,3)   0.0314
#> R.(5,3)   0.2919
#> R.(4,4)   0.3587
#> R.(5,4)   0.2605
#> R.(5,5)   0.8792
#> Initial states (x0) defined at t=1
#> 
#> Standard errors have not been calculated. 
#> Use MARSSparamCIs to compute CIs and bias estimates.
#> 
m1.tmb <- dfaTMB(dat, model=list(R='unconstrained', m=1))
c(m1.em$logLik, m1.tmb$logLik)
#> [1] -772.4017 -772.4011