Class for stock-recruitment models.
FLSR(model, ...)
# S4 method for ANY
FLSR(model, ...)
# S4 method for missing
FLSR(model, ...)Details
A series of commonly-used stock-recruitment models are already available,
including the corresponding likelihood functions and calculation of initial
values. See SRModels for more details and the exact
formulation implemented for each of them.
Slots
- name
Name of the object (
character).- desc
Description of the object (
character).- range
Range (
numeric).- rec
Recruitment series (
FLQuant).- ssb
Index of reproductive potential, e.g. SSB or egg oor egg production (
FLQuant).- fitted
Estimated values for rec (
FLQuant).- residuals
Residuals obtained from the model fit (
FLArray).- covar
Covariates for SR model (
FLQuants).- model
Model formula (
formula).- gr
Function returning the gradient of the likelihood (
function).- logl
Log-likelihood function (
function).- initial
Function returning initial parameter values for the optimizer (
function).- params
Estimated parameter values (
FLPar).- logLik
Value of the log-likelihood (
logLik).- vcov
Variance-covariance matrix (
array).- details
Extra information on the model fit procedure (
list).- logerror
Is the error on a log scale (
logical).- distribution
(
factor).- hessian
Resulting Hessian matrix from the fit (
array).
Examples
# Create an empty FLSR object.
sr1 <- FLSR()
# Create an FLSR object using the existing SR models.
sr2 <- FLSR(model = 'ricker')
sr2@model
#> rec ~ a * ssb * exp(-b * ssb)
#> <environment: 0x557227629738>
sr2@initial
#> function(rec, ssb) {
#> # The function to provide initial values
#> res <-coefficients(lm(log(c(rec)/c(ssb))~c(ssb)))
#> return(FLPar(a=max(exp(res[1])), b=-max(res[2])))}
#> <bytecode: 0x557227621ee0>
#> <environment: 0x557227629738>
#> attr(,"lower")
#> [1] -Inf -Inf
#> attr(,"upper")
#> [1] Inf Inf
sr2@logl
#> function(a, b, rec, ssb)
#> loglAR1(log(rec), log(a*ssb*exp(-b*ssb)))
#> <bytecode: 0x557227620370>
#> <environment: 0x557227629738>
sr3 <- FLSR(model = 'bevholt')
sr3@model
#> rec ~ a * ssb/(b + ssb)
#> <environment: 0x557227acabf0>
sr3@initial
#> function(rec, ssb) {
#> a <- max(quantile(c(rec), 0.75, na.rm = TRUE))
#> b <- max(quantile(c(rec)/c(ssb), 0.9, na.rm = TRUE))
#> return(FLPar(a = a, b = a/b))}
#> <bytecode: 0x557227ac3240>
#> <environment: 0x557227acabf0>
#> attr(,"lower")
#> [1] -Inf -Inf
#> attr(,"upper")
#> [1] Inf Inf
sr3@logl
#> function(a, b, rec, ssb)
#> loglAR1(log(rec), log(a*ssb/(b+ssb)))
#> <bytecode: 0x557227abd730>
#> <environment: 0x557227acabf0>
# Create an FLSR using a function.
mysr1 <- function(){
model <- rec ~ a*ssb^b
return(list(model = model))}
sr4 <- FLSR(model = mysr1)
# Create an FLSR using a function and check that it works.
mysr2 <- function(){
formula <- rec ~ a+ssb*b
logl <- function(a, b, sigma, rec, ssb) sum(dnorm(rec,
a + ssb*b, sqrt(sigma), TRUE))
initial <- structure(function(rec, ssb) {
a <- mean(rec)
b <- 1
sigma <- sqrt(var(rec))
return(list(a=a, b=b, sigma=sigma))},
lower = c(0, 1e-04, 1e-04), upper = rep(Inf, 3))
return(list(model = formula, initial = initial, logl = logl))
}
ssb <- FLQuant(runif(10, 10000, 100000))
rec <- 10000 + 2*ssb + rnorm(10,0,1)
sr5 <- FLSR(model = mysr2, ssb = ssb, rec = rec)
sr5.mle <- fmle(sr5)
#> Nelder-Mead direct search function minimizer
#> function value for initial parameters = 300009.608149
#> Scaled convergence tolerance is 0.00447049
#> Stepsize computed as 10894.415305
#> BUILD 4 35604233172799.054688 246594.512118
#> LO-REDUCTION 6 8027994845809.416992 246594.512118
#> HI-REDUCTION 8 2039725245416.724609 246594.512118
#> HI-REDUCTION 10 513931463016.055481 246594.512118
#> HI-REDUCTION 12 128892303912.929276 246594.512118
#> HI-REDUCTION 14 32226144724.193943 246594.512118
#> HI-REDUCTION 16 8032781791.753624 246594.512118
#> HI-REDUCTION 18 1993248612.371147 246594.512118
#> HI-REDUCTION 20 490557443.511403 246594.512118
#> HI-REDUCTION 22 118829945.657251 246594.512118
#> HI-REDUCTION 24 27914415.365704 246594.512118
#> HI-REDUCTION 26 6199169.697422 246594.512118
#> HI-REDUCTION 28 1277860.003481 246594.512118
#> HI-REDUCTION 30 419003.785329 246594.512118
#> REFLECTION 32 301368.860949 169563.447589
#> HI-REDUCTION 34 300009.608149 169563.447589
#> LO-REDUCTION 36 246594.512118 154961.674216
#> HI-REDUCTION 38 176793.742088 153601.478392
#> EXTENSION 40 169563.447589 96638.628829
#> LO-REDUCTION 42 154961.674216 96638.628829
#> LO-REDUCTION 44 153601.478392 96638.628829
#> EXTENSION 46 119295.008370 66080.769087
#> EXTENSION 48 115364.613500 56420.892581
#> LO-REDUCTION 50 96638.628829 56420.892581
#> LO-REDUCTION 52 77531.719425 56420.892581
#> EXTENSION 54 66080.769087 37593.887867
#> LO-REDUCTION 56 61167.866773 37593.887867
#> EXTENSION 58 56420.892581 29758.469273
#> REFLECTION 60 38434.634697 19972.061004
#> HI-REDUCTION 62 37593.887867 19972.061004
#> EXTENSION 64 29758.469273 13292.146429
#> EXTENSION 66 27552.362509 2764.858404
#> LO-REDUCTION 68 19972.061004 2764.858404
#> LO-REDUCTION 70 13292.146429 2764.858404
#> REFLECTION 72 8828.272171 147.072099
#> LO-REDUCTION 74 3014.837648 147.072099
#> HI-REDUCTION 76 2764.858404 147.072099
#> HI-REDUCTION 78 584.469211 147.072099
#> HI-REDUCTION 80 499.581626 143.777028
#> HI-REDUCTION 82 461.176536 108.387971
#> LO-REDUCTION 84 164.272917 108.387971
#> HI-REDUCTION 86 147.072099 91.756835
#> HI-REDUCTION 88 143.777028 77.062808
#> HI-REDUCTION 90 108.387971 77.062808
#> LO-REDUCTION 92 91.756835 71.331379
#> HI-REDUCTION 94 80.893401 71.331379
#> HI-REDUCTION 96 77.062808 71.331379
#> HI-REDUCTION 98 74.333977 69.116436
#> LO-REDUCTION 100 71.388240 69.116436
#> LO-REDUCTION 102 71.331379 69.116436
#> HI-REDUCTION 104 70.398595 69.116436
#> LO-REDUCTION 106 69.486916 68.672613
#> HI-REDUCTION 108 69.468860 68.672613
#> LO-REDUCTION 110 69.116436 68.672613
#> LO-REDUCTION 112 68.787389 68.548236
#> HI-REDUCTION 114 68.732991 68.548236
#> HI-REDUCTION 116 68.672613 68.548236
#> LO-REDUCTION 118 68.568859 68.524579
#> REFLECTION 120 68.548520 68.481495
#> HI-REDUCTION 122 68.548236 68.481495
#> LO-REDUCTION 124 68.524579 68.479702
#> HI-REDUCTION 126 68.498588 68.479702
#> REFLECTION 128 68.488963 68.471117
#> LO-REDUCTION 130 68.481495 68.471117
#> EXTENSION 132 68.479702 68.466844
#> LO-REDUCTION 134 68.478520 68.466844
#> Exiting from Nelder Mead minimizer
#> 136 function evaluations used
sr5.nls <- nls(sr5)
# NS Herring stock-recruitment dataset
data(nsher)
# already fitted with a Ricker SR model
summary(nsher)
#> An object of class "FLSR"
#>
#> Name:
#> Description:
#> Quant: age
#> Dims: age year unit season area iter
#> 1 45 1 1 1 1
#>
#> Range: min minyear max maxyear
#> 0 1960 0 2004
#>
#>
#> Model: rec ~ a * ssb * exp(-b * ssb)
#> params
#> iter a b
#> 1 119 0.00945
#> Log-likelihood: 15.862(0)
#> Variance-covariance:
#> a b
#> a 255.3388181 1.808870e-02
#> b 0.0180887 1.992659e-06
plot(nsher)
# change model
model(nsher) <- bevholt()
# fit through MLE
nsher <- fmle(nsher)
#> Nelder-Mead direct search function minimizer
#> function value for initial parameters = -10.336211
#> Scaled convergence tolerance is 1.54022e-07
#> Stepsize computed as 501.110000
#> BUILD 3 44.842344 -11.603908
#> Warning: NaNs produced
#> HI-REDUCTION 5 31.685209 -11.603908
#> Warning: NaNs produced
#> HI-REDUCTION 7 17.913114 -11.603908
#> Warning: NaNs produced
#> HI-REDUCTION 9 5.415279 -11.603908
#> Warning: NaNs produced
#> HI-REDUCTION 11 -3.412974 -11.603908
#> HI-REDUCTION 13 -8.018030 -11.603908
#> LO-REDUCTION 15 -10.336211 -11.603908
#> LO-REDUCTION 17 -11.081040 -11.603908
#> EXTENSION 19 -11.295930 -12.061705
#> LO-REDUCTION 21 -11.603908 -12.061705
#> REFLECTION 23 -11.813826 -12.087620
#> REFLECTION 25 -12.061705 -12.199591
#> LO-REDUCTION 27 -12.087620 -12.199591
#> LO-REDUCTION 29 -12.158184 -12.199591
#> LO-REDUCTION 31 -12.191726 -12.199591
#> HI-REDUCTION 33 -12.192269 -12.199591
#> HI-REDUCTION 35 -12.197784 -12.199591
#> LO-REDUCTION 37 -12.198015 -12.199591
#> HI-REDUCTION 39 -12.199555 -12.199776
#> REFLECTION 41 -12.199591 -12.200058
#> HI-REDUCTION 43 -12.199776 -12.200092
#> HI-REDUCTION 45 -12.200058 -12.200142
#> HI-REDUCTION 47 -12.200092 -12.200155
#> HI-REDUCTION 49 -12.200142 -12.200160
#> HI-REDUCTION 51 -12.200155 -12.200177
#> HI-REDUCTION 53 -12.200160 -12.200177
#> LO-REDUCTION 55 -12.200171 -12.200179
#> HI-REDUCTION 57 -12.200177 -12.200179
#> HI-REDUCTION 59 -12.200178 -12.200179
#> HI-REDUCTION 61 -12.200179 -12.200179
#> HI-REDUCTION 63 -12.200179 -12.200179
#> HI-REDUCTION 65 -12.200179 -12.200179
#> Exiting from Nelder Mead minimizer
#> 67 function evaluations used
plot(nsher)
