R Under development (unstable) (2024-10-28 r87274 ucrt) -- "Unsuffered Consequences"
Copyright (C) 2024 The R Foundation for Statistical Computing
Platform: x86_64-w64-mingw32/x64
R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.
R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.
Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.
> library(tsDyn)
>
> mod.lstar <- lstar(log10(lynx), m=2, mTh=c(0,1), control=list(maxit=3000))
Using maximum autoregressive order for low regime: mL = 2
Using maximum autoregressive order for high regime: mH = 2
Performing grid search for starting values...
Starting values fixed: gamma = 11.15385 , th = 3.337486 ; SSE = 4.337664
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 11.15383 , th = 3.339199 ; SSE = 4.337643
> mod.lstar
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
const.L phiL.1 phiL.2
0.4891014 1.2465399 -0.3664328
High regime:
const.H phiH.1 phiH.2
-1.0240758 0.4232669 -0.2546088
Smoothing parameter: gamma = 11.15
Threshold
Variable: Z(t) = + (0) X(t) + (1) X(t-1)
Value: 3.339
> deviance(mod.lstar)
[,1]
[1,] 4.337643
> c(AIC(mod.lstar),BIC(mod.lstar))
[1] -356.6509 -334.7613
>
> mod.lstar2 <- lstar(log10(lynx), m=1, control=list(maxit=3000))
Using maximum autoregressive order for low regime: mL = 1
Using maximum autoregressive order for high regime: mH = 1
Using default threshold variable: thDelay=0
Performing grid search for starting values...
Starting values fixed: gamma = 100 , th = 3.23744 ; SSE = 12.59824
Grid search selected lower/upper bound gamma (was: 1 100 ]).
Might try to widen bound with arg: 'starting.control=list(gammaInt=c(1,200))'
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 100.0031 , th = 3.235652 ; SSE = 12.59746
> mod.lstar2
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
const.L phiL.1
0.2884595 0.9280287
High regime:
const.H phiH.1
-0.6726318 0.1331190
Smoothing parameter: gamma = 100
Threshold
Variable: Z(t) = + (1) X(t)
Value: 3.236
> deviance(mod.lstar2)
[,1]
[1,] 12.59746
> c(AIC(mod.lstar2),BIC(mod.lstar2))
[1] -239.1082 -222.6910
>
> ## include: none
> mod.lstar_noConst <- lstar(log10(lynx), m=2, control=list(maxit=1000), include="none")
Using maximum autoregressive order for low regime: mL = 2
Using maximum autoregressive order for high regime: mH = 2
Using default threshold variable: thDelay=0
Performing grid search for starting values...
Starting values fixed: gamma = 3.538462 , th = 2.719919 ; SSE = 4.933078
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 2.456908 , th = 2.707843 ; SSE = 4.907455
> mod.lstar_noConst
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
phiL.1 phiL.2
1.4005126 -0.2253725
High regime:
phiH.1 phiH.2
0.7660504 -1.0552844
Smoothing parameter: gamma = 2.457
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)
Value: 2.708
> deviance(mod.lstar_noConst)
[,1]
[1,] 4.907455
> c(AIC(mod.lstar_noConst),BIC(mod.lstar_noConst))
[1] -346.5805 -330.1633
>
> ## include: trend
> mod.lstar_trend <- lstar(log10(lynx), m=2, control=list(maxit=1000), include="trend")
Using maximum autoregressive order for low regime: mL = 2
Using maximum autoregressive order for high regime: mH = 2
Using default threshold variable: thDelay=0
Performing grid search for starting values...
Starting values fixed: gamma = 3.538462 , th = 2.712901 ; SSE = 4.899802
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 2.575672 , th = 2.697974 ; SSE = 4.880102
> mod.lstar_trend
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
trend.L phiL.1 phiL.2
1.825193e-05 1.403688e+00 -2.333263e-01
High regime:
trend.H phiH.1 phiH.2
0.0006932109 0.7390089035 -1.0305204430
Smoothing parameter: gamma = 2.576
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)
Value: 2.698
> deviance(mod.lstar_trend)
[,1]
[1,] 4.880102
> c(AIC(mod.lstar_trend),BIC(mod.lstar_trend))
[1] -343.2177 -321.3281
>
> ## include: both
> mod.lstar_both <- lstar(log10(lynx), m=2, control=list(maxit=1000), include="both")
Using maximum autoregressive order for low regime: mL = 2
Using maximum autoregressive order for high regime: mH = 2
Using default threshold variable: thDelay=0
Performing grid search for starting values...
Starting values fixed: gamma = 100 , th = 2.565527 ; SSE = 4.581837
Grid search selected lower/upper bound gamma (was: 1 100 ]).
Might try to widen bound with arg: 'starting.control=list(gammaInt=c(1,200))'
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 100.0001 , th = 2.568063 ; SSE = 4.580635
> mod.lstar_both
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
const.L trend.L phiL.1 phiL.2
0.3869417359 0.0002595822 1.2409824738 -0.3277026131
High regime:
const.H trend.H phiH.1 phiH.2
0.7981219886 0.0001995693 0.3034500606 -0.6346479944
Smoothing parameter: gamma = 100
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)
Value: 2.568
> deviance(mod.lstar_both)
[,1]
[1,] 4.580635
> c(AIC(mod.lstar_both),BIC(mod.lstar_both))
[1] -346.4371 -319.0751
>
> ## grid attributes
> mod.lstar3 <- lstar(log10(lynx), m=2, control=list(maxit=3000), starting.control=list(gammaInt=c(1,1000), nTh=100))
Using maximum autoregressive order for low regime: mL = 2
Using maximum autoregressive order for high regime: mH = 2
Using default threshold variable: thDelay=0
Performing grid search for starting values...
Starting values fixed: gamma = 871.9231 , th = 2.560424 ; SSE = 4.564989
Optimization algorithm converged
Optimized values fixed for regime 2 : gamma = 871.9231 , th = 2.560332 ; SSE = 4.564987
> mod.lstar3
Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
const.L phiL.1 phiL.2
0.3992404 1.2466373 -0.3316956
High regime:
const.H phiH.1 phiH.2
0.7821303 0.3009462 -0.6246129
Smoothing parameter: gamma = 871.9
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)
Value: 2.56
> deviance(mod.lstar3)
[,1]
[1,] 4.564987
> c(AIC(mod.lstar3),BIC(mod.lstar3))
[1] -350.8272 -328.9377
>
>
> mod.lstar_ALL <- list(mod.lstar=mod.lstar, mod.lstar2=mod.lstar2,
+ mod.lstar_noConst=mod.lstar_noConst,mod.lstar_trend=mod.lstar_trend,
+ mod.lstar_both=mod.lstar_both,mod.lstar3=mod.lstar3)
>
> vec_min_size <- function(x, size = 1, enforce = TRUE) {
+ l <- length(x)
+ if(l<size) x <- c(x, rep(NA, size-l))
+ if(enforce) x <- head(x, size)
+ x
+ }
>
>
> sapply(mod.lstar_ALL, function(x) c(AIC=AIC(x), BIC=BIC(x), deviance=deviance(x)))
mod.lstar mod.lstar2 mod.lstar_noConst mod.lstar_trend
AIC -356.650870 -239.10818 -346.580507 -343.217691
BIC -334.761283 -222.69098 -330.163316 -321.328104
deviance 4.337643 12.59746 4.907455 4.880102
mod.lstar_both mod.lstar3
AIC -346.437120 -350.827247
BIC -319.075136 -328.937660
deviance 4.580635 4.564987
> sapply(mod.lstar_ALL, function(x) vec_min_size(coef(x),8))
mod.lstar mod.lstar2 mod.lstar_noConst mod.lstar_trend mod.lstar_both
const.L 0.4891014 0.2884595 1.4005126 1.825193e-05 0.3869417359
phiL.1 1.2465399 0.9280287 -0.2253725 1.403688e+00 0.0002595822
phiL.2 -0.3664328 -0.6726318 0.7660504 -2.333263e-01 1.2409824738
const.H -1.0240758 0.1331190 -1.0552844 6.932109e-04 -0.3277026131
phiH.1 0.4232669 100.0031347 2.4569077 7.390089e-01 0.7981219886
phiH.2 -0.2546088 3.2356515 2.7078431 -1.030520e+00 0.0001995693
gamma 11.1538344 NA NA 2.575672e+00 0.3034500606
th 3.3391985 NA NA 2.697974e+00 -0.6346479944
mod.lstar3
const.L 0.3992404
phiL.1 1.2466373
phiL.2 -0.3316956
const.H 0.7821303
phiH.1 0.3009462
phiH.2 -0.6246129
gamma 871.9230770
th 2.5603319
> sapply(mod.lstar_ALL, getTh)
mod.lstar.th mod.lstar2.th mod.lstar_noConst.th
3.339199 3.235652 2.707843
mod.lstar_trend.th mod.lstar_both.th mod.lstar3.th
2.697974 2.568063 2.560332
> sapply(mod.lstar_ALL, function(x) vec_min_size(coef(x,hyperCoef=FALSE),8))
mod.lstar mod.lstar2 mod.lstar_noConst mod.lstar_trend mod.lstar_both
const.L 0.4891014 0.2884595 1.4005126 1.825193e-05 0.3869417359
phiL.1 1.2465399 0.9280287 -0.2253725 1.403688e+00 0.0002595822
phiL.2 -0.3664328 -0.6726318 0.7660504 -2.333263e-01 1.2409824738
const.H -1.0240758 0.1331190 -1.0552844 6.932109e-04 -0.3277026131
phiH.1 0.4232669 NA NA 7.390089e-01 0.7981219886
phiH.2 -0.2546088 NA NA -1.030520e+00 0.0001995693
NA NA NA NA 0.3034500606
NA NA NA NA -0.6346479944
mod.lstar3
const.L 0.3992404
phiL.1 1.2466373
phiL.2 -0.3316956
const.H 0.7821303
phiH.1 0.3009462
phiH.2 -0.6246129
NA
NA
> sapply(mod.lstar_ALL, function(x) vec_min_size(coef(x,hyperCoef=FALSE, regime = "L"), 4))
mod.lstar mod.lstar2 mod.lstar_noConst mod.lstar_trend mod.lstar_both
const.L 0.4891014 0.2884595 1.4005126 1.825193e-05 0.3869417359
phiL.1 1.2465399 0.9280287 -0.2253725 1.403688e+00 0.0002595822
phiL.2 -0.3664328 NA NA -2.333263e-01 1.2409824738
NA NA NA NA -0.3277026131
mod.lstar3
const.L 0.3992404
phiL.1 1.2466373
phiL.2 -0.3316956
NA
> sapply(mod.lstar_ALL, function(x) vec_min_size(coef(x,hyperCoef=FALSE, regime = "H"), 4))
mod.lstar mod.lstar2 mod.lstar_noConst mod.lstar_trend mod.lstar_both
const.H -1.0240758 -0.6726318 0.7660504 0.0006932109 0.7981219886
phiH.1 0.4232669 0.1331190 -1.0552844 0.7390089035 0.0001995693
phiH.2 -0.2546088 NA NA -1.0305204430 0.3034500606
NA NA NA NA -0.6346479944
mod.lstar3
const.H 0.7821303
phiH.1 0.3009462
phiH.2 -0.6246129
NA
> sapply(mod.lstar_ALL, function(x) head(x$model,2))
$mod.lstar
yy const.L phiL.1 phiL.2 const.H phiH.1 phiH.2
1 2.767156 1 2.506505 2.429752 3.931706e-05 0.0000985484 9.553071e-05
2 2.940018 1 2.767156 2.506505 9.254479e-05 0.0002560859 2.319640e-04
$mod.lstar2
yy const.L phiL.1 const.H phiH.1
1 2.506505 1 2.429752 9.980313e-36 2.424969e-35
2 2.767156 1 2.506505 2.150693e-32 5.390722e-32
$mod.lstar_noConst
yy phiL.1 phiL.2 phiH.1 phiH.2
1 2.767156 2.506505 2.429752 0.9494498 0.9203763
2 2.940018 2.767156 2.506505 1.4842116 1.3444070
$mod.lstar_trend
yy trend.L phiL.1 phiL.2 trend.H phiH.1 phiH.2
1 2.767156 1 2.506505 2.429752 0.3791494 0.9503398 0.921239
2 2.940018 2 2.767156 2.506505 1.0888603 1.5065231 1.364617
$mod.lstar_both
yy const.L trend.L phiL.1 phiL.2 const.H trend.H phiH.1
1 2.767156 1 1 2.506505 2.429752 0.002116716 0.002116716 0.00530556
2 2.940018 1 2 2.767156 2.506505 0.999999998 1.999999995 2.76715586
phiH.2
1 0.005143096
2 2.506505027
$mod.lstar3
yy const.L phiL.1 phiL.2 const.H phiH.1 phiH.2
1 2.767156 1 2.506505 2.429752 4.142784e-21 1.038391e-20 1.006594e-20
2 2.940018 1 2.767156 2.506505 1.000000e+00 2.767156e+00 2.506505e+00
>
> proc.time()
user system elapsed
6.89 1.23 8.17