Create lag variables and evaluate models with different number of lags
Source:R/models.R
evaluateLags.RdThis function creates the desired number of lags and tests consecutive models from a model with no lags (lag 0), lag 0 + lag1, etc. and reports model performance. This helps evaluate how many lags are needed.
Arguments
- formula
A
characterstring giving thelmer()formula to use as a base. The variable to be tested with lags gets added as fixed effects only to this, currently.- lagvar
A
characterstring giving the name of the variable to test lags for.- nlags
An
integer(e.g., 0L, 3L) giving the number of lags to test. Defaults to 0L but really should be more. Must be a positive integer.- idvar
A
characterstring giving the name o the ID variable.- data
A
data.tabledataset ideally or at least adata.frame.- ...
Additional arguments passed to
lmer, used to control model fitting.
Examples
## these examples are slow to run
data(aces_daily, package = "JWileymisc")
evaluateLags(
"NegAff ~ Female + Age + BornAUS + (1 | UserID)",
"STRESS",
4L,
"UserID",
aces_daily)
#> Lag Model Estimator N_Obs N_Groups AIC BIC LL LLDF
#> <int> <char> <char> <num> <char> <num> <num> <num> <num>
#> 1: 0 merMod ML 5372 UserID (190) 6751.190 6797.313 -3368.595 7
#> 2: 1 merMod ML 5372 UserID (190) 6681.458 6734.169 -3332.729 8
#> 3: 2 merMod ML 5372 UserID (190) 6648.660 6707.960 -3315.330 9
#> 4: 3 merMod ML 5372 UserID (190) 6638.343 6704.233 -3309.172 10
#> 5: 4 merMod ML 5372 UserID (190) 6640.292 6712.771 -3309.146 11
#> Sigma MarginalR2 ConditionalR2 MarginalF2 ConditionalF2
#> <num> <num> <num> <num> <num>
#> 1: 0.4301553 0.2941101 0.5707243 0.4166516 1.329505
#> 2: 0.4276613 0.3233266 0.5801990 0.4778178 1.382081
#> 3: 0.4264497 0.3399432 0.5867621 0.5150211 1.419913
#> 4: 0.4260388 0.3483062 0.5903241 0.5344629 1.440954
#> 5: 0.4260312 0.3478455 0.5901544 0.5333790 1.439943
#> Formula
#> <char>
#> 1: NegAff ~ Female + Age + BornAUS + (1 | UserID) + STRESS
#> 2: NegAff ~ Female + Age + BornAUS + (1 | UserID) + STRESS + STRESSLag1
#> 3: NegAff ~ Female + Age + BornAUS + (1 | UserID) + STRESS + STRESSLag1 + STRESSLag2
#> 4: NegAff ~ Female + Age + BornAUS + (1 | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3
#> 5: NegAff ~ Female + Age + BornAUS + (1 | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3 + STRESSLag4
# \donttest{
## not run, more complex example with random slope, fails to converge
evaluateLags(
"NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID)",
"STRESS",
5L,
"UserID",
aces_daily)
#> Warning: Model failed to converge with max|grad| = 0.00302696 (tol = 0.002, component 1)
#> Lag Model Estimator N_Obs N_Groups AIC BIC LL LLDF
#> <int> <char> <char> <num> <char> <num> <num> <num> <num>
#> 1: 0 merMod ML 5144 UserID (190) 5903.510 5962.420 -2942.755 9
#> 2: 1 merMod ML 5144 UserID (190) 5837.591 5903.046 -2908.795 10
#> 3: 2 merMod ML 5144 UserID (190) 5816.798 5888.799 -2897.399 11
#> 4: 3 merMod ML 5144 UserID (190) 5806.522 5885.069 -2891.261 12
#> 5: 4 merMod ML 5144 UserID (190) 5808.423 5893.515 -2891.211 13
#> 6: 5 merMod ML 5144 UserID (190) 5810.158 5901.796 -2891.079 14
#> Sigma MarginalR2 ConditionalR2 MarginalF2 ConditionalF2
#> <num> <num> <num> <num> <num>
#> 1: 0.4021802 0.2441862 0.5872656 0.3230772 1.422865
#> 2: 0.4001612 0.2785727 0.5956186 0.3861410 1.472913
#> 3: 0.3994679 0.2943976 0.6000910 0.4172287 1.500569
#> 4: 0.3990020 0.3040614 0.6039227 0.4369085 1.524760
#> 5: 0.3989923 0.3033384 0.6036954 0.4354171 1.523311
#> 6: 0.3989851 0.3044128 0.6041170 0.4376342 1.525999
#> Formula
#> <char>
#> 1: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS
#> 2: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1
#> 3: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2
#> 4: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3
#> 5: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3 + STRESSLag4
#> 6: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3 + STRESSLag4 + STRESSLag5
## use different control to fit model and now converges
strictControl <- lme4::lmerControl(optCtrl = list(
algorithm = "NLOPT_LN_NELDERMEAD",
xtol_abs = 1e-10,
ftol_abs = 1e-10))
evaluateLags(
"NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID)",
"STRESS",
5L,
"UserID",
aces_daily,
control = strictControl)
#> Lag Model Estimator N_Obs N_Groups AIC BIC LL LLDF
#> <int> <char> <char> <num> <char> <num> <num> <num> <num>
#> 1: 0 merMod ML 5144 UserID (190) 5903.510 5962.420 -2942.755 9
#> 2: 1 merMod ML 5144 UserID (190) 5837.591 5903.046 -2908.795 10
#> 3: 2 merMod ML 5144 UserID (190) 5816.798 5888.799 -2897.399 11
#> 4: 3 merMod ML 5144 UserID (190) 5806.522 5885.069 -2891.261 12
#> 5: 4 merMod ML 5144 UserID (190) 5808.423 5893.515 -2891.211 13
#> 6: 5 merMod ML 5144 UserID (190) 5810.158 5901.796 -2891.079 14
#> Sigma MarginalR2 ConditionalR2 MarginalF2 ConditionalF2
#> <num> <num> <num> <num> <num>
#> 1: 0.4021799 0.2441830 0.5872689 0.3230717 1.422885
#> 2: 0.4001612 0.2785749 0.5956152 0.3861453 1.472892
#> 3: 0.3994694 0.2944043 0.6000779 0.4172423 1.500487
#> 4: 0.3990019 0.3040583 0.6039258 0.4369019 1.524779
#> 5: 0.3989926 0.3033362 0.6036964 0.4354127 1.523318
#> 6: 0.3989852 0.3044130 0.6041160 0.4376347 1.525992
#> Formula
#> <char>
#> 1: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS
#> 2: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1
#> 3: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2
#> 4: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3
#> 5: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3 + STRESSLag4
#> 6: NegAff ~ Female + Age + BornAUS + (1 + STRESS | UserID) + STRESS + STRESSLag1 + STRESSLag2 + STRESSLag3 + STRESSLag4 + STRESSLag5
# }