https://github.com/palday/lme4-vs-mixedmodels.jl
Convergence in lme4 and MixedModels.jl
https://github.com/palday/lme4-vs-mixedmodels.jl
Last synced: 3 months ago
JSON representation
Convergence in lme4 and MixedModels.jl
- Host: GitHub
- URL: https://github.com/palday/lme4-vs-mixedmodels.jl
- Owner: palday
- Created: 2015-06-05T06:35:15.000Z (about 10 years ago)
- Default Branch: master
- Last Pushed: 2015-06-05T08:42:47.000Z (about 10 years ago)
- Last Synced: 2025-01-12T09:12:12.577Z (5 months ago)
- Size: 2.46 MB
- Stars: 0
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.Rmd
Awesome Lists containing this project
README
---
title: "Model Fits in lme4 and MixedModels.jl"
author: "Phillip Alday"
date: "June 2015"
output:
html_document:
keep_md: yes
toc: yes
---
```{r setup, include=FALSE}
options(width=120)
```
R session info:
```{r rsession,echo=TRUE,prompt=TRUE,comment=''}
sessionInfo()
```
Julia session info:
```
julia> println(versioninfo())
Julia Version 0.3.8
Commit 79599ad (2015-04-30 23:40 UTC)
Platform Info:
System: Linux (x86_64-linux-gnu)
CPU: Intel(R) Core(TM) i5 CPU M 480 @ 2.67GHz
WORD_SIZE: 64
BLAS: libblas.so.3
LAPACK: liblapack.so.3
LIBM: libopenlibm
LLVM: libLLVM-3.3
nothing
julia> println(Pkg.installed("MixedModels"))
0.3.22
```
# Models fit by Maximum Likelihood
## R
```{r rml,cache=TRUE,echo=TRUE,prompt=TRUE,comment=''}
load("aldayetal2014.RData")
library(lme4)
d.small <- subset(d,chan %in% c("CZ","CPZ","PZ"))
n400 <- subset(d.small, win=="N400")
system.time(m0 <- lmer(meanuv ~ wordOrder*ambiguity*np1type*np2type + (1|subj) + (1|item),data=n400,REML=FALSE)
)
print(summary(m0),correlation=FALSE)
system.time(m1 <- lmer(meanuv ~ wordOrder*ambiguity*np1type*np2type + (wordOrder+ambiguity|subj) + (wordOrder+ambiguity|item),data=n400,REML=FALSE)
)
print(summary(m1),correlation=FALSE)
system.time(m2 <- lmer(meanuv ~ wordOrder*ambiguity*np1type*np2type + (wordOrder+ambiguity+np1type+np2type|subj) + (wordOrder+ambiguity+np1type+np2type|item),data=n400,REML=FALSE)
)
print(summary(m2),correlation=FALSE)
```
## Julia
```{r eval=FALSE,include=FALSE}
using DataFrames, MixedModels
d = DataFrame(read_rda("aldayetal2014.RData")["d"]);
dsmall = d[bool([x in ["CZ","CPZ","PZ"] for x in d[:chan]]),:];
n400 = dsmall[dsmall[:win] .== "N400",:];
@time m0 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1|item) + (1|subj), n400));
print(m0)
@time m1 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1+wordOrder+ambiguity|item) + (1+wordOrder+ambiguity|subj), n400));
print(m1);
@time m2 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1+wordOrder+ambiguity+np1type+np2type|item) + (1+wordOrder+ambiguity+np1type+np2type|subj), n400));
print(m2)
```
```
julia> using DataFrames, MixedModels
julia> d = DataFrame(read_rda("aldayetal2014.RData")["d"]);
julia> dsmall = d[bool([x in ["CZ","CPZ","PZ"] for x in d[:chan]]),:];
julia> n400 = dsmall[dsmall[:win] .== "N400",:];
julia> @time m0 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1|item) + (1|subj), n400));
elapsed time: 20.650423511 seconds (1259493348 bytes allocated, 24.05% gc time)
julia> print(m0)
Linear mixed model fit by maximum likelihood
Formula: meanuv ~ 1 + wordOrder * ambiguity * np1type * np2type + (1 | item) + (1 | subj)
logLik: -158029.004553, deviance: 316058.009106
Variance components:
Variance Std.Dev.
item 0.25483327 0.50481013
subj 1.46886921 1.21196915
Residual 24.83310014 4.98328207
Number of obs: 52191; levels of grouping factors: 60, 37
Fixed-effects parameters:
Estimate Std.Error z value
(Intercept) -0.170135 0.227059 -0.749301
wordOrdersubject 0.810334 0.123448 6.56415
ambiguityunambig 0.872724 0.123444 7.06978
np1typepronoun 0.0996264 0.123553 0.806347
np2typepronoun 1.46224 0.123776 11.8136
wordOrdersubject&ambiguityunambig -0.626527 0.174739 -3.5855
wordOrdersubject&np1typepronoun -0.201881 0.174717 -1.15548
wordOrdersubject&np2typepronoun -0.300485 0.174889 -1.71815
ambiguityunambig&np1typepronoun -0.112868 0.174758 -0.645855
ambiguityunambig&np2typepronoun -0.108149 0.17483 -0.618594
np1typepronoun&np2typepronoun -0.432515 0.17523 -2.46827
wordOrdersubject&ambiguityunambig 0.0269254 0.247212 0.108916
wordOrdersubject&ambiguityunambig -0.0601712 0.247325 -0.243288
wordOrdersubject&np1typepronoun 0.880053 0.247806 3.55139
ambiguityunambig&np1typepronoun 0.446759 0.24746 1.80538
wordOrdersubject&ambiguityunambig -0.850636 0.350306 -2.42827
julia> @time m1 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1+wordOrder+ambiguity|item) + (1+wordOrder+ambiguity|subj), n400));
elapsed time: 13.114836832 seconds (1768485232 bytes allocated, 50.79% gc time)
julia> print(m1);
Linear mixed model fit by maximum likelihood
Formula: meanuv ~ 1 + wordOrder * ambiguity * np1type * np2type + ((1 + wordOrder + ambiguity) | item) + ((1 + wordOrder + ambiguity) | subj)
logLik: -157888.376486, deviance: 315776.752972
Variance components:
Variance Std.Dev. Corr.
item 0.35132918 0.59273027
0.23458607 0.48434086 -0.30
0.24918868 0.49918802 -0.42 -0.42
subj 1.32575435 1.15141406
0.19840059 0.44542182 0.16
0.33737448 0.58083947 -0.11 -0.11
Residual 24.58280904 4.95810539
Number of obs: 52191; levels of grouping factors: 60, 37
Fixed-effects parameters:
ambiguityunambig&np2typepronoun -0.112348 0.174184 -0.644994
np1typepronoun&np2typepronoun -0.468859 0.174777 -2.68261
wordOrdersubject&ambiguityunambig 0.0581888 0.246326 0.236226
wordOrdersubject&ambiguityunambig -0.0308533 0.24644 -0.125196
wordOrdersubject&np1typepronoun 0.918377 0.247248 3.71439
ambiguityunambig&np1typepronoun 0.452145 0.246883 1.83141
wordOrdersubject&ambiguityunambig -0.903582 0.349536 -2.58509
julia> @time m2 = fit(lmm(meanuv ~ 1 + wordOrder*ambiguity*np1type*np2type + (1+wordOrder+ambiguity+np1type+np2type|item) + (1+wordOrder+ambiguity+np1type+np2type|subj), n400));
elapsed time: 28.146366217 seconds (2426816680 bytes allocated, 37.34% gc time)
julia> print(m2)
Linear mixed model fit by maximum likelihood
Formula: meanuv ~ 1 + wordOrder * ambiguity * np1type * np2type + ((1 + wordOrder + ambiguity + np1type + np2type) | item) + ((1 + wordOrder + ambiguity + np1type + np2type) | subj)
logLik: -157234.785839, deviance: 314469.571677
Variance components:
Variance Std.Dev. Corr.
item 0.61958276 0.78713580
0.23555342 0.48533846 -0.21
0.25190194 0.50189833 -0.33 -0.33
0.30335998 0.55078125 -0.22 -0.22 -0.22
0.61963811 0.78717096 -0.61 -0.61 -0.61 -0.61
subj 2.02160641 1.42183206
0.20111041 0.44845335 0.11
0.33079992 0.57515208 -0.24 -0.24
0.16601203 0.40744574 0.03 0.03 0.03
2.02052768 1.42145267 -0.55 -0.55 -0.55 -0.55
Residual 23.81509690 4.88007140
Number of obs: 52191; levels of grouping factors: 60, 37
Fixed-effects parameters:
Estimate Std.Error z value
(Intercept) -0.163416 0.268861 -0.607807
wordOrdersubject 0.810187 0.155077 5.2244
ambiguityunambig 0.847608 0.166738 5.08349
np1typepronoun 0.129621 0.155731 0.832338
np2typepronoun 1.45031 0.282292 5.13763
wordOrdersubject&ambiguityunambig -0.598355 0.171535 -3.48825
wordOrdersubject&np1typepronoun -0.230189 0.17169 -1.34072
wordOrdersubject&np2typepronoun -0.290458 0.171717 -1.69149
ambiguityunambig&np1typepronoun -0.101199 0.171542 -0.589941
ambiguityunambig&np2typepronoun -0.0967029 0.17158 -0.563601
np1typepronoun&np2typepronoun -0.477141 0.172047 -2.77332
wordOrdersubject&ambiguityunambig 0.0132734 0.2429 0.0546457
wordOrdersubject&ambiguityunambig -0.0620641 0.242925 -0.255487
wordOrdersubject&np1typepronoun 0.929022 0.243358 3.81751
ambiguityunambig&np1typepronoun 0.45835 0.243008 1.88615
wordOrdersubject&ambiguityunambig -0.910206 0.344039 -2.64565
```