CODEWITHSUNDEEP
rfunctionlme4
Steen Harsted
Steen Harsted

Reputation: 1932

lmer returns error when run inside a function

I am trying to make a function that fits a mixed model on the chosen variable.

If I run this command inside the function: 

lme4::lmer(VAR ~ cVAR + (1|ID), data=df)

I get the error:

Error in model.frame.default(data = df, drop.unused.levels = TRUE, formula = VAR ~  : 
  variable lengths differ (found for 'ID')

I have tested my code in several ways, and I just cant find out whats wrong. I can make plots from inside the function without problems. If I try to just print the dataframe used inside the function it looks exactly as I want it to look. Furthermore - If I run the lmer command outside the function (substituting VAR and cVAR for X and cX), the command runs without problems.

I therefore believe the error is caused by running lmer() inside the function. This Question has a similar problem, but is having a different error message than me.

Is there any way to run lmer inside a function without getting the error?

Reproducible example:

  set.seed(1)

org <- data.frame(
  X = rep(1:10+rnorm(10,mean=1,sd=0.5),2),
  cX = rep(1:10+rnorm(10,mean=1,sd=0.5),2),
  Y = rep(1:10+rnorm(10,mean=1,sd=0.5), 2),
  cY = rep(1:10-rnorm(10,mean=1,sd=0.5),2),
  Frame = rep(1:5, 4),
  JN = rep(1:2, each=5),
  ID = rep(1:2, each=10))

FUNC <- function(VAR) {
  VAR <- deparse(substitute(VAR))
  cVAR <- paste0("c", VAR)
  df <- dplyr::select(org, ID, JN, Frame, VAR, cVAR) # Keeping only a subset of the data
  df <- df %>% tidyr::drop_na() #Keeping only complete cases

  stopifnot(all(c(VAR, cVAR) %in% names(df)))

  MixedModel <- lme4::lmer(VAR ~ cVAR + (1|ID), data=df)
  b <- plot(MixedModel)
}

#It works fine outside the function:
MixedModel_no_FUNC <- lme4::lmer(X ~ cX + (1|ID), data=org)
plot(MixedModel_no_FUNC)

#But not inside the function
FUNC(X)

My sessionInfo in case it is relevant

R version 3.4.1 (2017-06-30)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

Matrix products: default

locale:
[1] LC_COLLATE=Danish_Denmark.1252  LC_CTYPE=Danish_Denmark.1252   
[3] LC_MONETARY=Danish_Denmark.1252 LC_NUMERIC=C                   
[5] LC_TIME=Danish_Denmark.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] lme4_1.1-14          Matrix_1.2-10        ggExtra_0.7          BlandAltmanLeh_0.3.1
 [5] gganimate_0.1.0.9000 gridExtra_2.3        bindrcpp_0.2         dplyr_0.7.3         
 [9] purrr_0.2.3          readr_1.1.1          tidyr_0.7.1          tibble_1.3.4        
[13] tidyverse_1.1.1      ggmap_2.7            ggplot2_2.2.1       

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.13      lubridate_1.6.0   lattice_0.20-35   png_0.1-7         assertthat_0.2.0 
 [6] digest_0.6.12     psych_1.7.8       mime_0.5          R6_2.2.2          cellranger_1.1.0 
[11] plyr_1.8.4        httr_1.3.1        RgoogleMaps_1.4.1 rlang_0.1.2       lazyeval_0.2.0   
[16] readxl_1.0.0      minqa_1.2.4       geosphere_1.5-7   miniUI_0.1.1      nloptr_1.0.4     
[21] splines_3.4.1     proto_1.0.0       labeling_0.3      stringr_1.2.0     foreign_0.8-69   
[26] munsell_0.4.3     shiny_1.0.5       broom_0.4.2       compiler_3.4.1    httpuv_1.3.5     
[31] modelr_0.1.1      pkgconfig_2.0.1   mnormt_1.5-5      htmltools_0.3.6   tidyselect_0.2.1 
[36] MASS_7.3-47       bitops_1.0-6      grid_3.4.1        nlme_3.1-131      jsonlite_1.5     
[41] xtable_1.8-2      gtable_0.2.0      magrittr_1.5      scales_0.5.0      stringi_1.1.5    
[46] mapproj_1.2-5     reshape2_1.4.2    sp_1.2-5          xml2_1.1.1        rjson_0.2.15     
[51] tools_3.4.1       forcats_0.2.0     glue_1.1.1        maps_3.2.0        hms_0.3          
[56] jpeg_0.1-8        parallel_3.4.1    yaml_2.1.14       colorspace_1.3-2  rvest_0.3.2      
[61] knitr_1.17        bindr_0.1         haven_1.1.0

Upvotes: 0

Views: 284

Answers (1)

trosendal
trosendal

Reputation: 1234

You have some conflicting names of objects in your dataframe and the environment in your function. Here's your data:

org <- data.frame(
    X = rep(1:10+rnorm(10,mean=1,sd=0.5),2),
    cX = rep(1:10+rnorm(10,mean=1,sd=0.5),2),
    Y = rep(1:10+rnorm(10,mean=1,sd=0.5), 2),
    cY = rep(1:10-rnorm(10,mean=1,sd=0.5),2),
    Frame = rep(1:5, 4),
    JN = rep(1:2, each=5),
    ID = rep(1:2, each=10))

I choose to just pass a string to the VAR argument instead of deparsing like you did. Then check if VAR and cVAR are in the dataframe. Then I reduce the dataframe and rename VAR and cVAR to Y and X so that when you call lmer() you're sure that the formula is referring to the variables in the data.frame and not the objects in the function environment.

library(lme4)
FUNC <- function(VAR = "VAR", df) {
    cVAR <- paste0("c", VAR)
    stopifnot(all(c(VAR, cVAR) %in% names(df)))
    df <- df[,c("ID", "JN", "Frame", VAR, cVAR)]
    names(df) <- c(c("ID", "JN", "Frame", "Y", "X"))
    df <- df[complete.cases(df),]
    MixedModel <- lme4::lmer(Y ~ X + (1|ID), data=df)
    plot(MixedModel)
}

FUNC(VAR = "X", df = org)

Upvotes: 1

Related Questions