Better error message when step_dummy on two or more variables and one of them have NA

[LluisRamon]

Thank you for the package, I find it very useful!

There is an error when using step_dummy with two or more variables and one of them have NA (not sure if it is the expected behaviour).

Error message relates to differing number of rows which is hard to understand which is the problem. Especially if you have a long or complex recipe.

I attach a reproducible example.

library("recipes")
library("dplyr")

data(okc)
okc <- okc[complete.cases(okc),]

# Two variables to dummy -> Works fine
okc$sunny_location <- sample(c("Florida", "Barcelona", "California"), nrow(okc), replace = TRUE)

rec <- recipe(age ~ ., data = okc)

dummies <- rec %>% step_dummy(diet, sunny_location)
dummies <- prep(dummies, training = okc)

dummy_data <- bake(dummies, newdata = okc)

# Variable with some missing values -> Not clear error
okc$sunny_location <- sample(c("Florida", "Barcelona", "California"), nrow(okc), replace = TRUE)

# Add some missing values
okc$sunny_location[1:5] <- NA

rec <- recipe(age ~ ., data = okc)

dummies <- rec %>% step_dummy(diet, sunny_location)
dummies <- prep(dummies, training = okc)

dummy_data <- bake(dummies, newdata = okc)
# Error in data.frame(..., check.names = FALSE) : 
#   arguments imply differing number of rows: 35495, 35490

My sessionInfo in case you need it.

R version 3.4.3 (2017-11-30)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS High Sierra 10.13.3

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

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

other attached packages:
[1] recipes_0.1.2 broom_0.4.2   dplyr_0.7.4  

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.15      ddalpha_1.3.1     gower_0.1.2       pillar_1.2.1      compiler_3.4.3    DEoptimR_1.0-8    plyr_1.8.4        bindr_0.1        
 [9] class_7.3-14      tools_3.4.3       rpart_4.1-11      ipred_0.9-6       lubridate_1.6.0   tibble_1.4.2      nlme_3.1-131      lattice_0.20-35  
[17] pkgconfig_2.0.1   rlang_0.2.0       Matrix_1.2-12     psych_1.7.8       cli_1.0.0         rstudioapi_0.7    yaml_2.1.14       parallel_3.4.3   
[25] RcppRoll_0.2.2    prodlim_1.6.1     bindrcpp_0.2      stringr_1.2.0     tidyselect_0.2.2  nnet_7.3-12       CVST_0.2-1        grid_3.4.3       
[33] glue_1.2.0        robustbase_0.92-7 R6_2.2.2          survival_2.41-3   foreign_0.8-69    lava_1.5.1        kernlab_0.9-25    DRR_0.0.2        
[41] tidyr_0.7.2       reshape2_1.4.2    purrr_0.2.4       magrittr_1.5      splines_3.4.3     MASS_7.3-47       sfsmisc_1.1-1     dimRed_0.1.0     
[49] assertthat_0.2.0  mnormt_1.5-5      timeDate_3012.100 utf8_1.1.3        stringi_1.1.5     crayon_1.3.4   

[jlopezper]

Hi @LluisRamon, how do you finally dealt with it? Is there any way to overcome this issue without renounce of using recipes? Thanks!

[topepo]

This should solve the problem by not generating an error but instead assigning missing values to the resulting indicator variables.

[jlopezper]

Thank you!

[LluisRamon]

Great, thanks!