如果使用 doParallel 和 recipes 则出现插入错误

Posted 2023-02-19

【中文标题】如果使用 doParallel 和 recipes 则出现插入错误

【英文标题】：caret error if doParallel and recipes is used

【发布时间】：2018-08-28 04:46:34

【问题描述】：

当我将 caret 与新的食谱包（即 caret::train.recipe()）一起使用时，如果 doParallel 也用于注册并行后端，我会收到一条错误消息。附件是一个可重现的示例（插入符号文档中官方食谱示例的简短版本）：

### Error only when I use doParallel
library(doParallel)
cl <- makeCluster(3)
registerDoParallel(cl)
###

library(caret)
library(recipes)
library(dplyr)
library(QSARdata)

data(AquaticTox)
tox <- AquaticTox_moe2D
ncol(tox)

tox$Activity <- AquaticTox_Outcome$Activity

tox <- tox %>%
  select(-Molecule) %>%
  mutate(manufacturability  = 1/moe2D_Weight) %>%
  mutate(manufacturability = manufacturability/sum(manufacturability))

wt_rmse <- function (pred, obs, wts, na.rm = TRUE) 
  sqrt(weighted.mean((pred - obs)^2, wts, na.rm = na.rm))

model_stats <- function(data, lev = NULL, model = NULL) 
  stats <- defaultSummary(data, lev = lev, model = model)
  res <- wt_rmse(pred = data$pred,
                 obs = data$obs, 
                 wts = data$manufacturability)
  c(wRMSE = res, stats)


tox_recipe <- recipe(Activity ~ ., data = tox) %>%
  add_role(manufacturability, new_role = "performance var")

tox_recipe

tox_ctrl <- trainControl(method = "cv", summaryFunction = model_stats, allowParallel = TRUE)
set.seed(888)
tox_svm <- train(tox_recipe, tox,
                 method = "svmRadial", 
                 metric = "wRMSE",
                 maximize = FALSE,
                 tuneLength = 10,
                 trControl = tox_ctrl)
tox_svm

如果我评论前三行，这会顺利进行。但如果不是：

Error in e$fun(obj, substitute(ex), parent.frame(), e$data) : 
unable to find variable "optimism_boot"

同样的错误出现在旧版本的 caret 中（在没有使用食谱的情况下），参见。这里： https://github.com/topepo/caret/issues/706

它在不使用 doParallel 或使用 doParallel 但没有配方时不会出现的意义上已修复，例如通过 train.formula。我正在使用所有软件包的最新版本：

   > sessionInfo()
R version 3.4.4 (2018-03-15)
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=German_Germany.1252  LC_CTYPE=German_Germany.1252    LC_MONETARY=German_Germany.1252
[4] LC_NUMERIC=C                    LC_TIME=German_Germany.1252    

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

other attached packages:
 [1] doParallel_1.0.11 iterators_1.0.9   foreach_1.4.4     kernlab_0.9-25    bindrcpp_0.2      QSARdata_1.3     
 [7] recipes_0.1.2     broom_0.4.3       dplyr_0.7.4       caret_6.0-78      ggplot2_2.2.1     lattice_0.20-35  

loaded via a namespace (and not attached):
 [1] tidyselect_0.2.4   purrr_0.2.4        reshape2_1.4.3     splines_3.4.4      colorspace_1.3-2   stats4_3.4.4      
 [7] yaml_2.1.18        survival_2.41-3    prodlim_1.6.1      rlang_0.2.0        ModelMetrics_1.1.0 pillar_1.2.1      
[13] withr_2.1.2        foreign_0.8-69     glue_1.2.0         bindr_0.1.1        plyr_1.8.4         dimRed_0.1.0      
[19] lava_1.6           robustbase_0.92-8  stringr_1.3.0      timeDate_3043.102  munsell_0.4.3      gtable_0.2.0      
[25] codetools_0.2-15   psych_1.7.8        class_7.3-14       DEoptimR_1.0-8     Rcpp_0.12.16       scales_0.5.0      
[31] ipred_0.9-6        CVST_0.2-1         mnormt_1.5-5       stringi_1.1.7      RcppRoll_0.2.2     ddalpha_1.3.1.1   
[37] grid_3.4.4         tools_3.4.4        magrittr_1.5       lazyeval_0.2.1     tibble_1.4.2       tidyr_0.8.0       
[43] DRR_0.0.3          pkgconfig_2.0.1    MASS_7.3-49        Matrix_1.2-12      lubridate_1.7.2    gower_0.1.2       
[49] assertthat_0.2.0   R6_2.2.2           rpart_4.1-13       sfsmisc_1.1-2      nnet_7.3-12        nlme_3.1-131.1    
[55] compiler_3.4.4   

该错误也出现在运行 Windows 10 的第二台 PC 上。非常感谢任何帮助或建议！

【问题讨论】：

在github上安装devel版本（今天要cran了）

【参考方案1】：

安装开发版：

devtools::install_github('topepo/caret/pkg/caret')

今天要去 CRAN。

【讨论】：

更多详情在这里https://github.com/topepo/caret/issues/860