tune_bayes() error: Gaussian process model: data length exceeds size of matrix

[jcpsantiago]

The problem

I'm having trouble with using tune_bayes() in my pipeline. I realized it also fails with the example found in the docs https://www.tidymodels.org/learn/work/bayes-opt/, with the exact same error.

Reproducible example

library(tidymodels)
#> ── Attaching packages ───────────────────────────────────────────── tidymodels 0.1.1 ──
#> ✓ broom     0.7.0      ✓ recipes   0.1.13
#> ✓ dials     0.0.8      ✓ rsample   0.0.7 
#> ✓ dplyr     1.0.1      ✓ tibble    3.0.3 
#> ✓ ggplot2   3.3.2      ✓ tidyr     1.1.1 
#> ✓ infer     0.5.3      ✓ tune      0.1.1 
#> ✓ modeldata 0.0.2      ✓ workflows 0.1.3 
#> ✓ parsnip   0.1.3      ✓ yardstick 0.0.7 
#> ✓ purrr     0.3.4
#> ── Conflicts ──────────────────────────────────────────────── tidymodels_conflicts() ──
#> x purrr::discard() masks scales::discard()
#> x dplyr::filter()  masks stats::filter()
#> x dplyr::lag()     masks stats::lag()
#> x recipes::step()  masks stats::step()
library(modeldata)

# Load data
data(cells)

set.seed(2369)
tr_te_split <- initial_split(cells %>% select(-case), prop = 3/4)
cell_train <- training(tr_te_split)
cell_test  <- testing(tr_te_split)

set.seed(1697)
folds <- vfold_cv(cell_train, v = 10)

library(themis)
#> Registered S3 methods overwritten by 'themis':
#>   method               from   
#>   bake.step_downsample recipes
#>   bake.step_upsample   recipes
#>   prep.step_downsample recipes
#>   prep.step_upsample   recipes
#>   tidy.step_downsample recipes
#>   tidy.step_upsample   recipes
#> 
#> Attaching package: 'themis'
#> The following objects are masked from 'package:recipes':
#> 
#>     step_downsample, step_upsample, tunable.step_downsample,
#>     tunable.step_upsample

cell_pre_proc <-
  recipe(class ~ ., data = cell_train) %>%
  step_YeoJohnson(all_predictors()) %>%
  step_normalize(all_predictors()) %>%
  step_pca(all_predictors(), num_comp = tune()) %>%
  step_downsample(class)

svm_mod <-
  svm_rbf(mode = "classification", cost = tune(), rbf_sigma = tune()) %>%
  set_engine("kernlab")

svm_wflow <-
  workflow() %>%
  add_model(svm_mod) %>%
  add_recipe(cell_pre_proc)

svm_set <- parameters(svm_wflow)
svm_set
#> Collection of 3 parameters for tuning
#> 
#>         id parameter type object class
#>       cost           cost    nparam[+]
#>  rbf_sigma      rbf_sigma    nparam[+]
#>   num_comp       num_comp    nparam[+]

svm_set <- 
  svm_set %>% 
  update(num_comp = num_comp(c(0L, 20L)))

set.seed(12)
search_res <-
  svm_wflow %>% 
  tune_bayes(
    resamples = folds,
    # To use non-default parameter ranges
    param_info = svm_set,
    # Generate five at semi-random to start
    initial = 1,
    iter = 50,
    # How to measure performance?
    metrics = metric_set(roc_auc),
    control = control_bayes(no_improve = 30, verbose = TRUE)
  )
#> 
#> >  Generating a set of 1 initial parameter results
#> ✓ Initialization complete
#> 
#> Optimizing roc_auc using the expected improvement
#> 
#> ── Iteration 1 ────────────────────────────────────────────────────────────────────────
#> 
#> i Current best:      roc_auc=0.8678 (@iter 0)
#> i Gaussian process model
#> ! Gaussian process model: X should be in range (0, 1), data length exceeds...
#> x Gaussian process model: Error in GP_deviance(beta = row, X = X, Y = Y, n...
#> ! An error occurred when creating candidates parameters:  Error in GP_deviance(beta = row, X = X, Y = Y, nug_thres = nug_thres,  : 
#>   Infinite values of the Deviance Function, 
#>             unable to find optimum parameters
#> x Skipping to next iteration
#> Error in eval(expr, p): no loop for break/next, jumping to top level
#> x Optimization stopped prematurely; returning current results.

^{Created on 2020-08-30 by the reprex package (v0.3.0)}

Session info

sessionInfo()
#> R version 4.0.2 (2020-06-22)
#> Platform: x86_64-apple-darwin17.0 (64-bit)
#> Running under: macOS Catalina 10.15.6
#> 
#> Matrix products: default
#> BLAS:   /Library/Frameworks/R.framework/Versions/4.0/Resources/lib/libRblas.dylib
#> LAPACK: /Library/Frameworks/R.framework/Versions/4.0/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 datasets  utils     methods   base     
#> 
#> other attached packages:
#>  [1] themis_0.1.2     yardstick_0.0.7  workflows_0.1.3  tune_0.1.1      
#>  [5] tidyr_1.1.1      tibble_3.0.3     rsample_0.0.7    recipes_0.1.13  
#>  [9] purrr_0.3.4      parsnip_0.1.3    modeldata_0.0.2  infer_0.5.3     
#> [13] ggplot2_3.3.2    dplyr_1.0.1      dials_0.0.8      scales_1.1.1    
#> [17] broom_0.7.0      tidymodels_0.1.1
#> 
#> loaded via a namespace (and not attached):
#>  [1] splines_4.0.2      foreach_1.5.0      prodlim_2019.11.13 assertthat_0.2.1  
#>  [5] highr_0.8          unbalanced_2.0     GPfit_1.0-8        renv_0.11.0       
#>  [9] yaml_2.2.1         globals_0.12.5     ipred_0.9-9        pillar_1.4.6      
#> [13] backports_1.1.8    lattice_0.20-41    glue_1.4.1         pROC_1.16.2       
#> [17] digest_0.6.25      checkmate_2.0.0    hardhat_0.1.4      colorspace_1.4-1  
#> [21] mlr_2.17.1         htmltools_0.5.0    Matrix_1.2-18      plyr_1.8.6        
#> [25] timeDate_3043.102  pkgconfig_2.0.3    lhs_1.0.2          DiceDesign_1.8-1  
#> [29] listenv_0.8.0      parallelMap_1.5.0  RANN_2.6.1         gower_0.2.2       
#> [33] lava_1.6.7         generics_0.0.2     ellipsis_0.3.1     withr_2.2.0       
#> [37] furrr_0.1.0        ROSE_0.0-3         nnet_7.3-14        cli_2.0.2         
#> [41] survival_3.2-3     magrittr_1.5       crayon_1.3.4       evaluate_0.14     
#> [45] future_1.18.0      fansi_0.4.1        doParallel_1.0.15  MASS_7.3-51.6     
#> [49] class_7.3-17       FNN_1.1.3          data.table_1.13.0  tools_4.0.2       
#> [53] BBmisc_1.11        lifecycle_0.2.0    stringr_1.4.0      kernlab_0.9-29    
#> [57] munsell_0.5.0      compiler_4.0.2     rlang_0.4.7        grid_4.0.2        
#> [61] iterators_1.0.12   rstudioapi_0.11    rmarkdown_2.3      gtable_0.3.0      
#> [65] codetools_0.2-16   R6_2.4.1           ParamHelpers_1.14  lubridate_1.7.9   
#> [69] knitr_1.29         fastmatch_1.1-0    stringi_1.4.6      parallel_4.0.2    
#> [73] Rcpp_1.0.5         vctrs_0.3.2        rpart_4.1-15       tidyselect_1.1.0  
#> [77] xfun_0.16

Setting initial argument to > 1 fixes this issue.

[topepo]

I think the issue is related to:

    # Generate five at semi-random to start
    initial = 1,

You can't fit a model to one data point (not without getting infinite solutions). I try to make the initial set a minimum of the number of parameters + 1.

Can you try that and let us know the results?

[jcpsantiago]

Yes that works :) it was not obvious at the time, but now it makes sense of course. I had a wrong understand of the system. Maybe it should have an if guard-rail for distracted people like me?

[topepo]

That's a good point!

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.