No `collect_metric()` exists for this type of object.

Hello,

Thank you for developing this amazing package!

I am trying to run your example code provided in your tidysdm overview vignette but I am getting an issue when I call the collect_metric() on the lacerta_ensemble object. Could you please take a look?

Reprex

library(tidysdm)
#> Loading required package: tidymodels
#> Loading required package: spatialsample
library(sf)
#> Linking to GEOS 3.12.0, GDAL 3.7.1, PROJ 9.2.1; sf_use_s2() is TRUE
library(pastclim)
#> Loading required package: terra
#> terra 1.7.55
#> 
#> Attaching package: 'terra'
#> The following object is masked from 'package:tidyr':
#> 
#>     extract
#> The following object is masked from 'package:scales':
#> 
#>     rescale
#> 
#> Attaching package: 'pastclim'
#> The following objects are masked from 'package:tidysdm':
#> 
#>     date2ybp, ybp2date
library(tidyterra)
#> 
#> Attaching package: 'tidyterra'
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(ggplot2)

data(lacerta)
lacerta
#> # A tibble: 1,297 × 3
#>           ID latitude longitude
#>        <dbl>    <dbl>     <dbl>
#>  1 858029749     42.6     -7.09
#>  2 858029738     42.6     -7.09
#>  3 614631090     41.4     -7.90
#>  4 614631085     41.3     -7.81
#>  5 614631083     41.3     -7.81
#>  6 614631080     41.4     -7.83
#>  7 614631072     41.4     -7.81
#>  8 614559731     40.3     -7.70
#>  9 614559728     40.4     -7.70
#> 10 614559657     40.4     -7.56
#> # ℹ 1,287 more rows

lacerta <- st_as_sf(lacerta, coords = c("longitude", "latitude"))
st_crs(lacerta) <- 4326

# Iberia peninsula extension
iberia_poly <- terra::vect("POLYGON((-9.8 43.3,-7.8 44.1,-2.0 43.7,3.6 42.5,3.8 41.5,1.3 40.8,0.3 39.5,0.9 38.6,-0.4 37.5,-1.6 36.7,-2.3 36.3,-4.1 36.4,-4.5 36.4,-5.0 36.1,-5.6 36.0,-6.3 36.0,-7.1 36.9,-9.5 36.6,-9.4 38.0,-10.6 38.9,-9.5 40.8,-9.8 43.3))")

crs(iberia_poly) <- "lonlat"
gdal(warn = 3)
land_mask <- rast(system.file("extdata/lacerta_land_mask.nc",
                              package = "tidysdm"))

ggplot() +
  geom_spatraster(data = land_mask, aes(fill = land_mask_1985)) +
  geom_sf(data = lacerta)


# Thinning step

set.seed(1234567)
lacerta <- thin_by_cell(lacerta, raster = land_mask)
nrow(lacerta)
#> [1] 231

ggplot() +
  geom_spatraster(data = land_mask, aes(fill = land_mask_1985)) +
  geom_sf(data = lacerta)


set.seed(1234567)
lacerta_thin <- thin_by_dist(lacerta, dist_min = km2m(20))
nrow(lacerta_thin)
#> [1] 113

ggplot() +
  geom_spatraster(data = land_mask, aes(fill = land_mask_1985)) +
  geom_sf(data = lacerta_thin)


set.seed(1234567)
lacerta_thin <- sample_pseudoabs(lacerta_thin,
                                 n = 3 * nrow(lacerta_thin),
                                 raster = land_mask,
                                 method = c("dist_min", km2m(50))
)

ggplot() +
  geom_spatraster(data = land_mask, aes(fill = land_mask_1985)) +
  geom_sf(data = lacerta_thin, aes(col = class))


gdal(warn = 3)
climate_present <-
  rast(system.file("extdata/lacerta_climate_present_10m.nc",
                   package = "tidysdm"))
climate_vars <- names(climate_present)

lacerta_thin <- lacerta_thin %>%
  bind_cols(terra::extract(climate_present, lacerta_thin, ID = FALSE))

lacerta_thin %>% plot_pres_vs_bg(class)

lacerta_thin %>% dist_pres_vs_bg(class)
#>     bio19     bio12     bio16     bio02     bio13     bio07     bio05     bio04 
#> 0.6935026 0.6929654 0.6885097 0.6776561 0.6679894 0.6542063 0.6243068 0.6071527 
#>     bio09     bio10     bio17     bio15     bio18     bio06     bio08     bio14 
#> 0.5506741 0.4855064 0.4318872 0.4241999 0.3600587 0.3454245 0.3108560 0.2954351 
#>     bio11     bio03     bio01  altitude 
#> 0.2317079 0.2173934 0.2045136 0.1554400

vars_to_keep <- lacerta_thin %>% dist_pres_vs_bg(class)
vars_to_keep <- names(vars_to_keep[vars_to_keep > 0.30])
lacerta_thin <- lacerta_thin %>% select(all_of(c(vars_to_keep, "class")))
vars_to_keep
#>  [1] "bio19" "bio12" "bio16" "bio02" "bio13" "bio07" "bio05" "bio04" "bio09"
#> [10] "bio10" "bio17" "bio15" "bio18" "bio06" "bio08"

suggested_vars <- c("bio05", "bio06", "bio13", "bio14", "bio15")
pairs(climate_present[[suggested_vars]])


climate_present <- climate_present[[suggested_vars]]
vars_uncor <- filter_high_cor(climate_present, cutoff = 0.7)
vars_uncor
#> [1] "bio15" "bio05" "bio13" "bio06"
#> attr(,"to_remove")
#> [1] "bio14"

lacerta_thin <- lacerta_thin %>% select(all_of(c(vars_uncor, "class")))
climate_present <- climate_present[[vars_uncor]]

lacerta_rec <- recipe(lacerta_thin, formula = class ~ .)
lacerta_rec
#> 
#> ── Recipe ──────────────────────────────────────────────────────────────────────
#> 
#> ── Inputs
#> Number of variables by role
#> outcome:   1
#> predictor: 4
#> coords:    2

lacerta_thin %>% check_sdm_presence(class)
#> [1] TRUE

lacerta_models <-
  # create the workflow_set
  workflow_set(
    preproc = list(default = lacerta_rec),
    models = list(
      # the standard glm specs
      glm = sdm_spec_glm(),
      # rf specs with tuning
      rf = sdm_spec_rf(),
      # boosted tree model (gbm) specs with tuning
      gbm = sdm_spec_boost_tree(),
      # maxent specs with tuning
      maxent = sdm_spec_maxent()
    ),
    # make all combinations of preproc and models,
    cross = TRUE
  ) %>%
  # tweak controls to store information needed later to create the ensemble
  option_add(control = control_ensemble_grid())

set.seed(100)
lacerta_cv <- spatial_block_cv(lacerta_thin, v = 5)
autoplot(lacerta_cv)


set.seed(1234567)
lacerta_models <-
  lacerta_models %>%
  workflow_map("tune_grid",
               resamples = lacerta_cv, grid = 3,
               metrics = sdm_metric_set(), verbose = TRUE
  )
#> i    No tuning parameters. `fit_resamples()` will be attempted
#> i 1 of 4 resampling: default_glm
#> ✔ 1 of 4 resampling: default_glm (188ms)
#> i 2 of 4 tuning:     default_rf
#> i Creating pre-processing data to finalize unknown parameter: mtry
#> ✔ 2 of 4 tuning:     default_rf (763ms)
#> i 3 of 4 tuning:     default_gbm
#> i Creating pre-processing data to finalize unknown parameter: mtry
#> ✔ 3 of 4 tuning:     default_gbm (4s)
#> i 4 of 4 tuning:     default_maxent
#> ✔ 4 of 4 tuning:     default_maxent (1.4s)

autoplot(lacerta_models)


lacerta_ensemble <- simple_ensemble() %>%
  add_member(lacerta_models, metric = "boyce_cont")
lacerta_ensemble
#> A simple_ensemble of models
#> 
#> Members:
#> • default_glm
#> • default_rf
#> • default_gbm
#> • default_maxent
#> 
#> Available metrics:
#> • boyce_cont
#> • roc_auc
#> • tss_max
#> 
#> Metric used to tune workflows:
#> • boyce_cont

autoplot(lacerta_ensemble)


lacerta_ensemble %>% collect_metrics()
#> Error in `collect_metrics()`:
#> ! No `collect_metric()` exists for this type of object.
#> Backtrace:
#>     ▆
#>  1. ├─lacerta_ensemble %>% collect_metrics()
#>  2. ├─tune::collect_metrics(.)
#>  3. └─tune:::collect_metrics.default(.)
#>  4.   └─rlang::abort("No `collect_metric()` exists for this type of object.")

Session info

R version 4.3.1 (2023-06-16)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Arch Linux

Matrix products: default
BLAS:   /usr/lib/libblas.so.3.11.0 
LAPACK: /usr/lib/liblapack.so.3.11.0

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8    LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C             LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

time zone: Europe/Lisbon
tzcode source: system (glibc)

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

other attached packages:
 [1] maxnet_0.1.4        xgboost_1.7.6.1     ranger_0.16.0       tidyterra_0.4.0     pastclim_2.0.0      terra_1.7-55       
 [7] sf_1.0-14           tidysdm_0.9.1       spatialsample_0.4.0 yardstick_1.2.0     workflowsets_1.0.1  workflows_1.1.3    
[13] tune_1.1.2          tidyr_1.3.0         tibble_3.2.1        rsample_1.2.0       recipes_1.0.8       purrr_1.0.2        
[19] parsnip_1.1.1       modeldata_1.2.0     infer_1.0.5         ggplot2_3.4.4       dplyr_1.1.3         dials_1.2.0        
[25] scales_1.2.1        broom_1.0.5         tidymodels_1.1.1   

loaded via a namespace (and not attached):
 [1] DBI_1.1.3           s2_1.1.4            testthat_3.2.0      rlang_1.1.1         magrittr_2.0.3      furrr_0.3.1        
 [7] e1071_1.7-13        compiler_4.3.1      vctrs_0.6.4         lhs_1.1.6           shape_1.4.6         pkgconfig_2.0.3    
[13] wk_0.8.0            backports_1.4.1     ellipsis_0.3.2      labeling_0.4.3      utf8_1.2.3          prodlim_2023.08.28 
[19] glmnet_4.1-8        reprex_2.0.2        overlapping_2.1     jsonlite_1.8.7      parallel_4.3.1      prettyunits_1.2.0  
[25] R6_2.5.1            parallelly_1.36.0   rpart_4.1.19        brio_1.1.3          lubridate_1.9.3     Rcpp_1.0.11        
[31] iterators_1.0.14    future.apply_1.11.0 Matrix_1.6-1.1      splines_4.3.1       nnet_7.3-19         timechange_0.2.0   
[37] tidyselect_1.2.0    rstudioapi_0.15.0   timeDate_4022.108   codetools_0.2-19    listenv_0.9.0       lattice_0.21-8     
[43] withr_2.5.1         future_1.33.0       survival_3.5-7      units_0.8-4         proxy_0.4-27        pillar_1.9.0       
[49] KernSmooth_2.23-22  foreach_1.5.2       generics_0.1.3      rprojroot_2.0.3     munsell_0.5.0       globals_0.16.2     
[55] class_7.3-22        glue_1.6.2          tools_4.3.1         data.table_1.14.8   gower_1.0.1         fs_1.6.3           
[61] grid_4.3.1          DALEX_2.4.3         ipred_0.9-14        colorspace_2.1-0    patchwork_1.1.3     cli_3.6.1          
[67] DiceDesign_1.9      fansi_1.0.5         lava_1.7.2.1        gtable_0.3.4        GPfit_1.0-8         digest_0.6.33      
[73] classInt_0.4-10     farver_2.1.1        lifecycle_1.0.3     hardhat_1.3.0       here_1.0.1          MASS_7.3-60

EvolEcolGroup / tidysdm

No `collect_metric()` exists for this type of object. #29

Reprex

Session info