The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

xgboost: Multiclass Classification

# nolint start
library(mlexperiments)
library(mllrnrs)

See https://github.com/kapsner/mllrnrs/blob/main/R/learner_xgboost.R for implementation details.

Preprocessing

Import and Prepare Data

library(mlbench)
data("DNA")
dataset <- DNA |>
  data.table::as.data.table() |>
  na.omit()

feature_cols <- colnames(dataset)[160:180]
target_col <- "Class"

General Configurations

seed <- 123
if (isTRUE(as.logical(Sys.getenv("_R_CHECK_LIMIT_CORES_")))) {
  # on cran
  ncores <- 2L
} else {
  ncores <- ifelse(
    test = parallel::detectCores() > 4,
    yes = 4L,
    no = ifelse(
      test = parallel::detectCores() < 2L,
      yes = 1L,
      no = parallel::detectCores()
    )
  )
}
options("mlexperiments.bayesian.max_init" = 4L)
options("mlexperiments.optim.xgb.nrounds" = 20L)
options("mlexperiments.optim.xgb.early_stopping_rounds" = 5L)

Generate Training- and Test Data

data_split <- splitTools::partition(
  y = dataset[, get(target_col)],
  p = c(train = 0.7, test = 0.3),
  type = "stratified",
  seed = seed
)

train_x <- model.matrix(
  ~ -1 + .,
  dataset[data_split$train, .SD, .SDcols = feature_cols]
)
train_y <- as.integer(dataset[data_split$train, get(target_col)]) - 1L


test_x <- model.matrix(
  ~ -1 + .,
  dataset[data_split$test, .SD, .SDcols = feature_cols]
)
test_y <- as.integer(dataset[data_split$test, get(target_col)]) - 1L

Generate Training Data Folds

fold_list <- splitTools::create_folds(
  y = train_y,
  k = 3,
  type = "stratified",
  seed = seed
)

Experiments

Prepare Experiments

# required learner arguments, not optimized
learner_args <- list(
  objective = "multi:softprob",
  eval_metric = "mlogloss",
  num_class = 3
)

# set arguments for predict function and performance metric,
# required for mlexperiments::MLCrossValidation and
# mlexperiments::MLNestedCV
predict_args <- list(reshape = TRUE)
performance_metric <- metric("ACC")
performance_metric_args <- NULL
return_models <- FALSE

# required for grid search and initialization of bayesian optimization
parameter_grid <- expand.grid(
  subsample = seq(0.6, 1, .2),
  colsample_bytree = seq(0.6, 1, .2),
  min_child_weight = seq(1, 5, 4),
  learning_rate = seq(0.1, 0.2, 0.1),
  max_depth = seq(1, 5, 4)
)
# reduce to a maximum of 10 rows
if (nrow(parameter_grid) > 10) {
  set.seed(123)
  sample_rows <- sample(seq_len(nrow(parameter_grid)), 10, FALSE)
  parameter_grid <- kdry::mlh_subset(parameter_grid, sample_rows)
}

# required for bayesian optimization
parameter_bounds <- list(
  subsample = c(0.2, 1),
  colsample_bytree = c(0.2, 1),
  min_child_weight = c(1L, 10L),
  learning_rate = c(0.1, 0.2),
  max_depth =  c(1L, 10L)
)
optim_args <- list(
  n_iter = ncores,
  kappa = 3.5,
  acq = "ucb"
)

Hyperparameter Tuning

tuner <- mlexperiments::MLTuneParameters$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  strategy = "grid",
  ncores = ncores,
  seed = seed
)

tuner$parameter_grid <- parameter_grid
tuner$learner_args <- learner_args
tuner$split_type <- "stratified"

tuner$set_data(
  x = train_x,
  y = train_y
)

tuner_results_grid <- tuner$execute(k = 3)
#>
#> Parameter settings [============================>-------------------------------------------------------------------] 3/10 ( 30%)
#> Parameter settings [=====================================>----------------------------------------------------------] 4/10 ( 40%)
#> Parameter settings [===============================================>------------------------------------------------] 5/10 ( 50%)
#> Parameter settings [=========================================================>--------------------------------------] 6/10 ( 60%)
#> Parameter settings [==================================================================>-----------------------------] 7/10 ( 70%)
#> Parameter settings [============================================================================>-------------------] 8/10 ( 80%)
#> Parameter settings [=====================================================================================>----------] 9/10 ( 90%)
#> Parameter settings [===============================================================================================] 10/10 (100%)

head(tuner_results_grid)
#>    setting_id metric_optim_mean nrounds subsample colsample_bytree min_child_weight learning_rate max_depth      objective
#>         <int>             <num>   <int>     <num>            <num>            <num>         <num>     <num>         <char>
#> 1:          1         1.0106675      38       0.6              0.8                5           0.2         1 multi:softprob
#> 2:          2         0.9828797      37       1.0              0.8                5           0.1         5 multi:softprob
#> 3:          3         1.0102800      76       0.8              0.8                5           0.1         1 multi:softprob
#> 4:          4         0.9867769      20       0.6              0.8                5           0.2         5 multi:softprob
#> 5:          5         0.9815158      32       1.0              0.8                1           0.1         5 multi:softprob
#> 6:          6         0.9741743      50       0.8              0.8                5           0.1         5 multi:softprob
#>    eval_metric num_class
#>         <char>     <num>
#> 1:    mlogloss         3
#> 2:    mlogloss         3
#> 3:    mlogloss         3
#> 4:    mlogloss         3
#> 5:    mlogloss         3
#> 6:    mlogloss         3

Bayesian Optimization

tuner <- mlexperiments::MLTuneParameters$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  strategy = "bayesian",
  ncores = ncores,
  seed = seed
)

tuner$parameter_grid <- parameter_grid
tuner$parameter_bounds <- parameter_bounds

tuner$learner_args <- learner_args
tuner$optim_args <- optim_args

tuner$split_type <- "stratified"

tuner$set_data(
  x = train_x,
  y = train_y
)

tuner_results_bayesian <- tuner$execute(k = 3)
#>
#> Registering parallel backend using 4 cores.

head(tuner_results_bayesian)
#>    Epoch setting_id subsample colsample_bytree min_child_weight learning_rate max_depth gpUtility acqOptimum inBounds Elapsed
#>    <num>      <int>     <num>            <num>            <num>         <num>     <num>     <num>     <lgcl>   <lgcl>   <num>
#> 1:     0          1       0.6              0.8                5           0.2         1        NA      FALSE     TRUE   0.995
#> 2:     0          2       1.0              0.8                5           0.1         5        NA      FALSE     TRUE   1.057
#> 3:     0          3       0.8              0.8                5           0.1         1        NA      FALSE     TRUE   1.088
#> 4:     0          4       0.6              0.8                5           0.2         5        NA      FALSE     TRUE   1.035
#> 5:     0          5       1.0              0.8                1           0.1         5        NA      FALSE     TRUE   0.282
#> 6:     0          6       0.8              0.8                5           0.1         5        NA      FALSE     TRUE   0.314
#>         Score metric_optim_mean nrounds errorMessage      objective eval_metric num_class
#>         <num>             <num>   <int>       <lgcl>         <char>      <char>     <num>
#> 1: -1.0106675         1.0106675      38           NA multi:softprob    mlogloss         3
#> 2: -0.9828797         0.9828797      37           NA multi:softprob    mlogloss         3
#> 3: -1.0102800         1.0102800      76           NA multi:softprob    mlogloss         3
#> 4: -0.9867769         0.9867769      20           NA multi:softprob    mlogloss         3
#> 5: -0.9815158         0.9815158      32           NA multi:softprob    mlogloss         3
#> 6: -0.9741743         0.9741743      50           NA multi:softprob    mlogloss         3

k-Fold Cross Validation

validator <- mlexperiments::MLCrossValidation$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  fold_list = fold_list,
  ncores = ncores,
  seed = seed
)

validator$learner_args <- tuner$results$best.setting[-1]

validator$predict_args <- predict_args
validator$performance_metric <- performance_metric
validator$performance_metric_args <- performance_metric_args
validator$return_models <- return_models

validator$set_data(
  x = train_x,
  y = train_y
)

validator_results <- validator$execute()
#>
#> CV fold: Fold1
#>
#> CV fold: Fold2
#>
#> CV fold: Fold3
#> CV progress [========================================================================================================] 3/3 (100%)
#>

head(validator_results)
#>      fold performance subsample colsample_bytree min_child_weight learning_rate max_depth nrounds      objective eval_metric
#>    <char>       <num>     <num>            <num>            <num>         <num>     <num>   <int>         <char>      <char>
#> 1:  Fold1   0.5685484 0.7220588        0.8330246                1     0.1099728        10      19 multi:softprob    mlogloss
#> 2:  Fold2   0.5587045 0.7220588        0.8330246                1     0.1099728        10      19 multi:softprob    mlogloss
#> 3:  Fold3   0.5483871 0.7220588        0.8330246                1     0.1099728        10      19 multi:softprob    mlogloss
#>    num_class
#>        <num>
#> 1:         3
#> 2:         3
#> 3:         3

Nested Cross Validation

validator <- mlexperiments::MLNestedCV$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  strategy = "grid",
  fold_list = fold_list,
  k_tuning = 3L,
  ncores = ncores,
  seed = seed
)

validator$parameter_grid <- parameter_grid
validator$learner_args <- learner_args
validator$split_type <- "stratified"

validator$predict_args <- predict_args
validator$performance_metric <- performance_metric
validator$performance_metric_args <- performance_metric_args
validator$return_models <- return_models

validator$set_data(
  x = train_x,
  y = train_y
)

validator_results <- validator$execute()
#>
#> CV fold: Fold1
#>
#> Parameter settings [============================>-------------------------------------------------------------------] 3/10 ( 30%)
#> Parameter settings [=====================================>----------------------------------------------------------] 4/10 ( 40%)
#> Parameter settings [===============================================>------------------------------------------------] 5/10 ( 50%)
#> Parameter settings [=========================================================>--------------------------------------] 6/10 ( 60%)
#> Parameter settings [==================================================================>-----------------------------] 7/10 ( 70%)
#> Parameter settings [============================================================================>-------------------] 8/10 ( 80%)
#> Parameter settings [=====================================================================================>----------] 9/10 ( 90%)
#> Parameter settings [===============================================================================================] 10/10 (100%)
#> CV fold: Fold2
#> CV progress [====================================================================>-----------------------------------] 2/3 ( 67%)
#>
#> Parameter settings [============================>-------------------------------------------------------------------] 3/10 ( 30%)
#> Parameter settings [=====================================>----------------------------------------------------------] 4/10 ( 40%)
#> Parameter settings [===============================================>------------------------------------------------] 5/10 ( 50%)
#> Parameter settings [=========================================================>--------------------------------------] 6/10 ( 60%)
#> Parameter settings [==================================================================>-----------------------------] 7/10 ( 70%)
#> Parameter settings [============================================================================>-------------------] 8/10 ( 80%)
#> Parameter settings [=====================================================================================>----------] 9/10 ( 90%)
#> Parameter settings [===============================================================================================] 10/10 (100%)
#> CV fold: Fold3
#> CV progress [========================================================================================================] 3/3 (100%)
#>
#> Parameter settings [============================>-------------------------------------------------------------------] 3/10 ( 30%)
#> Parameter settings [=====================================>----------------------------------------------------------] 4/10 ( 40%)
#> Parameter settings [===============================================>------------------------------------------------] 5/10 ( 50%)
#> Parameter settings [=========================================================>--------------------------------------] 6/10 ( 60%)
#> Parameter settings [==================================================================>-----------------------------] 7/10 ( 70%)
#> Parameter settings [============================================================================>-------------------] 8/10 ( 80%)
#> Parameter settings [=====================================================================================>----------] 9/10 ( 90%)
#> Parameter settings [===============================================================================================] 10/10 (100%)

head(validator_results)
#>      fold performance nrounds subsample colsample_bytree min_child_weight learning_rate max_depth      objective eval_metric
#>    <char>       <num>   <int>     <num>            <num>            <num>         <num>     <num>         <char>      <char>
#> 1:  Fold1   0.5591398      32       0.6              1.0                1           0.1         5 multi:softprob    mlogloss
#> 2:  Fold2   0.5344130      34       0.8              0.8                5           0.1         5 multi:softprob    mlogloss
#> 3:  Fold3   0.5510753      24       0.6              1.0                1           0.1         5 multi:softprob    mlogloss
#>    num_class
#>        <num>
#> 1:         3
#> 2:         3
#> 3:         3

Inner Bayesian Optimization

validator <- mlexperiments::MLNestedCV$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  strategy = "bayesian",
  fold_list = fold_list,
  k_tuning = 3L,
  ncores = ncores,
  seed = seed
)

validator$parameter_grid <- parameter_grid
validator$learner_args <- learner_args
validator$split_type <- "stratified"


validator$parameter_bounds <- parameter_bounds
validator$optim_args <- optim_args

validator$predict_args <- predict_args
validator$performance_metric <- performance_metric
validator$performance_metric_args <- performance_metric_args
validator$return_models <- TRUE

validator$set_data(
  x = train_x,
  y = train_y
)

validator_results <- validator$execute()
#>
#> CV fold: Fold1
#>
#> Registering parallel backend using 4 cores.
#>
#> CV fold: Fold2
#> CV progress [====================================================================>-----------------------------------] 2/3 ( 67%)
#>
#> Registering parallel backend using 4 cores.
#>
#> CV fold: Fold3
#> CV progress [========================================================================================================] 3/3 (100%)
#>
#> Registering parallel backend using 4 cores.

head(validator_results)
#>      fold performance subsample colsample_bytree min_child_weight learning_rate max_depth nrounds      objective eval_metric
#>    <char>       <num>     <num>            <num>            <num>         <num>     <num>   <int>         <char>      <char>
#> 1:  Fold1   0.5591398 0.6000000        1.0000000                1     0.1000000         5      32 multi:softprob    mlogloss
#> 2:  Fold2   0.5641026 0.6936177        0.7695365                1     0.1099728        10      20 multi:softprob    mlogloss
#> 3:  Fold3   0.5537634 0.5955781        0.8688622                1     0.1099728        10      19 multi:softprob    mlogloss
#>    num_class
#>        <num>
#> 1:         3
#> 2:         3
#> 3:         3

Holdout Test Dataset Performance

Predict Outcome in Holdout Test Dataset

preds_xgboost <- mlexperiments::predictions(
  object = validator,
  newdata = test_x
)

Evaluate Performance on Holdout Test Dataset

perf_xgboost <- mlexperiments::performance(
  object = validator,
  prediction_results = preds_xgboost,
  y_ground_truth = test_y
)
perf_xgboost
#>     model performance
#>    <char>       <num>
#> 1:  Fold1   0.5590387
#> 2:  Fold2   0.5579937
#> 3:  Fold3   0.5611285

Appendix I: Grid-Search with Target Weigths

Here, xgboost’s weight-argument is used to rescale the case-weights during the training.

# define the target weights
y_weights <- ifelse(train_y == 1, 0.8, ifelse(train_y == 2, 1.2, 1))
head(y_weights)
#> [1] 1.2 1.2 0.0 0.8 0.8 0.0
tuner_w_weights <- mlexperiments::MLTuneParameters$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  strategy = "grid",
  ncores = ncores,
  seed = seed
)

tuner_w_weights$parameter_grid <- parameter_grid
tuner_w_weights$learner_args <- c(
  learner_args,
  list(case_weights = y_weights)
)
tuner_w_weights$split_type <- "stratified"

tuner_w_weights$set_data(
  x = train_x,
  y = train_y
)

tuner_results_grid <- tuner_w_weights$execute(k = 3)
#>
#> Parameter settings [============================>-------------------------------------------------------------------] 3/10 ( 30%)
#> Parameter settings [=====================================>----------------------------------------------------------] 4/10 ( 40%)
#> Parameter settings [===============================================>------------------------------------------------] 5/10 ( 50%)
#> Parameter settings [=========================================================>--------------------------------------] 6/10 ( 60%)
#> Parameter settings [==================================================================>-----------------------------] 7/10 ( 70%)
#> Parameter settings [============================================================================>-------------------] 8/10 ( 80%)
#> Parameter settings [=====================================================================================>----------] 9/10 ( 90%)
#> Parameter settings [===============================================================================================] 10/10 (100%)

head(tuner_results_grid)
#>    setting_id metric_optim_mean nrounds subsample colsample_bytree min_child_weight learning_rate max_depth      objective
#>         <int>             <num>   <int>     <num>            <num>            <num>         <num>     <num>         <char>
#> 1:          1         0.9442324      50       0.6              0.8                5           0.2         1 multi:softprob
#> 2:          2         0.9217258      35       1.0              0.8                5           0.1         5 multi:softprob
#> 3:          3         0.9443002      93       0.8              0.8                5           0.1         1 multi:softprob
#> 4:          4         0.9245540      20       0.6              0.8                5           0.2         5 multi:softprob
#> 5:          5         0.9212009      26       1.0              0.8                1           0.1         5 multi:softprob
#> 6:          6         0.9145242      39       0.8              0.8                5           0.1         5 multi:softprob
#>    eval_metric num_class
#>         <char>     <num>
#> 1:    mlogloss         3
#> 2:    mlogloss         3
#> 3:    mlogloss         3
#> 4:    mlogloss         3
#> 5:    mlogloss         3
#> 6:    mlogloss         3

Appendix II: k-Fold Cross Validation with Target Weigths

validator_w_weights <- mlexperiments::MLCrossValidation$new(
  learner = mllrnrs::LearnerXgboost$new(
    metric_optimization_higher_better = FALSE
  ),
  fold_list = fold_list,
  ncores = ncores,
  seed = seed
)

# append the optimized setting from above with the newly created weights
validator_w_weights$learner_args <- c(
  tuner_w_weights$results$best.setting[-1]
)

validator_w_weights$predict_args <- predict_args
validator_w_weights$performance_metric <- performance_metric
validator_w_weights$performance_metric_args <- performance_metric_args
validator_w_weights$return_models <- return_models

validator_w_weights$set_data(
  x = train_x,
  y = train_y
)

validator_results <- validator_w_weights$execute()
#>
#> CV fold: Fold1
#>
#> CV fold: Fold2
#>
#> CV fold: Fold3
#> CV progress [========================================================================================================] 3/3 (100%)
#>

head(validator_results)
#>      fold performance nrounds subsample colsample_bytree min_child_weight learning_rate max_depth      objective eval_metric
#>    <char>       <num>   <int>     <num>            <num>            <num>         <num>     <num>         <char>      <char>
#> 1:  Fold1   0.5658602      39       0.8              0.8                5           0.1         5 multi:softprob    mlogloss
#> 2:  Fold2   0.5222672      39       0.8              0.8                5           0.1         5 multi:softprob    mlogloss
#> 3:  Fold3   0.5577957      39       0.8              0.8                5           0.1         5 multi:softprob    mlogloss
#>    num_class
#>        <num>
#> 1:         3
#> 2:         3
#> 3:         3

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.