xgboost: Binary 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("PimaIndiansDiabetes2")
dataset <- PimaIndiansDiabetes2 |>
  data.table::as.data.table() |>
  na.omit()

feature_cols <- colnames(dataset)[1:8]
target_col <- "diabetes"

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" = 10L)
options("mlexperiments.optim.xgb.nrounds" = 100L)
options("mlexperiments.optim.xgb.early_stopping_rounds" = 10L)

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 = "binary:logistic",
  eval_metric = "logloss"
)

# set arguments for predict function and performance metric,
# required for mlexperiments::MLCrossValidation and
# mlexperiments::MLNestedCV
predict_args <- NULL
performance_metric <- metric("auc")
performance_metric_args <- list(positive = "1")
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(
  iters.n = 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 [==================>-----------------------------------------------------------------------------] 2/10 ( 20%)
#> 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
#> 1:          1         0.4121967      34       0.6              0.8                5           0.2         1 binary:logistic
#> 2:          2         0.3890956      57       1.0              0.8                5           0.1         5 binary:logistic
#> 3:          3         0.3925308     100       0.8              0.8                5           0.1         1 binary:logistic
#> 4:          4         0.4082505      34       0.6              0.8                5           0.2         5 binary:logistic
#> 5:          5         0.3975907      36       1.0              0.8                1           0.1         5 binary:logistic
#> 6:          6         0.3932451      66       0.8              0.8                5           0.1         5 binary:logistic
#>    eval_metric
#> 1:     logloss
#> 2:     logloss
#> 3:     logloss
#> 4:     logloss
#> 5:     logloss
#> 6:     logloss

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
#> 1:     0          1       0.6              0.8                5           0.2         1        NA      FALSE     TRUE   1.695
#> 2:     0          2       1.0              0.8                5           0.1         5        NA      FALSE     TRUE   1.702
#> 3:     0          3       0.8              0.8                5           0.1         1        NA      FALSE     TRUE   1.734
#> 4:     0          4       0.6              0.8                5           0.2         5        NA      FALSE     TRUE   1.724
#> 5:     0          5       1.0              0.8                1           0.1         5        NA      FALSE     TRUE   0.849
#> 6:     0          6       0.8              0.8                5           0.1         5        NA      FALSE     TRUE   0.850
#>         Score metric_optim_mean nrounds errorMessage       objective eval_metric
#> 1: -0.4089735         0.4089735      56           NA binary:logistic     logloss
#> 2: -0.3970937         0.3970937      49           NA binary:logistic     logloss
#> 3: -0.4013240         0.4013240     100           NA binary:logistic     logloss
#> 4: -0.4070968         0.4070968      69           NA binary:logistic     logloss
#> 5: -0.3819756         0.3819756      39           NA binary:logistic     logloss
#> 6: -0.3987643         0.3987643      99           NA binary:logistic     logloss

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

head(validator_results)
#>     fold performance subsample colsample_bytree min_child_weight learning_rate max_depth nrounds       objective eval_metric
#> 1: Fold1   0.8799577         1              0.8                1           0.1         5      39 binary:logistic     logloss
#> 2: Fold2   0.8635643         1              0.8                1           0.1         5      39 binary:logistic     logloss
#> 3: Fold3   0.9027699         1              0.8                1           0.1         5      39 binary:logistic     logloss

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 [==================================================================>-----------------------------] 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 [==================================================================>-----------------------------] 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 [===============================================>------------------------------------------------] 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
#> 1: Fold1   0.8675304      40       0.6                1                1           0.2         1 binary:logistic     logloss
#> 2: Fold2   0.8635643      44       1.0                1                5           0.1         5 binary:logistic     logloss
#> 3: Fold3   0.8793103      24       0.6                1                1           0.2         1 binary:logistic     logloss

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
#> 1: Fold1   0.8662084       0.6              1.0                1           0.2         1      28 binary:logistic     logloss
#> 2: Fold2   0.8746695       1.0              0.8                5           0.1         5      44 binary:logistic     logloss
#> 3: Fold3   0.8903335       0.6              1.0                1           0.1         5      30 binary:logistic     logloss

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,
  type = "binary"
)
perf_xgboost
#>    model performance       auc     prauc sensitivity specificity       ppv       npv tn tp fn fp       tnr       tpr       fnr
#> 1: Fold1   0.7922752 0.7922752 0.6016630   0.5128205   0.8734177 0.6666667 0.7840909 69 20 19 10 0.8734177 0.5128205 0.4871795
#> 2: Fold2   0.7687439 0.7687439 0.5601442   0.3846154   0.8860759 0.6250000 0.7446809 70 15 24  9 0.8860759 0.3846154 0.6153846
#> 3: Fold3   0.7594937 0.7594937 0.6142299   0.4871795   0.8481013 0.6129032 0.7701149 67 19 20 12 0.8481013 0.4871795 0.5128205
#>          fpr    bbrier       acc        ce     fbeta
#> 1: 0.1265823 0.1726355 0.7542373 0.2457627 0.5797101
#> 2: 0.1139241 0.1885316 0.7203390 0.2796610 0.4761905
#> 3: 0.1518987 0.1854326 0.7288136 0.2711864 0.5428571