
====================================================================
  Loading data
====================================================================
  Train : 630,000 samples x 13 features
  Test  : 270,000 samples
  Target: ['Absence' 'Presence'] -> [0, 1]
  Positive rate: 0.4483

====================================================================
  FOLD 1/5
====================================================================
  Train: 504,000  Holdout: 126,000

  [HVRT] Fitting on 504,000 samples...
  [HVRT] fit done in 1.8s
  [HVRT] Reducing to 100,000 samples (method=variance_ordered)...
  [HVRT] reduce done in 5.7s  | positive rate: 0.6240

  [HPO] Running 50 trials (3-fold inner CV on 100,000 samples)...
  [HPO] 22108s  |  Best inner CV AUC: 0.9571
  [HPO] Best params:
          n_rounds: 2000
          learning_rate: 0.2
          max_depth: 3
          refit_interval: 20
          y_weight: 0.9
          min_samples_leaf: 20
          hvrt_min_samples_leaf: 20
          reduce_ratio: 0.7
          auto_expand: True
          expand_ratio: 0.1
          convergence_tol: None

  [Refit] Fitting best model on full 100k reduced set...
  [Refit] done in 469.0s

  [Eval] Holdout AUC: 0.9542
  [Time] Fold total: 22593s (HVRT 7s, HPO 22108s, Refit 469s)
  [Sub]  Fold submission: sample_submission_cv_fold1.csv

====================================================================
  FOLD 2/5
====================================================================
  Train: 504,000  Holdout: 126,000

  [HVRT] Fitting on 504,000 samples...
  [HVRT] fit done in 1.5s
  [HVRT] Reducing to 100,000 samples (method=variance_ordered)...
  [HVRT] reduce done in 4.8s  | positive rate: 0.6234

  [HPO] Running 50 trials (3-fold inner CV on 100,000 samples)...
