Skip to main content

Table 7 Final AdaBoost model scores

From: Ada-WHIPS: explaining AdaBoost classification with applications in the health sciences

   Undiscretised: used by Discretised: used by
   Ada-WHIPS & LORE Anchors
Data ntree Accuracy κ Accuracy κ
SAMME
Breast cancer 200 0.98 0.96 0.96 0.92
Cardiotocography 800 0.94 0.84 0.89 0.70
Diabetic retinopathy 1000 0.68 0.36 0.66 0.33
Cleveland heart 200 0.77 0.52 0.80 0.59
Mental health survey ’16 200 0.88 0.76 0.88 0.75
Mental health survey ’14 200 0.83 0.65 0.81 0.62
Hospital readmission 800 0.62 0.22 0.60 0.18
Thyroid 1200 0.97 0.92 0.80 0.45
Understanding society 600 0.64 0.13 0.61 0.14
SAMME.R
Breast cancer 1000 0.98 0.96 0.95 0.90
Cardiotocography 1600 0.94 0.82 0.88 0.67
Diabetic retinopathy 200 0.69 0.38 0.65 0.30
Cleveland heart 400 0.76 0.50 0.82 0.63
Mental health survey ’16 800 0.87 0.73 0.86 0.72
Mental health survey ’14 200 0.80 0.60 0.81 0.63
Hospital readmission 200 0.62 0.22 0.63 0.23
Thyroid 1600 0.97 0.92 0.76 0.37
Understanding society 200 0.62 0.13 0.62 0.15
  1. Accuracy and Cohen’s kappa on Held Out Data