Figure 3.

Comparison of LMB using AUC error vs (a) different tree algorithms (ID3 and CART) and (b) different ensemble methods (LogitBoost and Random Forests) on 13 medical datasets. Points above the black line indicate results where LMB was better.