
320 ◾ Kweku-Muata Osei-Bryson
Problem P
SubTree-Aug
, which is the same as problem P
SubTree
but which also
includes any relevant threshold constraints (i.e., ACC≥0.90; STAB
A
≥0.90;
DSCPWR≥0.75; SIMPL
Leaf
>0.00; and SIMPL
Rule
>0.00), was formulated and
solved for two dierent weight vectors: (a) 1.00000, 0.00000, 0.00000, 0.00000,
and 0.00000, which corresponds to the traditional approach to post-pruning that
considers accuracy as the only objective and (b) 0.252763, 0.203765, 0.246988,
0.150937, and 0.145547, which is the weight vector that was generated in step2.
It should be noted that in generating the pruned DT1a and DT2a, we did not ...