Application of Random Forest Survival Models to Increase Generalizability of Decision Trees: A Case Study in Acute Myocardial Infarction

Table 2

Assessment of performance of different tree construction methods using either training or test sets.

	-index			IBS
	Training set	Test set	Percent change	Training set	Test set	Percent change
Saturated tree	0.872 (0.863, 0.882)	0.634 (0.528, 0.743)	27%	0.088 (0.082, 0.094)	0.224 (0.157, 0.298)	150%
Pruned tree	0.753 (0.740, 0.768)	0.699 (0.570, 0.824)	7%	0.145 (0.138, 0.151)	0.166 (0.113, 0.221)	14%
RSF	0.710 (0.693, 0.729)	0.716 (0.609, 0.857)	0.08%	0.163 (0.156, 0.169)	0.163 (0.114, 0.210)	0.1%