avgPPVtop10 statistics.

The quality of predictive models built from data in terms of avgPPVtop10 statistics and their distribution. Only models that passed the AUROC > 0.70 threshold are included. Higher avgPPVtop10 values reflect models that are stronger for predicting outliers and alerts. The N/A entry records the number of models for which less than 10 patient instances had action a =1.

avgPPVtop10 range	# of lab-order omission models	# of medication-order omission models	total # of models
(0.90, 1.00]	77	62	139
(0.80, 0.90]	12	22	34
(0.70, 0.80]	4	24	28
(0.60, 0.70]	4	18	22
(0.50, 0.60]	2	30	32
(0.40, 0.50]	5	28	33
(0.30, 0.40]	4	17	21
(0.20, 0.30]	6	33	39
(0.10, 0.20]	17	72	89
[0, 0.10]	38	304	342
N/A	8	80	88
all	177	690	867