Skip to main content
. 2016 Apr 23;17:299. doi: 10.1186/s12864-016-2619-0

Table 3.

The comparison of two classification models of SPAID using alternate control groupsa

Genotypes from marker(s) Control group Tree TP FP Precision Recall F-measure AUC
CNV_14.3 and CNV_16.1 C2 J48 0.715 0.389 0.706 0.715 0.708 0.605
RF 0.715 0.389 0.706 0.715 0.708 0.613
C1 J48 0.810 0.822 0.671 0.810 0.734 0.575
RF 0.810 0.822 0.671 0.810 0.734 0.575
CNV_14.3 C2 J48 0.715 0.389 0.706 0.715 0.708 0.605
RF 0.715 0.389 0.706 0.715 0.708 0.611
C1 J48 0.820 0.820 0.673 0.820 0.739 0.457
RF 0.820 0.820 0.673 0.820 0.739 0.575
CNV_16.1 C2 J48 0.681 0.473 0.661 0.681 0.661 0.599
RF 0.711 0.391 0.702 0.711 0.704 0.612
C1 J48 0.820 0.820 0.673 0.820 0.739 0.457
RF 0.810 0.822 0.671 0.810 0.734 0.580

aTwo control groups were considered. Control 2 (C2, n = 80) contained all dogs free from SPAID, irrespective of age, whilst Control 1 (C1, n = 34) included only those individuals older than 60 months with no signs of SPAID. The counts required to calculate the receiver operator curve (ROC) area are reported, including true positive (TP) and false positive (FP)