Predicting Mycobacterium tuberculosis Complex Clades Using Knowledge-Based Bayesian Networks

Table 3

Results of the 𝐹-measures of KBBN based on out of-sample test. The KBBN model was trained on SITVIT-Train (with 2714 records) and tested on SITVIT-Test with 7949 records. Overall average 𝐹-measure is 0.939.

Clade	𝐹-measure	Clade	𝐹-measure	Clade	𝐹-measure
AFRI	0.889	H	0.942	PINI	0.667
AFRI_1	0.975	H1	0.977	PINI1	0.923
AFRI_2	0.926	H2	0.625	PINI2	0.522
AFRI_3	1.000	H3	0.944	S	0.956
Beijing	0.980	H3-Ural-1	0.887	T	0.969
BOV	0.981	H37Rv	1.000	T1-RUS2	0.956
BOV_1	0.996	H4-Ural-2	0.960	T2	0.991
BOV_2	1.000	LAM	0.949	T2-Uganda	1.000
BOV_3	1.000	LAM1	0.986	T3	0.969
BOV_4-Caprae	0.914	LAM11-ZWE	0.976	T3-ETH	0.977
Cameroon	0.967	LAM12-Madrid1	1.000	T3-OSA	0.978
Canetti	0.500	LAM2	0.993	T4	0.984
CAS	0.990	LAM3	0.973	T4-CEU1	1.000
CAS1-Delhi	0.990	LAM4	0.967	T5	1.000
CAS1-Kili	0.846	LAM5	0.985	T5-Madrid2	1.000
CAS2	1.000	LAM6	0.889	T5-RUS1	0.883
EAI	0.989	LAM8	0.970	T-Tuscany	0.889
EAI1-SOM	1.000	Manu_ancestor	1.000	Turkey	0.941
EAI2-Manila	1.000	Manu1	0.995	X1	0.963
EAI2-Nonthaburi	0.933	Manu2	0.997	X2	0.944
EAI3-IND	1.000	Manu3	1.000	X3	0.971
EAI4-VNM	1.000	Microti	0.667	ZERO	0.800
EAI6-BGD1	1.000
EAI7-BGD2	0.993			Average	0.939
EAI8-MDG	1.000