Predicting foodborne pathogens and probiotics taxa within poultry-related microbiomes using a machine learning approach

Table 1 Comparison of random forest (RF), support vector machine (SVM) and logistic regression (LogReg). Model performance was considered good when 5-fold cross-validation value ≥ 0.7

Number	Target Variables	RF	SVM	LogReg
	Pathogens
1	Salmonella	0.85	0.85	0.78
2	Campylobacter	0.80	0.65	0.74
3	Listeria	0.86	0.86	0.79
	Common Farm Practice Variables
4	BroodBedding	0.91	0.91	0.88
5	BrGMOFree	0.78	0.69	0.72
6	BrSoyFree	0.84	0.83	0.8
7	BrMedicated	0.97	0.97	0.94
8	AvgAgeToPasture	0.73	0.61	0.71
9	PastureHousing	0.72	0.55	0.62
10	FreqHousingMove	0.98	0.98	0.96
11	AlwaysNewPasture	0.9	0.87	0.86
12	PaGMOFree	0.77	0.67	0.73
13	PaSoyFree	0.76	0.64	0.71
14	PaMedicated	0.98	0.98	0.95
15	LayersOnFarm	0.95	0.95	0.93
16	CattleOnFarm	0.77	0.6	0.7
17	SwineOnFarm	0.83	0.8	0.81
18	GoatsOnFarm	0.74	0.63	0.7
19	SheepOnFarm	0.78	0.55	0.7
20	WaterSource	0.7	0.52	0.62
21	FreqBirdHandling	0.9	0.87	0.86
22	AnyABXUse	0.98	0.98	0.95
23	AnimalSource	0.90	0.85	0.87
	Physicochemical Properties
24	pH	0.93	0.56	0.81
25	EC	0.96	0.65	0.89
26	Moisture	0.97	0.66	0.91
27	TotalC	0.94	0.6	0.85
28	TotalN	0.94	0.63	0.83
29	CNRatio	0.98	0.55	0.82
30	Al	0.92	0.55	0.73
31	B	0.94	0.58	0.77
32	Ca	0.95	0.63	0.84
33	Cd	0.98	0.65	0.92
34	Cr	0.98	0.9	0.92
35	Cu	0.94	0.6	0.84
36	Fe	0.95	0.61	0.87
37	K	0.97	0.55	0.9
38	Mg	0.95	0.69	0.85
39	Mn	0.91	0.63	0.81
40	Mo	0.98	0.7	0.92
41	Na	0.96	0.68	0.88
42	Ni	0.96	0.65	0.89
43	P	0.97	0.67	0.89
44	Pb	0.91	0.55	0.79
45	S	0.94	0.57	0.78
46	Si	0.93	0.54	0.73
47	Zn	0.92	0.64	0.85

ISSN: 2524-4671