Reliable genomic strategies for species classification of plant genetic resources

BMC Bioinformatics

Table 1 Median prediction accuracy in the 15 × 5,000 curated Helianthus datasets

Accessions per species	Misclassification rate (%)	RF	NB	NJ	1-NN	3-NN
	6.25	0.63	0.88	0.81	0.81	0.75
2	12.50	0.56	0.75	0.75	0.75	0.63
	18.75	0.50	0.69	0.69	0.69	0.63
	6.25	0.84	0.97	0.91	0.78	0.97
4	12.50	0.81	0.88	0.84	0.75	0.94
	18.75	0.78	0.81	0.78	0.69	0.88
	6.25	0.96	0.96	0.96	0.94	0.96
6	12.50	0.94	0.85	0.92	0.88	0.94
	18.75	0.90	0.77	0.88	0.81	0.88
	6.25	0.95	0.88	0.94	0.83	0.98
8	12.50	0.94	0.81	0.89	0.78	0.95
	18.75	0.92	0.70	0.84	0.72	0.91
	6.25	0.98	0.88	0.93	0.91	0.98
10	12.50	0.96	0.75	0.89	0.85	0.94
	18.75	0.96	0.65	0.85	0.79	0.89
	Median	0.92	0.81	0.88	0.79	0.94

Classifiers are Random Forest (RF), Naive Bayes (NB), Neighbour-Joining (NJ), 1-Nearest Neighbour (1-NN), and 3-Nearest Neighbours (3-NN) respectively. For each parameter combination, the highest median score is presented in bold

ISSN: 1471-2105