Genomic data sampling and its effect on classification performance assessment

BMC Bioinformatics

Table 3 Summary of cross-validation results for splice-junction sequence data: Mean ± standard error of classification accuracies

Train/test runs	50%–50% cross validation	75%–25% cross validation	95%–5% cross validation
10	0.940 ± 0.0051	0.962 ± 0.0032	0.969 ± 0.0050
25	0.952 ± 0.0035	0.964 ± 0.0026	0.970 ± 0.0036
50	0.953 ± 0.0024	0.964 ± 0.0020	0.968 ± 0.0027
100	0.953 ± 0.0050	0.967 ± 0.0014	0.971 ± 0.0020
500	0.956 ± 0.0026	0.966 ± 0.0010	0.969 ± 0.0013
1000	0.957 ± 0.0019	0.968 ± 0.0008	0.969 ± 0.0011
2000	0.957 ± 0.0019	0.968 ± 0.0007	0.969 ± 0.0009
3000	0.958 ± 0.0016	0.966 ± 0.0007	0.968 ± 0.0014
4000	0.959 ± 0.0010	0.968 ± 0.0005	0.969 ± 0.0010
5000	0.957 ± 0.0009	0.968 ± 0.0005	0.970 ± 0.0008

ISSN: 1471-2105