Table 4

The variance of F1 for four algorithms and four functional groups

algorithms

< 60

60-100

100-300

> 300


twoclass

0.0636

0.0929

0.1034

0.0536

twoclassbal

0.0142

0.0083

0.0462

0.0215

PSoL

0.0594

0.0813

0.0629

0.0456

SPE-RNE

0.0523

0.0485

0.0359

0.0420


Table 4 lists the variances of F1 for all algorithms on four functional terms groups. These values can evaluate the stability of algorithms on different groups. While only few training examples are used to learn classifier, over-fitting may occur and makes algorithm unstable.

Chen et al. BMC Genomics 2010 11(Suppl 2):S11   doi:10.1186/1471-2164-11-S2-S11

Open Data