Table 1 Optimized parameter values for trained classifiers.

SVM

SVM method

NU-SVC

Kernel type

RBF

Kernel function parameter γ

2^-5.5

Error parameter ν

0.536

Stopping criterion ε

0.00001

Sequence length

10

6 before cut site

4 after cut site

Random forest

Maximum depth

unlimited

Number of features

4

Number of trees

143

Sequence length

24

12 before cut site

12 after cut site

J48

Confidence factor

0.285

Minimum number of objects

5

Number of folds

3

Binary splits

true

Reduced error pruning

false

Subtree raising

true

Unpruned

false

Use Laplace

false

Sequence length

6

4 before cut site

2 after cut site