Skip to main content

Table 5 The algorithm to select the classifiers to be combined.

From: Machine learning integration for predicting the effect of single amino acid substitutions on protein stability

1:

Initialize the subset Z as empty set

2:

Initialize the subset R as all possible 102 (R.D.B) groups

3:

Remove the most accurate (R.D.B) from R and add to Z

4:

Perform McNemar's test for all pairs between Z and R

5:

Decrease the degree of confidence, α, for McNemar's test

6:

if There is at least one diverse (R.D.B) in R then

7:

Select the most accurate and most diverse (R.D.B) from R and add it to Z

8:

Go to Step 4

9:

else

10:

Use the (R.D.B) triplets in Z as the current base-learners to be combined

11:

end if

  1. The aim is to select the most accurate and at the same time the most diverse classifiers.