An Ant Colony Optimization Based Feature Selection for Web Page Classification

Table 9

The best, worst, and average 𝐹-measure values of classification using ACO selected 𝑛 features from tagged terms for all classes when naïve Bayes classifier is used in the test phase.

No. of features selected (𝑛)	𝐹-measure	Dataset
		Course	Student	Project	Faculty	Conference
10	Best	0.933	0.864	0.917	0.936	0.897
	Avg.	0.856	0.863	0.88	0.926	0.855
	Worst	0.618	0.859	0.809	0.909	0.838
100	Best	0.868	0.38	0.912	0.875	0.797
	Avg.	0.858	0.354	0.901	0.762	0.695
	Worst	0.829	0.324	0.889	0.602	0.645
500	Best	0.543	0.54	0.502	0.527	0.826
	Avg.	0.51	0.539	0.484	0.51	0.821
	Worst	0.478	0.53	0.472	0.501	0.816