Table 11.

Average accuracy of different feature subsets of the three-class classification experiment.

Number of features	Accuracy of SVM	Accuracy of J48	The average accuracy of each set of features
Using 20 Features	100%	92.81%	96.40%
Using 10 Features	100%	93.50%	96.75%
Using 5 Features	93.14%	91.48%	92.31%
Using 3 Features	89.63%	89.96%	89.79%
Using 2 Features	87.11%	88.66%	87.89%
Using 1 Feature	85.18%	86.77%	85.98%