Table 4. Performance measures for the algorithms implemented.

	Class average (Macro)*			Weighted average (micro)
Algorithm	recall	precision	F-score	recall	precision	F-score
Manually modified rules	.994	1.000	.997	1.000	1.000	1.000
Rule Induction**	.626	.793	.677	.991	.981	.979
Naïve Bayes	.983	.939	.955	.994	.996	.994
Decision Trees**	.290	.416	.311	.936	.937	.923

^*The total number of groups in the training data was 16, and the total number of instances 53,146.

^**The Rule Induction algorithms failed to learn 3 classes, and the Decision Tree 11 classes.