Table 6. Nine attribute evaluators in WEKA are used for feature selection comparison, corresponding classification accuracy (%) based on Naive Bayes (NB), SVM, Decision Tree c4.5 (DT) and Random Forest (RF) of the selected features are listed.

Attribute Evaluators	Selected Features	NB	SVM	DT	FR	Avg. (%)
Correlation Ranking	16,19,29,37,2,20,28,18,30	80	80.0	63.3	73.3	74.2
Information Gain	16,13,12,11,40,15,14,17,10	63.3	63.3	56.7	66.7	62.5
OneR Evaluator	9,20,16,29,30,8,7,37,22	83.3	80.0	66.7	73.3	75.8
ReliefF Ranking	16,37,29,19,40,1, 20,2,28	80	76.7	60.0	76.7	73.4
Symmetrical Uncertainty	16,13,12,11,40,15,14,17,19	63.3	66.7	50.0	66.7	61.7
Decision tree (DT)	29,16,28,2,8,31, 37,22,35	86.7	80.0	53.3	73.3	73.3
RandomForest (RF)	29,28,16,9,8,18,20,25,40	76.7	70.0	63.3	70.0	70.0
Logistic Regression	29, 16,1, 5,7,9,10,13,14	70	70.0	70.0	76.7	71.7
Lazy KStar	16,30, 18,29,40, 38,7,8,13	76.7	66.7	66.7	76.7	71.7
Avg. (%)		76.7	72.7	61.3	72.0	70.5