Software Defect Prediction through Neural Network and Feature Selections

Table 3

RBF training and testing results using all features in the data set.

Data set	Feature number	Training results				Testing results
		Precision	Recall	F-measure	Accuracy	Precision	Recall	F-measure	Accuracy
CM1	22	94.47	95.86	0.95	93.20	92.18	95.53	0.93	90.89
JM1	22	85.34	86.82	0.89	84.38	81.99	83.98	0.85	81.18
KC1	22	83.02	83.87	0.86	82.99	84.78	90.10	0.84	80.05
KC2	22	81.83	81.08	0.84	79.59	79.82	81.01	0.79	76.49
KC3	39	80.99	81.97	0.85	80.09	81.47	80.76	0.81	78.25
KC4	39	84.45	85.72	0.84	83.89	81.08	82.09	0.83	80.08
MC1	38	99.91	100	1.00	100	99.01	99.41	0.99	98.89
MC2	39	72.63	76.82	0.77	70.08	70.59	77.13	0.73	67.29
MW1	37	89.91	90.09	0.93	88.67	88.82	90.92	0.91	86.89
PC1	22	98.99	99.98	1.00	100	97.99	98.91	0.99	97.78
PC2	36	98.09	98.99	1.00	100	99.01	99.32	0.99	98.42
PC3	37	94.42	97.73	0.98	93.27	92.23	93.12	0.95	90.07
PC4	37	94.19	96.76	0.97	94.20	91.89	93.78	0.93	91.40
PC5	38	80.82	83.79	0.82	79.84	78.01	81.40	0.80	77.32