Software Defect Prediction through Neural Network and Feature Selections

Table 1

NASA data set.

Data set	Attributes number	All samples number	Defective samples number	Nondefective samples number	Missing attributes number	Defect rate, %
CM1	22	498	49	449	0	9.83
JM1	22	10885	2106	8779	0	19.35
KC1	22	2109	326	1783	0	15.45
KC2	22	522	105	415	0	20.50
KC3	39	194	36	158	0	18.55
KC4	39	125	44	81	0	35.20
MC1	38	1988	46	1942	0	2.31
MC2	39	125	44	81	0	35.2
MW1	37	253	27	226	0	10.67
PC1	22	1109	77	1032	0	6.94
PC2	36	745	16	729	0	2.14
PC3	37	1077	134	943	0	12.44
PC4	37	1287	177	1110	0	13.75
PC5	38	1711	471	1240	0	27.52