Defect Prediction Technology of Aerospace Software Based on Deep Neural Network and Process Measurement

Table 3

Experimental data set.

Software project	Version number	Number of modules	Number of metrics	Number of defective modules	Defect rate (%)
XX flight control software	V1.01	1837	31	207	11.26
	V1.02	1995	31	215	10.77
	V1.03	2448	31	308	12.58
XX fire control software	V1.01	5973	31	1218	20.39
	V1.02	6898	31	1048	15.19
	V1.03	8420	31	1406	16.70
XX accused software	V1.01	3086	31	484	15.68
	V1.02	3317	31	723	21.79
	V1.03	3965	31	511	12.89