Research on Cross-Company Defect Prediction Method to Improve Software Security

Table 5

Performance results of different cluster numbers for target project Ant-1.7

Source	K	Ratio	Pd	Pf	F1	AUC	K	Ratio	Pd	Pf	F1	AUC
CM1	4	2.477	0.584	0.202	0.511	0.691	7	2.543	0.578	0.197	0.511	0.691
	5	2.278	0.687	0.196	0.58	0.746	8	3.022	0.488	0.18	0.462	0.654
	6	2.307	0.675	0.196	0.573	0.74	9	3.482	0.476	0.151	0.476	0.663
MW1	4	2.701	0.639	0.164	0.578	0.737	7	2.153	0.639	0.225	0.527	0.707
	5	2.56	0.633	0.18	0.56	0.726	8	2.012	0.645	0.242	0.518	0.701
	6	0.815	0.307	0.621	0.177	0.343	9	2.012	0.62	0.249	0.499	0.686
PC1	4	2.207	0.669	0.209	0.558	0.73	7	1.522	0.458	0.379	0.33	0.539
	5	1.918	0.681	0.246	0.537	0.718	8	2.153	0.536	0.254	0.443	0.641
	6	1.725	0.681	0.277	0.515	0.702	9	2.113	0.578	0.247	0.474	0.665
PC3	4	2.221	0.669	0.208	0.559	0.731	7	2.153	0.452	0.279	0.373	0.587
	5	2.413	0.645	0.192	0.557	0.726	8	2.074	0.705	0.216	0.574	0.744
	6	2.526	0.669	0.173	0.589	0.748	9	1.988	0.723	0.223	0.578	0.75
PC4	4	2.979	0.596	0.152	0.561	0.722	7	3.376	0.398	0.18	0.393	0.609
	5	2.012	0.289	0.344	0.232	0.472	8	2.351	0.663	0.194	0.567	0.734
	6	2.895	0.434	0.206	0.403	0.614	9	2.796	0.536	0.185	0.492	0.676