Software Defect Prediction through Neural Network and Feature Selections

Table 20

Comparison of RBF results with other previous methods after feature selection.

Data set	Source	Algorithm	F-measure	Accuracy
CM1	[43]	J star	N/A	81.69%
		RF	N/A	84.60%
	[49]	MLP	N/A	89.79%
	[50]	Boost-RF	N/A	89.97%
		Bag-RF-FS	N/A	89.97%
	[51]	DT	0.94	89.92%
		MLP	0.94	89.32%
		SVM	0.95	90.69%
	This research	RBF	0.95	93.99
JM1	[43]	J star	N/A	81.90%
		RF	N/A	73.90%
	[49]	MLP	0.17	80.44%
	[50]	Boost-RF	N/A	80.56%
		Bag-RF-FS	N/A	80.61%
	This research	RBF	0.87	84.87
KC1	[42]	J48	N/A	68.78%
		RF	N/A	71.08
	[43]	RF	N/A	84.16%
		J star	N/A	85.97%
	[50]	Boost-RF	N/A	78.51%
		Bag-RF-FS	N/A	78.51%
	[49]	MLP	0.43	77.65%
	[51]	DT	0.91	84.56%
		MLP	0.92	85.68%
		SVM	0.91	84.59%
	This research	RBF	0.83	83.25%
KC2	[52]	LSTSVM	N/A	86.12%
	This research	RBF	0.82	79.11%
KC3	[43]	J star	N/A	80.50%
		RF	N/A	80.50%
	[50]	Boost-RF	N/A	79.13%
		Bag-RF-FS	N/A	77.58%
	[49]	MLP	0.28	82.75%
	[51]	DT	0.96	93.11%
		MLP	0.96	93.50%
		SVM	0.96	93.28%
	This research	RBF	0.85	78.25%
KC4	This research	RBF	0.86	83.18%
MC1	[50]	Boost-RF	N/A	97.61%
		Bag-RF-FS	N/A	97.61%
	[49]	MLP	N/A	97.61%
	This research	RBF	0.99	99.01%
MC2	[50]	Boost-RF	N/A	64.86%
		Bag-RF-FS	N/A	62.16%
	[49]	MLP	N/A	97.60%
	This research	RBF	0.76	70.18
MW1	[50]	Boost-RF	N/A	89.33%
		Bag-RF-FS	N/A	89.33%
	[49]	MLP	0.40	92.00%
	This research	RBF	0.95	88.90%
PC1	[43]	J star	N/A	91.17%
		RF	N/A	91.17%
	[50]	Boost-RF	N/A	96.07%
		Bag-RF-FS	N/A	96.07%
	[49]	MLP	0.42	96.56%
	[51]	DT	0.96	93.58%
		MLP	0.96	93.59%
		SVM	0.96	93.10%
	This research	RBF	0.99	98.99%
PC2	[50]	Boost-RF	N/A	97.23%
		Bag-RF-FS	N/A	97.69%
	[49]	MLP	N/A	97.69%
	This research	RBF	0.99	99.80%
PC3	[50]	Boost-RF	N/A	87.34%
		Bag-RF-FS	N/A	87.34%
	[49]	MLP	0.14	85.12%
	This research	RBF	0.97	94.11
PC4	[50]	Boost-RF	N/A	91.60%
		Bag-RF-FS	N/A	90.81%
	[49]	MLP	0.44	88.97%
	This research	RBF	0.95	94.44
PC5	[50]	Boost-RF	N/A	75.78%
		Bag-RF-FS	N/A	76.96%
	[49]	MLP	0.24	74.80%
	This research	RBF	0.80	79.02%