An Improved Method for Cross-Project Defect Prediction by Simplifying Training Data

Table 4

Analyzing the factors similarity and normalization.

Factor	Method	Mean	St.d
Similarity	Cosine similarity	0.704	0.082	−0.133
	Euclidean distance	0.719	0.080	-
	Manhattan distance	0.682	0.098	−0.193
Normalization	Linear	0.706	0.087	−0.012
	Logistic	0.710	0.078	-
	Square root	0.699	0.091	−0.044
	Logarithmic	0.700	0.086	−0.064
	Inverse cotangent	0.696	0.097	−0.056