Cross-Project Defect Prediction Based on Two-Phase Feature Importance Amplification

Algorithm 1

CFS.

(i)	Input: , Feature set with features and a class;
(ii)	Output: , Subset of selected features,
1	, , ,.,
2	compute the merit of , ,., by (3)
3	put the feature with the biggest merit of into , and remove it from
4	compute merit of
5	put the new feature with the biggest merit of into , compute ,
6	if then
7	remove the new feature from
8	else
9	Update , remove the new feature from ,
10	end if
11	repeat step 5 to step 7 until or has no change by a certain number of times
12	return