Research Article
Research on Cross-Company Defect Prediction Method to Improve Software Security
Algorithm 2
Sample selection-based weight setting algorithm.
| Input: source and target , number of candidate samples N; | | Output: the final training dataset | | (1) | SCandidate = //array, which is used to store all the candidate training source samples | | (2) | For each sample of target project : | | (3) | Advanced = ;//Array, which is used to store the selected samples in each loop | | (4) | For each sample of source project : | | (5) | Calculate the Euclidean distance between and based on equation (2); | | (6) | Sort source samples according to the above Euclidean distance information; | | (7) | Select the Top N source samples with the smallest distance and store them in advanced; | | (8) | SCandidate SCandidate + Advanced; | | (9) | End | | (10) | End | | (11) | //array, which is used to store sample weight | | (12) | For each sample of source project : | | (13) | Statistic the sample frequency of in SCandidate and update the | | (14) | End | | (15) | Use MaxMinNormalization method to normalize ArrayOfWeight; | | (16) | Set the source sample weight based on ArrayOfWeight to obtain the final |
|
