Research Article
Surprise Bug Report Prediction Utilizing Optimized Integration with Imbalanced Learning Strategy
| Require: Training data (the data after balancing), the probability and of each bug, the constraint solving model . | | Ensure: The category of each bug. | (1) | = 0; //Initialization | (2) | = 0.5, = 0.5; //Initializes the weights. | (3) | /Weight training phase./ | (4) | for each | (5) | if belongs to minority class | (6) | = −1; | (7) | else | (8) | = 1; | (9) | end if | (10) | end for | (11) | | (12) | /Let the objective function be maximizing the highest achievable accuracy of the classifier./ | (13) | /Generate all the CONSTRAINTS./ | (14) | CONSTRAINTS ; | (15) | CONSTRAINTS ; | (16) | CONSTRAINTS ; | (17) | /The formulation is built successfully./ | (18) | for each classifier | (19) | ; // Obtain the most suitable weights by optimization. | (20) | end for | (21) | /Weight adjustment phase./ | (22) | for each do | (23) | ; // is the original majority probability of . | (24) | ; // is the original majority probability of . | (25) | end for | (26) | /Minimum selection/ | (27) | for each do | (28) | minimum value of majority class probabilities; | (29) | minimum value of majority class probabilities; | (30) | if | (31) | Category of majority class; | (32) | else | (33) | Category of minority class; | (34) | end if | (35) | end for | (36) | return the category of each bug report |
|