Software defect prediction-oriented unbalanced data generation method

Abstract

The invention discloses an unbalanced data generation method for software defect prediction, and belongs to the field of software testing. A software defect data set has a serious data imbalance problem, and generates a negative effect on the performance of a prediction model. Common methods for solving the data imbalance problem achieve inter-class balance by adjusting the number of samples, butnew sample distribution generally follows original distribution, and intra-class balance is not improved. According to the method, the sample distribution condition of the data set is considered, clustering division is performed on the original data set, and defective sample data generation is performed on the divided subareas by adopting different strategies according to different distribution conditions, so inter-class balance and intra-class balance of the data set samples are realized. Data imbalance can be effectively improved through data generation based on distribution, and accuracy ofa software defect prediction model is remarkably improved.

Description

technical field [0001] The invention relates to a method for generating unbalanced data oriented to software defect prediction, which belongs to the field of software development and testing. Background technique [0002] Software defect prediction technology (software defect prediction) identifies defective software modules by analyzing software historical data and using models such as classification and sorting. In the actual software defect prediction data set, the number of defective samples is usually much smaller than the number of non-defective samples, that is, there is an imbalance between classes. In addition, the distribution of defective samples in the data set is often uneven, which belongs to intra-class imbalance. Both inter-class imbalance and intra-class imbalance will degrade the performance of the prediction model on the prediction of defective samples. [0003] The existing data generation methods dealing with the problem of data imbalance generate new s...

AI Technical Summary

Problems solved by technology

In the actual software defect prediction data set, the number of defective samples is usually much smaller than the number of non-defective samples, that is, there is an imbalance between classes. In addition, the distribution of defective samples in the data set is often uneven, which belongs to intra-class imbalance.

Both inter-class imbalance and intra-class imbalance will reduce the performance of the prediction model on the prediction of defective samples.

[0003] Existing data generation methods that deal with data imbalances generate new samples that are close to the original distribution. Although the inter-class balance is achieved by increasing the number of defective samples, the intra-class imbalance problem is not resolved.

Images