Software defect prediction method and system

Abstract

The invention discloses a software defect prediction method. According to the method, a sample with a class label and a sample without a class label are processed together, semi-supervised learning is utilized in the laplacian eigenmaps (LE), the LE method is improved, and meanwhile, in order to avoid that samples different in classification are mapped to a small low-dimensional neighbourhood, especially that defect samples are mapped to a defect-free sample neighbourhood, cost-sensitive information is introduced when the sample point distance is calculated through an LE algorithm. On the basis, the LE mapping precision is improved, and the discriminating performance of discriminating performance can be effectively improved through the method. The invention further provides a software defect prediction system. When the software defect prediction method and system are applied to an NASA database, it is proved through experiments that the effectiveness of the method is improved, and compared with other comparative methods, the classification performance of the method is improved to a certain extent.

Description

technical field [0001] The invention relates to a software defect prediction method and a prediction system, belonging to the field of software engineering. Background technique [0002] Software defect prediction includes four links: data preprocessing, feature extraction, training prediction model, and identification. Feature extraction is one of the most basic problems in software defect prediction. For software defect prediction, extracting effective features is the primary task of recognition. [0003] Existing feature extraction methods can be divided into traditional dimensionality reduction methods and popular learning dimensionality reduction methods. Among them, the traditional dimensionality reduction methods include Principal Component Analysis (PCA) and Multidimensional Scaling Analysis (MDS). Popular learning dimension reduction methods: including isometric mapping (ISOMAP), Laplacian feature mapping (LE), locality-preserving projection (LPP), etc. [0004]...

AI Technical Summary

Problems solved by technology

[0024] For the traditional dimensionality reduction method, there are some deficiencies, such as Principal Component Analysis (PCA) method, its theory is perfect, and the calculation is effective, and it has a good dimensionality reduction effect on the data set whose internal structure is linear structure, but in the surface For linearly inseparable data, PCA cannot reflect the nonlinear nature of the data; and the multidimensional scaling algorithm (MDS) cannot handle sample data with nonlinear structures well.

For the popular learning dimension reduction method, although the isometric mapping method (ISOMAP) can reflect the intrinsic properties of nonlinear data, it is similar to MDS and is based on a global dimensionality reduction algorithm. This method does not take into account the local relationship of data samples. , and during dimensionality reduction, ISOMAP may produce an error of the "Elbow" phenomenon; although Laplacian eigenmap (LE) and local preservation projection (LPP) can process the local information of the sample, they do not take into account the category information of the sample

Images