An improved sorting learning method for software defect prediction

Abstract

The invention relates to an improved sorting learning method for software defect prediction, which comprises the following steps of: extracting a metric element and a corresponding defect number froma source code file with a known defect number as training data; preprocessing the training data; using the preprocessed training data to construct a prediction model; using a multi-objective optimization algorithm to optimize the sorting performance of the model at the same time; obtaining a group of model parameters according to the model parameters, the regression performance and the model complexity, then selecting appropriate model parameters according to actual needs, or giving corresponding weights to the sorting performance, the regression performance and the model complexity of the model according to the actual needs, and then obtaining the model parameters by using a single-objective optimization algorithm; and analyzing the test data by using the trained model to obtain defect information of the corresponding software module. The sorting performance, the regression performance and the model complexity of the model can be optimized, and different requirements under different application scenes can be better met.

Description

technical field [0001] The invention relates to the field of software analysis and defect prediction in software engineering, in particular to an improved sorting and learning method for software defect prediction. Background technique [0002] Software defect prediction began in the 1970s. It refers to the use of statistical learning techniques to predict the number and types of defects in software systems based on software measurement data such as historical data and discovered defects. The purpose of defect prediction technology is to count and calculate the number of defects in the software system and the number of defects that have not been discovered but may still exist, so as to determine whether the system can be delivered. Defect prediction technology has promoted the improvement of software quality, and at the same time, it has also made software engineering technology develop a step forward. [0003] The invention patent "Cost-sensitive semi-supervised software d...

AI Technical Summary

Problems solved by technology

The problem with constructing a software defect prediction model for ranking tasks in this way is that the module ranking given by the model with good fitting degree is not necessarily good.

Images