Research and Application of Software Defect Prediction Model based on Data Mining

Software quality has become an important factor affecting quality, and it is the key to ensure high reliability of the system. The concept of software engineering is highly valued in various R&D units. In the procedure of software development and operation, a large amount of historical fault data is generated and accumulated, but these historical data have not been fully researched and used. The concept of software engineering is highly valued in developers. In the procedure of software development, a large amount of historical fault data is generated and accumulated, but the data havs not been fully researched and used. Software failure prediction technology can forecast possible failures of a lot of software before testing, helping software engineers find error-prone modules and possible failure types as early as possible, and then repair them. However, through investigation, it is found that the construction of a high-performance software defect prediction model for system software faces the following problems. Firstly, modern system software fault problems are diverse and difficult to find. Secondly, the fault data is cluttered, repetitive, and incomplete. Thirdly, there is a lack of predictive models with good performance and strong model interpretability. In this paper, aiming at the fault prediction requirements of system software, the research on the construction technology of software defect prediction model is carried out.