Improved Particle Swarm Optimization Algorithm For Software Defect Prediction Model Of Support Vector Machine

With the continuous development of the new era and the rapid progress of science and technology,software has played an indispensable role in society.At the same time,software systems are becoming increasingly complex and large-scale,and the existence of software defects will bring unpredictable losses and problems to people's lives,which makes software quality assurance increasingly important.Software defect prediction technology is currently one of the most effective means to improve software quality.The combination of swarm intelligence algorithm and kernel function technology is an important research direction in software defect prediction.The main purpose of software defect prediction is to measure the possibility of defects in software modules by using related technologies.Aiming at this problem,the support vector machine,which has excellent performance in the binary classification problem,is used as the basic classifier,and the swarm intelligence optimization algorithm is used to optimize its parameters,so as to achieve the purpose of software defect prediction.The main work includes:(1)In view of the problem that the traditional particle swarm optimization algorithm is difficult to obtain the global optimum,an improvement is made.First,the fixed inertia weight value is improved to make it change linearly according to the particle iterative process;at the same time,A new learning factor change formula is proposed,which has a certain relationship with the inertia weight,and continuously adapts to the particles in the iteration;finally,the flight time factor is introduced to make the position of the particle adapt to the change and update iteratively.By using the improved particle swarm algorithm to optimize the parameters required in the support vector machine,the optimal parameters of the model are finally obtained,which makes it have a good performance in software defect prediction.(2)Although the traditional single-core support vector machine has good generalization ability,its adaptability and robustness are relatively poor.Therefore,the radial basis kernel function and the polynomial function are linearly combined to form a multi-core kernel function,which constitutes the support of multi-core.Vector machines to predict software defects for better performance.(3)From the perspective of practical application,based on the two software defect prediction models proposed in this paper,and then combining the existing software development techniques and tools,a software defect prediction system is designed and implemented.