Study on software defects prediction model based on machine learning

In the rapid development of network science and technology, the software, as the basic part of the network system operation, the practical application quality directly determines the realization of the function, so the users put forward higher requirements for the software quality. According to the application situation of network system software in recent years, software defects are the main factor affecting the application quality, and the relevant detection technology is the only way before the formal promotion of software. Therefore, researchers have put forward a defect prediction scheme based on the software code, which can not only reduce the cost, but also improve the practical efficiency. This paper focuses on the understanding of the machine learning algorithm and constructing automatic and comprehensive learning models according to the software defect prediction technology, thus discovering the defects in the software. The final experimental results prove that different algorithms have different advantages in different evaluation indicators. By using these advantages and the stacking integrated learning methods in machine learning, building a prediction model with combined machine learning algorithms as the core can find defects more accurately and perfectly.