Researching Algorithms Used to Predict Defects in the Software Code of Telecommunications Equipment

This article examines the problem of predicting defects in telecommunications equipment software. The goal of this work is to identify the most effective machine learning (ML) algorithm for accurately predicting defects in the source code of telecommunications equipment. The relevance of this work lies in the fact that traditional statistical algorithms and models do not provide a sufficiently accurate forecast of the quality of software used in various telecommunications equipment (including in space in Earth orbit), and static code analysis tools produce a large number of false positives and false negatives. Furthermore, given the high demands placed on the operation of telecommunications equipment in space in Earth orbit, it is especially important to identify defects in software code at all stages of its development. Therefore, the need for machine learning algorithms to predict code errors is increasing. The object of this study is the growing demand for high-quality software. The subject of this study is metrics for assessing the quality of machine learning algorithms and models designed to predict defects in source code. As a result of the experiments, conclusions were drawn about the effectiveness of each algorithm, and the choice of the most suitable algorithm for solving a specific applied problem of telecommunications equipment was determined.