ISSN 1991-2927
 

ACP № 3 (61) 2020

Author: "Vladimir Aleksandrovich Belov"

Pavel Vladimirovich Dudarin, graduated from Ulyanovsk State University; Postgraduate Student at the Department of Information Systems of Ulyanovsk State Technical University; an author of scientific papers in the field of data processing by means of neural networks. e-mail: p.dudarin@ulstu.ru.P. Dudarin,

Vadim Georgievich Tronin, Candidate of Science in Engineering; Associate Professor at the Department of Information Systems of UlSTU; an author of scientific papers in the field of economics, finance and information technologies. e-mail: v.tronin@ulstu.ru.V. Tronin,

Kirill Valerevich Sviatov, Candidate of Science in Engineering; graduated from UlSTU; Dean of the Faculty Information Systems and Technologies of UlSTU; an author of scientific papers in the field of automation of management processes. e-mail: k.svyatov@ulstu.ru.K. Sviatov,

Vladimir Aleksandrovich Belov, graduated from UlSTU with a bachelor degree in Software Engineering; Master Student at the Department of Information Systems of UlSTU; an author of an article in the field of computer operation monitoring. e-mail: v.belov@ulstu.ru.V. Belov,

Roman Azatovich Shakurov, graduated from UlSTU with a bachelor degree in Software Engineering; Master Student at the Department of Information Systems of UlSTU; an author of articles in the field of computer operation monitoring and developing of system for determining the winner in cyber security competitions. e-mail: r.shakurov@ulstu.ru.R. Shakurov

An Approach To Labor Intensity Evaluation In Software Development Process Based On Neural Networks 57_8.pdf

Software development process is actively studied by experts from different spheres of science and different viewpoints. However, the degree of success of projects in the development of software intensive systems (Software Intensive Systems, SIS) has changed insignificantly, remaining at the level of 50% inconsistency with the initial requirements (finance, time and functionality) for medium-sized projects. The annual financial losses in the world because of the total failures are counted by hundreds of billions of dollars. Its high complexity leads to constant mistakes in labor intensity evaluation, and even new agile development paradigm does not solve this problem. This paper shows that retrospective labor intensity estimation could be approximated by a function, implemented by neural network, with some amount of code complexity metrics as arguments. Also there is a described an approach of neural network training and data collection, which allows to automate a process of retrospective labor intensity evaluation in agile software developing process. Experiments performed on the real life software project show the effectiveness of proposed technique.

Software developing process, neural network, data augmentation, Halstead metrics, Cyclomatic metric.

2019_ 3

Sections: Artificial intelligence

Subjects: Artificial intelligence.



© FRPC JSC 'RPA 'Mars', 2009-2018 The web-site runs on Joomla!