当前位置: X-MOL 学术Comput. Sci. Rev. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Advancement from neural networks to deep learning in software effort estimation: Perspective of two decades
Computer Science Review ( IF 13.3 ) Pub Date : 2020-08-14 , DOI: 10.1016/j.cosrev.2020.100288
P. Suresh Kumar , H.S. Behera , Anisha Kumari K , Janmenjoy Nayak , Bighnaraj Naik

In the software engineering, estimation of the effort, time and cost required for the development of software projects is an important issue. It is a very difficult task for project managers to predict the cost and effort needed in the premature stages of planning. Software estimation ahead of development can reduce the risk and increase the success rate of the project. Many traditional and machine learning methods are used for software effort estimation by researchers, but always it has been a challenge to predict the effort accurately. In this study, different Artificial Neural Network (ANN) used for effort estimation is discussed. It is observed that the prediction of software effort by using ANN is more precise and better compared to traditional methods such as Function point, Use-case methods and COCOMO etc. Models based on neural networks are competitive in nature as compared to statistical and traditional regression methods. This paper explains the overview of various ANN such as basic NN, higher order NN, and deep learning networks used by the researchers for software effort estimation.



中文翻译:

从神经网络到深度学习的软件工作量估计的进步:二十年的前景

在软件工程中,估算软件项目开发所需的工作量,时间和成本是一个重要的问题。对于项目经理来说,预测计划的过早阶段所需的成本和工作量是一项非常艰巨的任务。开发前进行软件评估可以降低风险并提高项目的成功率。研究人员使用许多传统方法和机器学习方法来估算软件工作量,但准确预测工作量始终是一个挑战。在这项研究中,讨论了用于工作量估算的不同人工神经网络(ANN)。观察到,与传统方法(例如功能点,用例方法和COCOMO等)相比,使用ANN进行软件工作量的预测更为精确和更好。与统计和传统回归方法相比,基于神经网络的模型具有竞争优势。本文介绍了各种ANN的概述,例如基本NN,高阶NN和研究人员用于软件工作量估算的深度学习网络。

更新日期:2020-08-14
down
wechat
bug