Characterizing software development method using metrics. Issue 2 (16th December 2015)

• Record Type: Journal Article
• Authors: Shawky, Doaa M.; Abd‐El‐Hafiz, Salwa K.
• Is Part Of: Journal of software. Volume 28:Issue 2(2016)
• Page Start: 82
• Page End: 96
• Publication Date: 2015-12-16
• Subjects: agile development -- classification -- feature selection -- software metrics -- ANN; Software engineering -- Periodicals; Computer software -- Development -- Periodicals; Software maintenance -- Periodicals; 005.1
• DOI: 10.1002/smr.1766
• Languages: English
• ISSNs: 2047-7473

Abstract

Abstract: This work investigates whether the development methodology affects some metrics that are related to the quality of the generated code. Trying to characterize the development methodology using software metrics could be beneficial to developers who use agile approaches extensively. Therefore, this study sheds light on the metrics that might be highly affected by following agile approaches. The values of these metrics need to be tracked, because they might result in a quality‐related issue, such as error‐proneness, for example. The proposed approach employs different feature selection and classification methods, which include artificial neural networks. We used 11 object‐oriented systems of various sizes and functionalities. Five of the used systems were developed using agile approaches, while the rest were developed using non‐agile methodologies. The classes and methods of the used systems were represented by a set of metrics that shows some complexity, cohesion, and coupling features of the systems. The generated metrics are used as features to classify between the two different types of the development methodologies using a feed‐forward artificial neural network. The obtained high classification accuracy shows the strong relationship between the used metrics and the type of the followed development process. Copyright © 2015 John Wiley & Sons, Ltd. Abstract : This study employs different feature selection and classification methods. The classes and methods of the used, agile and non‐agile, systems were represented by a set of metrics that shows some complexity, cohesion, and coupling features. The generated metrics are used as features to classify between the two different types of the development methodologies using a feed‐forward artificial neural network. The obtained high classification accuracy shows the strong relationship between the used metrics and the type of the followed development process.