Simplified QMOOD Model Proposal Based on Correlation Analysis in Different Client Applications

Abstract

One of the commonly used models to measure the design quality of object-oriented software is the QMOOD model. The model works on a simple set of metrics; however, it may not be easy to obtain these metrics for different client applications even with the same/similar design. The main reason for this is that metric collection tools work differently in programming languages and development environments used to develop software for different clients. Moreover, metric aggregation tools used in some development environments may not be suitable for working with another development environment technology. In such cases, a quality metric that can be collected from the source code of a client application cannot be obtained for another client application. For this reason, it will be useful to simplify the model for some clients in order to match the quality assessment made using the QMOOD model with the real world. In this context, in this study, 9 versions of the Kubernetes project JAVA client have been examined. For these versions, all quality metrics needed in the QMOOD model were collected using different metric collection tools and the change of design quality attributes was obtained. In the evaluation, it was determined that the change in design quality characteristics coincided with the real world. From this point of view, the binary cross correlation of design properties was calculated. As a consequence, a QMOOD model proposal with 27% simplification that can be used in quality measurement for different client applications of the Kubernetes project is included.