Fuzzy Based Approach for Object-Oriented (OO) Software Maintainability Measurement

摘要

The quality of the software is the very important factor in the field of software development which can be determined by many quality attributes of the software. Thus, the quantification of the quality parameters and incorporating them into the quality models are essential for software maintainability. The ISO/IEC 25010:2011 standard is developed to integrate the quality model based on software attributes. In this study, a fuzzy based model is proposed to predict the software maintainability from UML class diagram. The outcomes are presented and the knowledge modeling using fuzzy logic is discussed. The development of this model is based on the factors that affect the maintainability like other software quality factors. This hierarchy consists of factors, attributes and metrics which are used for measuring the maintainability of object-oriented software. This proposed model captures the factors that determine maintainability at design level and expressed by coupling and size attributes. Some of the metrics which quantify these attributes (NA, NM, NAssoc, NAgg etc.) are then considered as the input parameters to the proposed model. The process is applied to a case study where the Mamdani fuzzy inference engine is used to predict software maintainability. The performance of the model was evaluated using RMSE. In order to estimate the software maintainability, the analysis of different membership functions defined in fuzzy inference system by MATLAB for the mentioned metrics are presented.