A Metrics-Based Framework for Estimating the Maintainability of Object-Oriented Software

Abstract

Software maintenance remains a resource-intensive activity, accounting for 60-70% of total software development costs, with object-oriented (OO) systems presenting unique challenges in maintainability assessment. Despite component-based OO development's prevalence, no comprehensive framework exists to measure maintainability as a unified construct during early design stages. This study addresses this gap by: (i) identifying key maintainability attributes (Understandability, Modifiability, and Reusability) through systematic analysis of OO principles; (ii) developing the Metric-Based Maintainability Estimation Framework for Object-Oriented Software (MEFOOS) using design-level metrics; and (iii) empirically validating the framework through a controlled case study of a Java-based student management system. The research employed mixed methods: a survey of 55 OO practitioners revealed current measurement shortcomings, while quantitative analysis of UML class diagrams (using Chidamber & Kemerer metrics) established statistical correlations between structural properties and maintainability attributes. Results demonstrated that class cohesion (LCOM) and coupling (CBO) were strongest predictors of modifiability (R²=0.82), while inheritance depth (DIT) significantly impacted reusability. The framework achieved 89% accuracy in maintainability classification during validation. This work contributes: (i) a novel measurement approach for early design phases, (ii) empirical evidence linking specific OO metrics to maintainability attributes, and (iii) practical guidelines for improving class designs. The framework’s evaluation confirmed its superiority, enabling early design optimization and its adoption can reduce maintenance costs by 23-41% based on case study outcomes, offering significant value to organizations adopting OO paradigms. By focusing on classes, MEFOOS provides developers a robust tool to enhance software quality, significantly impacting efficient software lifecycle management and reducing rework.

Keywords: Software maintainability, Object-oriented design metrics, Early-stage assessment, Maintainability prediction, Design quality