Hybrid Maintainability Prediction using Soft Computing Techniques

Manju Duhan,Pradeep Kumar Bhatia

doi:10.47839/ijc.20.3.2280

Abstract

Effective software maintenance is a crucial factor to measure that can be achieved with the help of software metrics. In this paper, authors derived a new approach for measuring the maintainability of software based on hybrid metrics that takes advantages of both i.e. static metrics and dynamic metrics in an object-oriented environment whereas, dynamic metrics capture the run time features of object-oriented languages i.e. run time polymorphism, dynamic binding etc. which is not covered by static metrics. To achieve this, the authors proposed a model based on static and hybrid metrics to measure maintainability factor by using soft computing techniques and it is found that the proposed neuro-fuzzy model was trained well and predict adequate results with MAE 0.003 and RMSE 0.009 based on hybrid metrics. Additionally, the proposed model was validated on two test datasets and it is concluded that the proposed model performed well, based on hybrid metrics.

Highlights

SOFTWARE maintenance is the most critical activity in the software development life cycle
The main purpose of the current study was to analyze the usefulness of hybrid metrics for maintainability estimation in an object-oriented environment that takes the advantages of both i.e. static metrics and dynamic metrics that takes less time and effort to compute maintainability of software with higher accuracy
Comparison of various soft computing techniques was done with the proposed neural network and neuro-fuzzy approach based on static and hybrid metrics by collecting metrics from HoDoKu, open-source software

Summary

Introduction

SOFTWARE maintenance is the most critical activity in the software development life cycle. We have chosen these metrics based on the correlation of these metrics with external quality factor, i.e. maintainability

Objectives

Results

Conclusion