Loading [a11y]/accessibility-menu.js
Predicting Software Maintenance Type, Change Impact, and Maintenance Time Using Machine Learning Algorithms | IEEE Conference Publication | IEEE Xplore

Predicting Software Maintenance Type, Change Impact, and Maintenance Time Using Machine Learning Algorithms


Abstract:

Software change requests can arise at any time and changes that are accepted by the change approval board goes to maintenance. Change impact analysis and implementation r...Show More

Abstract:

Software change requests can arise at any time and changes that are accepted by the change approval board goes to maintenance. Change impact analysis and implementation requires high cost because modules and artifacts in previous versions may not be the same as current versions. Changes usually cause impacts on other modules and artifacts; the time required to implement changes varied and becomes high. The type of maintenance, the impact of changes, and maintenance time can be determined by analyzing software repository data. But the selection of the important software repositories and extracting the relevant information from software repositories is challenging. Some studies were conducted on maintenance type prediction and change impact analysis but these studies focus on single maintenance tasks and specific software types, so the generality of the studies is in question mark both in maintenance task and type of software. In addition to this limited amount of data and only version history data is used. In this study, a method for predicting maintenance type, impacts of changes, and maintenance time is proposed so that relevant maintenance information is extracted from software repositories. Linear Support Vector Classifier, Random Forest, Logistic Regression, Multinomial Naïve Bayes, LSTM, and Bi-LSTM are used to predict maintenance tasks and their change impact. An Artificial Neural Network (ANN) is applied to estimate the time needed to resolve the changes. The experiments results show that Random Forest and LSTM have an accuracy of 94% and 95% respectively, and have better results compared with other models. ANN provides a Mean Square Error of 0.0028. Based on Pearson and Spearman correlation analysis, maintenance type and maintenance time show a positive correlation while change impact shows a negative correlation both with maintenance type and maintenance time.
Date of Conference: 28-30 November 2022
Date Added to IEEE Xplore: 08 December 2022
ISBN Information:
Conference Location: Bahir Dar, Ethiopia

References

References is not available for this document.