期刊名称：Software Engineering : an International Journal
电子版ISSN：2249-9342
出版年度：2012
卷号：2
期号：2
页码：19-36
出版社：Delhi Technological Universiity
摘要：Software maintainability is one of the most important aspects while evaluating quality of the software product. It is defined as the ease with which a software system or component can be modified to correct faults, improve performance or other attributes or adapt to a changed environment. Tracking the maintenance behaviour of the software product is very complex. This is precisely the reason that predicting the cost and risk associated with maintenance after delivery is extremely difficult which is widely acknowledged by the researchers and practitioners. In an attempt to address this issue quantitatively, the main purpose of this paper is to propose use of few machine learning algorithms with an objective to predict software maintainability and evaluate them. The proposed models are Group Method of Data Handling (GMDH), Genetic Algorithms (GA) and Probabilistic Neural Network (PNN) with Gaussian activation function. The prediction model is constructed using the above said machine learning techniques. In order to study and evaluate its performance, two commercial datasets UIMS (User Interface Management System) and QUES (Quality Evaluation System) are used. The code for these two systems was written in Classical Ada. The UIMS contains 39 classes and QUES datasets contains 71 classes. To measure the maintainability, number of “CHANGE” is observed over a period of three years. We can define CHANGE as the number of lines of code which were added, deleted or modified during a three year maintenance period. After conducting empirical study, performance of these three proposed machine learning algorithms was compared with prevailing models such as GRNN (General Regression Neural Network) Model, ANN (Artificial Neural Network) Model, Bayesian Model, RT (Regression Tree) Model, Backward Elimination Model, Stepwise Selection Model, MARS (Multiple Adaptive Regression Splines) Model, TreeNets Model, GN (Generalized Regression) Model, ANFIS (Adaptive Neuro Fuzzy inference System) Model, SVM (Support Vector Machine) Model and MLR (Multiple Linear Regressions) Model which were taken from the literature. Based on experiments conducted, it was found that GMDH can be applied as a sound alternative to the existing techniques used for software maintainability prediction since it assists in predicting the maintainability more accurately and precisely than prevailing models.
关键词：GMDH (Group Method of Data Handling);Genetic Algorithms; Probabilistic Neural Network (PNN);Software Maintainability; Software Maintainability;Prediction Metrics and Modeling.
