Software Maintainability Prediction Model Based on Fuzzy Neural Network
Lixin Jia, Bo Yang, Dong Ho Park, Feng Tan and Minjae Park
Due to the vast deployment of object-oriented software in our day-to-day livings, the issue of software maintainability prediction, which aims at ameliorating the software design process and planning the amount-constrained budget efficiently, calls attention to it. In this paper, a Fuzzy Neural Network (FNN) based software maintainability prediction model, which combines the Artificial Neural Network (ANN) and the Fuzzy Logic (FL), is proposed. To overcome the innate flaws of FNN, a statistical technique, e.g. Principle Component Analysis (PCA), is also used for the sake of computational simplicity. The proposed FNN, reinforced by PCA, can effectively reflect the complex relations among independent and dependable variables, that is, by showing relatively high prediction accuracy. The empirical experimental results verify this claim in the sense that, with respect to two disparate object-oriented software data sets, the model built by the proposed method prevails against three other typical counterparts, Multivariable Linear Regression (MLR), ANN and Support Vector Regression (SVR), in terms of the prediction accuracy.
Keywords: Object-oriented software, software maintainability prediction, fuzzy neural network, principle component analysis