Application of Machine Learning Techniques for Software Reliability Prediction (SRP)

Background

In literature, many empirical studies based on multivariate linear regression and neural network methods have been carried out for prediction of software reliability growth trends. Although, multivariate linear regression method can address linear relationship but require large sample size and more independent variables (Jung-Hua 2010). The use of support vector machine (SVM) approach in place of classical techniques has shown a remarkable improvement in the prediction of software reliability in the recent years (Xiang-Li 2007). The design of SVM is based on the extraction of a subset of the training data that serves as support vectors and therefore represents a stable characteristic of the data. SVM can be applied as an alternative approach because of their generalization performance, ease of usability and rigorous theoretical foundations that practically can be used for regression solving problems (Ping and Lyu 2005). However, the major limitation of support vector machines is the increasing computational and storage requirements with respect to the number of training examples (Chen 2005). The group method of data handling (GMDH) network based on the principle of heuristic self-organization has also been applied for predicting future software failure occurrence time and the optimal cost for software release instant during the testing phase (Dohi et al., 2000). They numerically illustrated that GMDH networks are capable to overcome the problem of determining a suitable network size in multilayer perceptron neural network. GMDH also can provide a more accurate measure in the software reliability assessment than other classical prediction methods. Another reliability prediction method, neural network based approach for predicting software reliability using Back-Propagation Neural Network (BPN) has been applied for estimating the failures of the software system in the maintaining phase (Chen et al, 2009).