Application of neural networks for software quality prediction using object-oriented metrics

Abstract

This paper presents the application of neural networks in software quality estimation using object-oriented metrics. Quality estimation includes estimating reliability as well as maintainability of a software. Reliability is typically measured as the number of defects. Maintenance effort can be measured as the number of lines changed per class. In this paper, two kinds of investigation are performed. The first on predicting the number of defects in a class and the second on predicting the number of lines change per class. Two neural network models are used, they are Ward neural network and General Regression neural network (GRNN). Objectoriented design metrics concerning inheritance related measures, complexity measures, cohesion measures, coupling measures and memory allocation measures are used as the independent variables. GRNN network model is found to predict more accurately than Ward network model.

Object-oriented software fault prediction using neural networks

“…Even though Khoshgoftaar et al [17] used Back Propagation Neural network to predict fault proneness for the software modules developed in procedural paradigm, Tong-Seng et al [24] first attempted neural network based prediction for OO environment. They applied General Regression Neural Networks to empirically validate nine Object-Oriented metrics to predict the value of fault count.…”

“…Many approaches for evaluating the quality of the prediction models are available for example, statistical significance of co-efficient [10], goodness of fit [7], cross validation [8], data splitting [24] etc.…”

“…Some of them worked on theoretical validation to prove the relationship but most of the studies were carried out on an empirical data set. Such studies are experimented by constructing prediction models for the quality attribute representatives such as fault proneness [3,6,10,12], productivity [9], effort [2], maintenance modifications [18] and fault count [24].…”

Estimating software readiness using predictive models

“…These models use characteristics in software codes such as lines of code, nesting of loops, external references, input/outputs, cyclomatic complexity and so forth to estimate the number of defects in the software [8][9][10][11][12][13][14][15]17,18,20,[24][25][26]38,39]. In two earlier studies, software development faults were predicted using object-oriented design metrics and SQL metrics [32,33].…”

“…Predictive models using neural networks and genetic net, etc. are more adept to modeling nonlinear functional relationships that are difficult to model with other techniques, and are attractive enhancements for software quality modeling [32,33]. The use of neural network model with genetic training strategy is therefore introduced to improve prediction results for estimating software readiness in this study.…”