Optimized Decision Tree-based Early Phase Software Dependability Analysis in Uncertain Environment

Because of rapid development of software-based technology, early phase software dependability analysis has become very essential. The main purpose of this paper is to develop a novel classification model based on software dependability attributes for classification of software fault-prone modules during early phase of development. To carry out dependability analysis, early phase dependability attribute prediction is very essential. During early phase software metric value collection, metric values become uncertain due to difficulty of collection. Fuzzy Inference System (FIS) has been designed here to predict dependability attributes. A new algorithm for rule base generation has been proposed initially based on expert opinions for fuzzy inference system. These predicted software dependability attributes are applied to carry out dependability analysis. To classify software modules in early phase of development, an optimized Decision tree-based classification algorithm has been proposed to meet the main goal. Finally, Mahalanobis distance-based ranking technique has been developed based on different dependability attributes for ranking most non-dependable software modules. The proposed model has been validated based on software data sets. Software engineers will find the proposed model useful in allocating testing resources for most non-dependable software modules.