Bayesian Software Failure Probability based Time-Invariant Tanimoto Random Testing for Software Quality Management

P. Saravanan and Dr.V. Sangeetha

The prediction of software defect is a considerable issue to be resolved for effective software quality management. For improving the software quality, recently, few research works have been designed with different techniques. However, software failure cause prediction accuracy (FCPA) and defect detection rate (DDR) of existing techniques were not adequate. In order to overcome such limitations, Bayesian Software Failure Probability Based Time-Invariant Tanimoto Random Testing (BSFP-TITRT) technique is proposed. The BSFP-TITRT technique estimates the probability of software failure prediction and identify where the defect density (DD) peaks. The BSFP-TITRT technique is introduced with the application of Bayesian Linear Regression Analysis (BLRA) and Time-Invariant Tanimoto Random Testing (TITRT). The designed BSFP-TITRT technique initially gets software program codes and event log files as input. After taking input, BSFP-TITRT technique utilizes BLRA to determine probability of software failure and discovering root cause of failure with higher accuracy. Based on evaluated software failure probability, finally BSFP-TITRT technique employs TITRT that correctly identifies errors, gaps or missing requirements in contrary to actual user needs with minimal time. From that, BSFP-TITRT technique achieves better software management performance. BSFP-TITRT technique conducts experimental process with metrics namely FCPA, defect detection rate (DDR), defect detection time (DDT) and false-positive rate (FPR). Experimental result evident that BSFP-TITRT technique enhances the DDR and lessen the DDT as compared to conventional works.

Volume 12 | 01-Special Issue

Pages: 473-485

DOI: 10.5373/JARDCS/V12SP1/20201094