Software quality analysis based on cost and error using fuzzy combined COCOMO model

Abstract

Software quality analysis and estimation is essential in developing a software to avoid faults and increase the reliability. Software quality model (SQM) is highly concerned with standard metrics to qualify the software modules to classify bug or no bug. By using these models, it is easy to identify the hurdles called as errors or faults Apriori to the development cycle. More likely the metrics will not follow the standard protocol in terms of size, performance, technology and the complexity involved. It will vary across the projects. Surprisingly there is no model-based architecture driven tool is available to intact the baseline estimates of the project based on the previous knowledge resource. In earlier research works, various quality assurance metrics are used for analysing the SQ. Also, there is no existing approaches can do earlier prediction of the faults/errors or reduced misclassification rates. But, the COCOMO (COnstructive COst MOdel) gives an approximate estimate in terms of the month constant will not be same for simulating the study. Hence By combining more than one model estimates COCOMO and Gaussian Membership Function software estimate relative error will be the best suite. A fuzzy-based analogy is obtained in the present study to select the nearest path from the history available to meet the project cost and time. Small or standard training sets were considered to deploy the estimate, and compare the performance with different estimators. From the experiment, it is concluded that the proposed fuzzy-COCOMO model outperforms than the existing approaches in terms of relative error.