Performance analysis of agile methodologies for software quality improvement

Abstract

Quality estimation is a pertinent step in the software development process. Software quality estimation is not a single variable analysis but depends upon number of factors that directly or indirectly govern the quality of a software project. Hence, this research work views quality as a construct to be decided by several factors which are related to Object-Oriented Programming Systems. In the present research, the quality of software is predicted by integrating soft computing approaches with agile technologies. The proposed algorithm utilizes agile based algorithmic structure of SWARA, FCRI and DDA methods for the attribute selection based on three formed classes in the list and has been evaluated under two statistical parameters namely Mean Squared Error and Standard Error. The clustering step is also modified using Cosine Similarity integrated into the basic architecture of K-means followed by the Feed Forward Back Propagation Neural Network based architecture. Aim is to P newlineenhance quality of the system based on class separations. In order to do so, the proposed algorithm architecture uses Swarm Intelligence based Grasshopper Algorithm for selection of most suitable records in the project database since more precise selection among the data row values will lead to precise quality. To justify that the proposed work outperforms the existing methods for prediction of software quality, comparative analysis of the proposed work is performed against the agile technology inspired techniques. The overall effect on predictive analysis of the techniques is evaluated in terms of precision, recall, f-measure and accuracy of software quality estimation when using 70%, 80% and 90% of the data for training of the system. Performance evaluation has also been done by implementing K-means with the grasshopper standard algorithm for comparative analysis. newline newline