Tuning of software cost drivers using BAT algorithm
Sanchi Girotra, Kapil Sharma · International Conference on Computing for Sustainable Global Development · 2016
Software effort estimation is an important process in software development as it predicts the no. of resources required to build the project. Lack of its accuracy and precision will impact timely delivery of project and its budget. There are numerous methods to estimate software effort. COCOMO II has been employed because of its wide acceptance as an industry standard and its applicability at diverse stages of software Engineering. And Accuracy in COCOMO II software effort estimation is highly dependent on its input parameters such as size of project, coefficients and cost drivers. Thus small changes in these parameters bring huge differences in Effort estimation. In this paper, we used Nature Inspired — Meta Heuristic Bat Algorithm to fine tune values of 15 cost drivers, which can effectively reduce error (MMRE) in effort estimation. It has been validated using NASA 93 dataset and its results are found better than COCOMO and genetic algorithm.