Chaos-based modified morphological genetic algorithm for effort estimation in agile software development

Abstract

One of the most critical and important aspects of any software development project is the estimation of cost and effort, as the success or failure of the entire project is largely dependent on the accuracy of these estimations. For any software development project, several methods such as waterfall, prototyping etc. exist, but the agile methods have prevailed in terms of its efficiency and implementation in solving the problems related to the projects thus substituting the traditional methodologies. Agile methods have become much popular recently because of its ability to adopt to the changing dynamics (requirements) of software projects. This dynamic nature makes the task of estimation even more challenging than the traditional methodologies present. Thus, it becomes convenient to accurately estimate the effort and cost while adopting the agile methods, for which various techniques have already been proposed such as analogy, dis-aggregation, expert opinion etc, but none among the same have a proper mathematical model. This work has presented a novice method from the domain of evolutionary algorithms. The work is based on mathematical morphology (MM) consisting of a hybrid-artificial neuron (Dilation-Erosion perceptron (DEP)) extended from the concept of complete lattice theory (CLT). Authors have presented a chaotically modified genetic algorithm (CMGA) to build the DEP-CMGA model for solving the software development effort estimation (SDEE) problem. Calibration of the proposed model was done using data collected from 21 software projects based on agile software development (ASD). Four different statistics were used for determining the precision of the model and the results were compared with the one’s obtained using the best available model in literature.