SELECTION OF OPTIMAL SOFTWARE COST ESTIMATION MODEL USING BACTERIAL FORAGING OPTIMIZATION ALGORITHM

Abstract

The main challenge that the software industry faces today is to estimate the cost required to develop the project in the early phase of software development life cycle. Cost estimation is difficult in the early phase because cost depends on factors like Line of Code, Methodology adopted etc. which cannot be stated accurately in the beginning. Various techniques had already been developed by the researchers which can be categorized into Algorithmic and Non- Algorithmic methods to determine the cost of the software precisely. In this thesis, Bacterial Foraging Optimization Algorithm (BFOA) technique is operated on NASA data set to estimate the cost for the NASA project. BFOA is a one of the soft computing techniques which is tolerant to factors like imprecision, approximation, partial truth, and uncertainty. It has drawn attention of many of the researchers and has been used in various engineering application. BFOA is employed to generate parameters of the COCOMO model and four of its variants. These five models were compared against the comparison criteria like BIAS, MSE, MAE, MEOP, PRR, Variance, RMPSE, RSQ, SSE, TS, ED, PA, SD, MD, MMRE, RMSE and, NRMS. Finally distance based approach (DBA) is used for optimal selection and ranking of COCOMO models. DBA do recognizes the importance of relative importance of the criteria for the given application, without it inter-criterion comparison could not have been accomplished. What all it requires is set of criteria for selection of model, set of model, and their level for the purpose of optimal selection; and finally it successfully rank on position one COCOMO_model1 as the best one with LSE of 651.2720.