OPTIMIZING EFFORT ESTIMATION MODEL USING BAT ALGORITHM

Abstract

Software development is most difficult as compared to other types of projects as there is uncertainty in the customer requirements, the process of development is complex, and the final product is intangible in nature. As per the IBM report, “31%of the project gets cancelled before they are completed, 53% overrun their cost estimates by an average of 189% and for every 100 projects, and there are 94 restarts”. In order to increase the likelihood of success of a software project, the project managers must do project planning well and for that they need proper effort estimation of the project. Software effort estimation is amongst the most important tasks in software project management as many decisions like cost estimation, deadline of submitting of project and timely planning a project are dependent on it. Many Algorithmic models are used for effort estimation like COCOMO, Function Points, Use case points etc. Most widely used software cost model or effort model is the Constructive Cost Model (COCOMO). This thesis introduces a new calibrated Intermediate COCOMO model (for all types of system i.e. organic, semi-detached and embedded) with Bat Algorithm, which is newest Algorithm amongst the category of Meta Heuristic and population based Algorithms. For estimation we have used NASA 63 dataset and Results show that Bat Algorithm gives better results in terms of MMRE (Mean Magnitude of Relative Error) for projects as compared to COCOMO Model.