Application of Fuzzy Logic Approach to Software Effort Estimation

Abstract

The most significant activity in software projectmanagement is Software development effort prediction.The literature shows several algorithmic costestimation models such as Boehm{'}s COCOMO, Albrecht's'Function Point Analysis, Putnam{'}s SLIM, ESTIMACSetc., but each model do have their own pros and cons inestimating development cost and effort. This is becauseproject data, available in the initial stages ofproject is often incomplete, inconsistent, uncertainand unclear. The need for accurate effort prediction insoftware project management is an ongoing challenge. Afuzzy model is more apt when the systems are notsuitable for analysis by conventional approach or whenthe available data is uncertain, inaccurate or vague.Fuzzy logic is a convenient way to map an input spaceto an output space. Fuzzy Logic is based on fuzzy settheory. A fuzzy set is a set without a crisp, clearlydefined boundary. It is characterized by a membershipfunction, which associates with each point in the fuzzyset a real number in the interval [0, 1], called degreeor grade of membership. The membership functions may beTriangular, GBell, Gauss and Trapezoidal etc. In thepresent paper, software development effort predictionusing Fuzzy Triangular Membership Function and GBellMembership Function is implemented and compared withCOCOMO. A case study based on the NASA93 datasetcompares the proposed fuzzy model with the IntermediateCOCOMO. The results were analyzed using differentcriterions like VAF, MARE, VARE, MMRE, Prediction andMean BRE. It is observed that the Fuzzy Logic Modelusing Triangular Membership Function provided betterresults than the other models.