Cuadrado: On the use of fuzzy regression in parametric software estimation models: Integrating imprecision in COCOMO cost drivers

Abstract:- Parametric software estimation techniques make use of mathematical models elaborated from regression techniques to obtain effort-of-development estimates. As a matter of fact, the values of the cost drivers that act as variables in these models are inherently vague and uncertain, since they’re in many cases elaborated from human judgments about highly abstract concepts (like ‘required reliability’) or even from collective attitudinal characterizations (like ‘analyst capability’). In this paper, fuzzy regression techniques based on fuzzification of input values are explored as an alternative to conventional regression for obtaining estimating equations. The available COCOMO-81 project database and the Furea fuzzy regression tool are used as a case study, emphasizing more realistic approaches to the expression of widely used cost driver values. Key-Words:- Software effort estimation, fuzzy regression, parametric estimation models, fuzzy variables