Knowledge representation using Bayesian networks — A case study in Web effort estimation

This paper's objective is twofold: i) to present a methodology employed to build company-specific effort estimation models that incorporate expert knowledge and the uncertainty inherent to this complex domain; ii) to report on a case study where one such model was completely built from expert-knowledge for a Multimedia company in Auckland (New Zealand). It contains 36 factors and 48 causal relationships, and was validated using data from 22 past finished projects. This model has to date been successfully used to estimate effort for the Web projects developed by that company.