User stories complexity estimation using Bayesian networks for inexperienced developers

Planning Poker is a complexity estimation technique for user stories through cards. This technique offers many advantages; however, it is not efficient enough as estimations are based on experts criteria, which is fuzzy regarding what factors are considered for estimation. This paper proposes a knowledge model to determine two of the most important aspects of estimation, the complexity, and importance of user stories based on Planning Poker in Scrum context. The goal of this work is to model the complex nature of user story estimation to facilitate this task to novice developers. A Bayesian network was built based on the proposed model that considers the complexity and importance of a user story. Students and professionals submitted their estimates to correlation tests to validate the applicability of the proposed model. Based on the results, the proposed model achieves a greater degree of correlation with the estimation from professionals than students, which means that the model includes factors considered in real world application. This proposal could be useful for guiding novice developers to evaluate the complexity and importance of user stories through questions. Students could use the proposal to estimate rather than the traditional Planning Poker.