Title: Evaluation of mean and variance approximations in three point estimation of task completion times using the beta and the Kumaraswamy distribution
Authors: Pablo Ortiz Bochard; Thomas Schwarz
Addresses: Área de Mecatrónica, Universidad Tecnológica del Uruguay, 11500 Montevideo, Uruguay ' Department of Mathematics, Statistics and Computer Science, Marquette University, Milwaukee, Wisconsin, USA
Abstract: Estimation of task and project completion times within IT projects remains one of the most error-prone, but also most critical duties of an IT project manager. Various three-point methods of PERT have been evaluated by assuming that the true distribution is a beta-distribution. We evaluate PERT methods by comparing additionally with the Kumaraswamy distribution, which has an equal claim to be the true a-priori distribution for project completion times. We use skew and kurtosis in order to define test sets instead of simply choosing a range of shape parameters. We validate various approximations proposed in the literature and show that valid approximations are possible.
Keywords: project management; expert judgement; mean and variance approximations; PERT; three-point estimations; task completion times; beta distribution; Kumaraswamy distribution.
DOI: 10.1504/IJITM.2019.103053
International Journal of Information Technology and Management, 2019 Vol.18 No.4, pp.389 - 406
Accepted: 20 May 2017
Published online: 15 Oct 2019 *