Better Estimation of PERT Activity Time Parameters

Abstract

This paper builds upon earlier work from the decision/risk analysis area in presenting simple, easy-to-use approximations for the mean and variance of PERT activity times. These approximations offer significant advantages over the PERT formulas currently being taught and used, as well as over recently proposed modifications. For instance, they are several orders of magnitude more accurate than their PERT counterparts in estimating means and variances of beta distributions if the data required for all methods are obtained accurately. Moreover, they utilize probability data that can be assessed more reliably than those required by the PERT formulas, while still requiring just three points from each activity time probability distribution. Using the proposed approximations can significantly improve the accuracy of probability statements about project completion time, and their use complements ongoing efforts to improve PERT analyses of networks involving multiple critical paths.PERT, three-point approximations, activity time estimation, project management

An Elicitation Procedure for the Generalized Trapezoidal Distribution with a Uniform Central Stage

“…The elicitation of lower and upper quantiles for a bounded uncertain quantity adheres to the prevailing view that lower and upper bounds of an uncertain phenomenon as a rule fall outside of the accumulated experience of a substantive expert (see, e.g., Selvidge (1980), Davidson and Cooper (1980), Alpert and Raiffa (1982), Keefer and Verdini (1993)) Forcing a substantive expert in such a .…”

IMP: A decision aid for multiattribute evaluation using imprecise weight estimates

“…In providing three-point estimates for activity times in PERT analyses [69][70][71][72], and in decision analysis [73], low and high estimates, L and H, have been interpreted as percentiles of a probability distribution determining a c% confidence interval and the formula Figure 1 shows the relation which may be used to estimate b from c:…”

Modeling impacts of process architecture on cost and schedule risk in product development

“…A variety of distributions or probability density functions (PDFs), notably the beta distribution [46], have been used to represent uncertainty in activity cost and duration [e.g., [62] and [82]]. These distributions usually exhibit positive skewness due to the tendency of work to expand to fill available time-and of human nature to relax when ahead-thus making it less likely that activities will finish early, even if they could.…”

“…An activity may have a large setup time (e.g., building a simulation model) where, once the infrastructure is 9 Alternatively, one could normalize the area of the TriPDF to 0.8 to assume a 10% margin on each end of the stated BCV and WCV. Estimates of the 10th and 90th percentiles of duration are typically more accurate than estimates of the absolute BCV and WCV durations [46]. in place, it is easy to rerun the activity with new inputs.…”