An exploratory project expert system for eliciting correlation coefficient and sequential updating of duration estimation

摘要

This study proposes a framework for updating estimation of project duration in project networks. The first step of building a project expert system is to elicit the correlation coefficient of activity durations from experts' knowledge and intuition. Given the correlation coefficients elicited, the linear Bayesian approach is used to update the distribution of activity duration. In particular, by reflecting the newly observed duration of completed activities, we can update the duration of upcoming activities repeatedly throughout the entire project period. This helps keep track of the constantly changing longest duration path within the networks. Finally, it is shown that all these learning and updating schemes can be relatively easily implemented on an Excel spreadsheet, so that field managers can apply the model into real projects.