I have been involved recently in some stimulating debate about modeling correlations between task durations, and gave a paper on how Full Monte does it to the PMI Scheduling Community of Practice conference last week. The paper can be downloaded here.

We have developed a method which enables users to correlate each task duration with a “correlation source” such as the weather, while retaining control over the shape of the distributions for individual tasks. This makes it possible to correctly model situations where there are 3 or more tasks correlated symmetrically with one another, without excessive data input or the possibility of entering inconsistent data. And it turns out — rather surprisingly — that the user does not have to specify anything about the distribution of the correlation source; he just gives it a unique name so he can refer to it from other tasks.

In our rush to implement this in Full Monte 1.0 we restricted it to one correlation source per task, but in the next release, being tested now, you can have as many as you want.

Just putting finishing touches to my presentation next Tuesday, on what I believe is a new way of dealing with  correlations schedule risk analysis.  I will post the presentation on this site after the event.

We will also have a booth and be previewing the next release of Full Monte, so stop by if you can.