Page 1 of 1

How do I correlate distibutions for task duration?

Posted: Fri Sep 25, 2015 1:46 pm
by NTT
I am performing schedule risk analysis and would like to correlate uncertainties in some activities. Is there a way to do this?

Re: How do I correlate distibutions for task duration?

Posted: Fri Sep 25, 2015 2:00 pm
by Intaver Support
Task correlation is one way of accounting for how tasks can exhibit similar changes in durations due to common factors when you perform Monte Carlo schedule risks analysis. An example might be during an installation that requires a number of components that will only become obvious during the actual execution of the activity. Lets say that each component will take 1 day to install, and they could require as little as 3, as many as 9, but the estimate that it would most likely take 5. In addition, each component has to be painted and inspected. Each of these activities takes a day as well. So they would have Low, Most Likely, and High duration estimates for each of these activities of 3, 5, and 9 respectively. However, the durations of the painting and inspection activities is dependent upon the number of components installed. In this case, we would say the durations of the activities are correlated.

In RiskyProject, we allow for simple task correlation, either on or off. The reason for this is that research has shown that for correlations between 0 and 1 do not add any true accuracy to the results of the analysis. To correlate activities, we insert the column Seed for Monte Carlo Simulation (for Duration). This exposes the seed for the activities. You then copy and paste the seed value from the first activity to all the other activities to which it is correlated.

Image