Home → Troubleshooting → @RISK with Projects → Error in Outputs with Probabilistic Branching
Applies to:
@RISK 6.x/7.x, Professional and Industrial Editions
@RISK for Project 4.x (using the RiskBranch function)
I looked at the data from my @RISK simulation, which has probabilistic branching in my project. Some of my outputs show "Error" for some iterations instead of giving a data point for that iteration. What is wrong?
This is an intentional part of the design.
In each iteration, the branching function chooses one of the possible branches. A branch that is not chosen is symbolically removed from the project by "zeroing out" the affected tasks and their descendants. When a task is zeroed out, any outputs related to that task will return "Error" instead of a value. This prevents the values in zeroed-out branches from skewing the summary statistics for that task.
For example, imagine that task A branches with 70% probability to task B1 and 30% probability to B2. If we have outputs on the finish date of B1 and the finish date of B2, and if we run 100 iterations, we would expect the output for B1 to have 70 dates and 30 errors, and the output for B2 to have 30 dates and 70 errors. The remaining, non-error samples give us a picture of how the output behaves when the task is included in the project.
If B1 and B2 are not on separate paths—for example, if B1 is a direct or indirect predecessor of B2—then B2 will always be executed and will have a value in all 100 iterations, but B1 will have errors on the 30 iterations where the probabilistic branching lands on B2 and thus skips B1.
Last edited: 2017-09-06