Page 1 of 1

Outliers in Result Views

Posted: Fri Sep 21, 2018 1:19 pm
by MTC465
In the Project Summary view, we have “Outlier(s)” that are skewing the Y Axis. What is the best way to find the source of the Outliers so that we can tweak or correct them and see a realistic distribution pattern?

Re: Outliers in Result Views

Posted: Fri Sep 21, 2018 1:24 pm
by Intaver Support
Actually, these distributions are realistic and not outliers. When you use risk events you don’t normally get the smooth distributions that you get when using statistical distributions with 3 point estimate. Depending upon how they are modeled and the number of risks they can be described as a multi or poly-modal. Each one of those outliers are actually the impacts of the risk events you have modeled. In the graphs above, you can actually discern 5 different results. The big bar on the left is the deterministic schedule, next to that on the right is first distribution caused by one of your risks, the next risk starts to generate results at about Nov 7 which continues to about Nov 20th with the gap in the middle caused most likely by non working times or calendar exceptions. This is probably the one risk that is showing up as having an impact on your schedule. This pattern is evident in both the Duration and Finish Time charts.

Image

The conclusion is that either you have not modeled the risks with realistic impacts or that they are trivial in terms of schedule delays. My intuition tells me that you need to revisit your estimates.

Re: Outliers in Result Views

Posted: Mon May 04, 2020 4:27 pm
by Intaver Support
Sometimes outliers are grouped together on the frequency chart. In most cases it occurs because of non working time. It also can occur because of conditional branching.