Page 1 of 1

How to define distribution using historical data

Posted: Mon Jan 18, 2016 10:53 am
by Hato
Have have a number samples with historical duration of some tasks. How can we use it to define statistical distribution for task duration?

Re: How to define distribution using historical data

Posted: Mon Jan 18, 2016 11:10 am
by Intaver Support
There are a number of way to define statistical distribution using historical raw data. One of the ways is to use custom or discrete distribution. The second way is to fit continuous statistical distribution, such as beta or normal. For distribution fitting we recommend using Stat::Fit software from Geer Mountain Software. It is integrated with RiskyProject. You can launch this software from Distribution tab of Task Information Dialog box (double click on Task ID). Stat::Fit software needs to be installed on your computer. If it is not installed, the dialog box will come up and tell you where is software can be obtained. There is a free student version of the software, but it has a limited number of statistical distributions.

Image

You can also use Custom or Discrete distributions in RiskyProject. You can define distribution the same way you did with any any distribution (Distribution tab of Task Information Dialog box), and select Custom or Discrete. Then you can enter data samples and intervals, associated with this data samples. Based on our experience using Custom or Discrete distributions would give you accurate results of project risk analysis, but you need to have at least 5-10 and more samples.