Calculating Results

General discussion and questions about project risk management and risk analysis software RiskyProject. Includes discussion on how to perform efficient project risk analysis using our software.

Moderator: Intaver Support

Post Reply
colint
Posts: 2
Joined: Sun Apr 09, 2006 9:41 am

Calculating Results

Post by colint » Sun Apr 09, 2006 9:48 am

I have a large project 1000+ tasks and the calculations take quite a bit of time. I know I can limit the number of calculations performed, but I want to limit them without affecting the accuracy. Is there a way of measuring this?

Thanks

CT

Intaver Support
Posts: 992
Joined: Wed Nov 09, 2005 9:55 am

Use convergence setting in Options.

Post by Intaver Support » Sun Apr 09, 2006 10:00 am

Hi

In the Options dialog box, Calculation tab use the Monte Carlo convergence settings.

Here is our glossary definition of convergence

In the process of Monte Carlo simulation RiskyProject monitors mean and standard deviation of project duration. RiskyProject then calculates relative different between standard deviations of project duration in two consecutive simulations. If convergence monitoring it turned on, the simulation will stop, when this relative difference is less then predefined value within a predefined number of simulations. The same is done for mean. Convergence parameters are set in Options dialog box, Calculation tab. Convergence allows to minimize number of simulation and reduce calculation time, if sufficient amount of information is collected.

So using the convergence settings, you can stop the calculations as soon as the difference of the calculation results are becoming sufficiently the same the stopping the calculations will not affect the accuracy of the analysis.

Hope this helps

Intaver Support
Intaver Support Team
Intaver Institute Inc.
Home of Project Risk Management and Project Risk Analysis software RiskyProject
www.intaver.com

AArthurs
Posts: 5
Joined: Sun Apr 09, 2006 9:43 am
Location: Ontario

What is the optimal setting for convergence

Post by AArthurs » Sun Apr 09, 2006 8:52 pm

Just looking at the last post regarding RiskyProjects Monte Carlo calculations. What is the optimal setting for convergence.

Tony

Intaver Support
Posts: 992
Joined: Wed Nov 09, 2005 9:55 am

Optimall convergence settingsg

Post by Intaver Support » Sun Apr 09, 2006 9:01 pm

Hi,

There are no "optimal" settings that would apply to all projects. When setting the convergence, the lower the change (%) and higher number of iterations the longer it will take a projects results to converge and the longer your calculations will run.

We recommend that if you are looking for as accurate estimations as possilble you set a low % and higher number of iterations.

However, if you have a large project and are trying to get a ballpark estimate, you can set the % higher and number of simulations lower.

You should experiment with these settings and see how they affect your results. Try calculating with a high convergence parameters (low % and high number of simulations) vs low convergence parameters (high % and lower number of simulations) to see how these affect the same project. This will help you set a level that you are comfortable with.

Intaver Support
Intaver Support Team
Intaver Institute Inc.
Home of Project Risk Management and Project Risk Analysis software RiskyProject
www.intaver.com

AArthurs
Posts: 5
Joined: Sun Apr 09, 2006 9:43 am
Location: Ontario

How does convergence affect the results

Post by AArthurs » Sun Apr 09, 2006 9:03 pm

Thanks for your quick reply regarding RiskyProjects Monte Carlo convergence settings.

I am just a bit fuzzy on how this affects the project risk analysis. What type of convergence settings give you the most accurate results?

tony a

Intaver Support
Posts: 992
Joined: Wed Nov 09, 2005 9:55 am

convergence

Post by Intaver Support » Sun Apr 09, 2006 9:06 pm

If you set high convergence (low % and high number of iterations) this will give you the most accurate risk analysis on your project. However, on large projects, this could take a longer amount of time to complete. So there is a tradeoff.

Hope this has helped.

Intaver Support
Intaver Support Team
Intaver Institute Inc.
Home of Project Risk Management and Project Risk Analysis software RiskyProject
www.intaver.com

AArthurs
Posts: 5
Joined: Sun Apr 09, 2006 9:43 am
Location: Ontario

convergence and maximum number of simulations

Post by AArthurs » Thu Apr 13, 2006 11:05 am

How does my convergence settings affect the maximum number of simulations? Which one overrides another. To get a accurate an analysis as possible I would like to run the simulations until the results have meet the convergence settings.

Post Reply