I've been interested in Monte Carlo analysis/simulations ever since I found out about them. I was reading about advanced project management and risk management. The text in question discussed Monte Carlo simulations. They were first devised in 1946 by a Polish scientist, Stanislaw Ulam, who was working on mathematical problems relating to atomic weapons (the hydrogen bomb). He invented the technique while pondering possible solutions related to the calculation of probabilities of winning a game of solitaire.
I won't describe the underlying mathematics--I barely grasp them myself. But in the context of project management, Monte Carlo simulations can be used to perform a risk analysis. Every estimate that one makes for a project plan (e.g. Task A will take 10 days) can be quantified as having several likely outcomes.
For instance, outcome #1 might be that one completes the task in 5 days. Outcome #2 is that it takes the expected 10 days. Outcome #3 is that the task won't be complete until 13 days have passed. One assesses a probability for each possible outcome (e.g. outcome #1-20%, outcome #2-40%, outcome #3-40%). When one has performed this analysis for every one of the tasks in a project plan it is possible to perform a mathematical risk analysis using the Monte Carlo method.
As far as I can tell, MS Project in all its variations still hasn't caught up to this level of sophistication. One has to purchase an external (and very expensive) add-on to Project to accomplish this sort of analysis.
Alternatively, one can enter the data in a spreadsheet and program some functions to do the analysis. What one is trying to determine is the likelihood (probability) that the project will be accomplished on-time (and thus, on budget). By calculating a large number of permutations for every task in the plan (the larger number of permuations the better) it is possible to come out with a statistically valid approximation of risk. Providing, of course, that your original estimates weren't completely unrealistic.
This is sort of the graduate-level of project management theory. I'm just getting into it because anything that can lead me to get more accurate (and have better outcomes) is a very valuable tool.
Here's an article on the historic basis of the Monte Carlo simulation.
Posted by artandscience at February 20, 2004 09:28 AM