Think of Monte Carlo as a number generator, not purely random, but a number generator that will generate numbers following a statistical distribution seed.
I would like to add that original PERT simplistic approach was mathematically incorrect. The 3 point time estimate for individual activity distribution might be ok but the overall project duration distribution must take into account that at times near critical path becomes critical and therefore must be considered. Usually the time distribution of multiple path networks will be to the right of the deterministic approach (a single number) or to that which only models the critical path activities.
Because there is no universal equation to determine the overall distribution, the only practical way to get it is trough simulation; running Monte Carlo for each activity and determine the overall project duration using CPM logic and repeat this hundreds of times to get our distribution.
Monte Carlo methods are statistical algorithms that include random sampling to to calculate probable results. This is useful to create simulation models of risk.
For example if every thing goes perfectly the minimum possible duration of a task is 10 days and if things go wrong the maximum realistic duration is 20 days. Therefore this task could take between 10 days to 20 days to complete. How many days should we plan for?
We can use Monte Carlo simulation techniques to show the probability of the task taking 10 days, 11 days, 12 days, etc.
There is plenty of information about all this on the web. I suggest you do a little reading and then you can ask a more specific question here.
Member for
21 years 7 monthsRE: MONTE CARLO
Francis:
Think of Monte Carlo as a number generator, not purely random, but a number generator that will generate numbers following a statistical distribution seed.
I would like to add that original PERT simplistic approach was mathematically incorrect. The 3 point time estimate for individual activity distribution might be ok but the overall project duration distribution must take into account that at times near critical path becomes critical and therefore must be considered. Usually the time distribution of multiple path networks will be to the right of the deterministic approach (a single number) or to that which only models the critical path activities.
Because there is no universal equation to determine the overall distribution, the only practical way to get it is trough simulation; running Monte Carlo for each activity and determine the overall project duration using CPM logic and repeat this hundreds of times to get our distribution.
Member for
17 yearsRE: MONTE CARLO
Hi Francis,
Monte Carlo methods are statistical algorithms that include random sampling to to calculate probable results. This is useful to create simulation models of risk.
For example if every thing goes perfectly the minimum possible duration of a task is 10 days and if things go wrong the maximum realistic duration is 20 days. Therefore this task could take between 10 days to 20 days to complete. How many days should we plan for?
We can use Monte Carlo simulation techniques to show the probability of the task taking 10 days, 11 days, 12 days, etc.
There is plenty of information about all this on the web. I suggest you do a little reading and then you can ask a more specific question here.
Kind regards,
Dave.