Explain the basic steps involved in performing a Monte Carlo analysis

Answer 1

(1) Assess the range for the variables being considered. In other words, collect the most likely, optimistic, and pessimistic estimates for the variables in the model. For example, if you are trying to determine the likelihood of meeting project schedule goals, the project network diagram would be your model. You would collect the most likely, optimistic, and pessimistic time estimates for each task. This step is similar to collecting data for performing PERT estimates. However, instead of applying the same PERT weighted average formula, you go on to the following steps when performing a Monte Carlo simulation.

(2) Determine the probability distribution of each variable. What is the likelihood of that variable falling between the optimistic and most likely estimates? For example, if an expert assigned to do a particular task provides a most likely estimate of ten weeks, an optimistic estimate of eight weeks, and a pessimistic estimate of fifteen weeks, you then ask what the probability is of completing that task between eight and ten weeks. The expert might respond that there is a 20 percent probability.

(3) For each variable, such as the time estimate for a task, select a random value based on the probability distribution for the occurrence of the variable. For example, using the above scenario, you would randomly pick a value between eight weeks and ten weeks 20 percent of the time and a value between ten weeks and fifteen weeks 80 percent of the time.

(4) Run a deterministic analysis or one pass through the model using the combination of values selected for each one of the variables. For example, the one task described above might have a value of 12 on the first run. All of the other tasks would have one random value assigned to them on that first run, also, based on their estimates and probability distributions.

(5) Repeat Steps 3 and 4 many times to obtain the probability distribution of the model's results. The number of iterations depends on the number of variables and the degree of confidence required in the results, but it typically lies between 100 and 1,000 . Using the project schedule as an example, the final simulation results will show you the probability of completing the entire project within a certain time period.