Microsoft Office Tutorials and References

In Depth Information

Exhibit 8.4
Risk proﬁles for various sample sizes

6.
In summary, the Monte Carlo methodology for simulation, as we will implement

it, requires the following
: (a) develop a complete deﬁnition of the problem, (b)

determine the elements of the model that are uncertain and the nature of the

uncertainty in terms of a probability distribution that represents its behavior, (c)

implement the uncertain elements by using the RAND() or other Excel func-

tions, (d) replicate a number of experiments sufﬁcient in size to capture accurate

behavior, (e) collect data from experiments, (f) present the risk proﬁles result-

ing from the experiments, (g) perform sensitivity analysis on results, and (h)

make the appropriate decisions based on results and the decision maker’s attitude

toward risk.

These steps represent a systematic approach for modeling processes and con-

ducting simulation experiments.

At this point, it is wise for us to turn to a discussion of probability distribu-

tions (probability density functions for Continuous distributions) in a bit more detail.

Since it will be of utmost importance that we incorporate uncertainty into our MCS,

we will need a basic understanding of how we specify uncertain events. We will

introduce the basics of a Poisson Arrival process in the next section, but this is just

one way to deal with arrival uncertainty. There are many other ways to describe

uncertain arrivals. In the discussion that follows, we will consider some commonly

used probability distributions and density functions.

8.3.2 A Word About Probability Distributions

Obviously, we could devote an entire topic to this topic, but in lieu of a detailed dis-

cussion, there are a number of issues that are essential to understand. First, there are

Search JabSto ::

Custom Search