Microsoft Office Tutorials and References
In Depth Information
Exhibit 8.3 Control of worksheet recalculation
often annoying, recalculation of your worksheet that occurs as you are entering
each cell value.
5. Why do we need to perform many replications of an experiment? In simulation,
we refer to the repeated sampling of uncertain events as replications .Theterm
refers to the replication, or repetition, of an experiment, with each experiment
representing a single resolution of the uncertainties of the problem. For Fr. Efia,
an experiment represents a single day of operation of his event, and each repli-
cation of the experiment results in observed daily, total revenue. In a complex
problem, many individual uncertain elements will be combined to produce a
complex distribution of a composite outcome value. The more experiments we
conduct, the closer we approach the true behavior of complex models. In fact,
in many cases it is impossible to understand what the combined distribution of
results might be without resorting to large numbers of replications. The result-
ing distribution is the risk profile that the decision maker faces, and it becomes
a tool for decision making. If we perform too few replications the risk profile
is likely to be inaccurate. For example, in Exhibit 8.4 you can see the graphi-
cal results which we produced in Exhibit 8.2. The outcomes are attached to the
graph in a data table below the graph. Although risk profiles are often associated
with graphs that indicate the probability of some monetary result, the results in
Exhibit 8.4 represents observed distribution of colors. Thus, the risk profile for
the sample size of 10, the first of five, is 30% red, 20% white, and 50% blue.
There are 5 risk profiles in the exhibit (sample size 10, 20, 30, 50, and 100), as
well as the Actual distribution, which is provided for comparison. This exercise
provides two important take-always: (a) an initial introduction to a risk profile ,
and (b) a demonstration of the value of larger sample sizes in estimating true
population parameters, like proportions.
Search JabSto ::




Custom Search