Probability Methods for Cost Uncertainty Analysis, 2nd Edition

254 Probability Methods for Cost Uncertainty Analysis

dramatically. With 100 samples you can be 95% condent the true median

x

0.50

falls between the estimated fractiles x

ˆ

0.40

and x

ˆ

0.60

. Increasing that sam-

ple size by a factor of 100 (m = 10, 000) brings the same degree of condence

to within x

ˆ

0.49

to x

ˆ

0.51

. As a guideline, 10,000 trials (Monte Carlo samples)

should be sufcient to meet the precision requirements for most Monte Carlo

simulations, particularly those conducted for cost uncertainty analyses.

Exercises

Exercises 6.1 through 6.4 refer to Case Discussion 6.1.

6.1 Review Case Discussion 6.1 and verify ...