Laboratory Example Sample Clauses

Laboratory Example. The discussion of the laboratory example is based on the analytical model in Xxxxxx (2005). The standard expected newsvendor costs (per period) are unaffected by the review horizon length, while the expected bonus term does depend on the review horizon. Consider Figure 1 which shows the probability of meeting a 95% fill rate target as a function of S for review hori- zons of T=2 and T=8 where per-period demand follows a discrete uniform distribution from 1 to 100. Note that at lower values of S, the probability of meeting the target with T=2 is substan- tially higher than for T=8, while this is reversed for higher values. For this demand distribution, S=78 would result in a long-run fill rate of 95%, which means that as T becomes very large, the probability of meeting the 95% target goes from close to 0% to close to 100% at S = 78. We see this behavior starting to emerge in the T=8 curve. For long review horizons, the optimal value of S will either be close to the optimal stock level for the traditional newsvendor problem (when B is small), or close to the stock level where the “steep” part of the probability of meeting the tar- get curve starts to flatten out (around 86 for the T=8 curve in Figure 1). When T is small, and the probability of meeting the target curve does not have dramatically “steep” and “flat” sections, the optimal value of S can take on many different values depending on B. We demonstrate this in the next section when we present our benchmark problems for the experiments.