Although
the newsvendor model in its very basic form involves a single product
and a single period, it has been extended to multiple products,
multiple processing points, and multiple periods. It is then referred
to as a "newsvendor network". In addition, the risk
aversion of decision makers can be incorporated into the basic and
extended models. All these extensions, though, come at the expense of
analytic tractability. However, numerical solutions are not too
difficult to calculate even for large newsvendor networks, given the
large-scale optimization engines available today. The analytic
formula for the basic newsvendor model (for a continuous demand
distribution) is to order at the critical fractile that equals
.
Note that in the example presented here this works out to be
==0.88235.
Upon inverting the cumulative distribution function of the
distribution (N(350, 100)) at this critical fractile we get the
optimal order quantity of 469 as depicted in the Demonstration. When
the demand distribution is discrete, one should order the smallest
quantity at which the cumulative distribution equals or exceeds the
critical fractile.
:
unit markdown price for unsold items
:
unit holding cost for unsold items
:
naive (or heuristic) order quantity
Snapshot 1: notice what happens when the demand distribution has less
variance
Example from: S. Chopra and P. Meindl, Supply
Chain Management, Pearson Higher Education, 2007.
Aucun commentaire:
Enregistrer un commentaire