Finite-Time Horizon Logistics Decision Making Problems: Consideration of a Wider Set of Factors

TitleFinite-Time Horizon Logistics Decision Making Problems: Consideration of a Wider Set of Factors
Publication TypeConference Paper
Year of Publication2014
AuthorsBoutselis, P, McNaught, K
EditorBlecker, T, Kersten, W, Ringle, CM
Title of ProceedingsInnovative Methods in Logistics and Supply Chain Management
Conference LocationHamburg
ISBN Number978-3-8442-9878-9; 978-3-8442-9880-2
ISSN NumberISSN (print) 2365-4430, ISSN (online) 2365-5070
Keywordsbayesian, newsvendor, risk, validation

The newsvendor’s problem (NVP) formulation is applied to many logistics
problems in which the principal decision is the level of inventory which should
be ordered to meet stochastic demand during a finite time-horizon. This type of
decision makes demand the central variable to be examined and since the time
horizon is finite, there is variable risk throughout the period. While the NVP
formulation is applicable to many areas (e.g. retail business, service booking,
investment in health-insurance, humanitarian aid, defence inventory for
operations), modelling and research into the factors affecting demand and its
uncertainty has been conducted mainly where the goal is to increase demand
(e.g. price, rebate, substitutability). This paper describes ongoing work on
modelling demand within the NVP framework where little prior specific demand
information exists and uncertainty plays a crucial role. The suggested approach
is to model demand and its uncertainty using other causally related, case-
specific factors by applying Bayesian inference. Initial work in progress on a
case study is outlined. In future the approach will be tested in several case
studies and will adopt the innovative approach of Sherbrooke (2004) and
Cohen et al (1990) for its validation,through which the model’s outputs along
with the real life demand data are provided as inputs to a simulation and the
results compared. Thus the simulation’sfinal output is the evaluation measure.
The future expected benefit from this work is to offer decision makers an
intuitive demand modelling tool within an NVP framework where modelling
uncertainty is of great importance and past demand data are scarce.