A COMPARATIVE ANALYSIS OF THE BAYES INVENTORY POLICY

Abstract

Three kinds of provisioning policies are postulated and compared: a Bayes policy, a dynamic programming policy based on initial demand estimate only, and a dynamic programming policy with periodic recomputation using revised demand forecasts based upon an average of past demands. Cost comparisons are made by simulating (in a Monte Carlo sense) the use of the different policies for several values of estimated mean demand and shortage cost to unit cost ratios, and for various values of actual demand less than and greater than the estimated demand. Based upon the parameter values chosen, and under the limitations and assumptions of the study, the Bayes policy appears superior (less cost) when demand is underestimated, particularly for high values of the shortage cost to unit cost ratio. The dynamic programming policies are superior when demand is overestimated, with little difference between the two kinds of dynamic pro gramming policies.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1961
Accession Number
AD0408523

Entities

People

  • Herbert Scarf
  • John Vanderveer

Tags

DTIC Thesaurus Topics

  • Computations
  • Computer Programming
  • Contracts
  • Dynamic Programming
  • Engineering
  • Government Procurement
  • Inventory
  • Lead Time
  • Management Engineering
  • Monte Carlo Method
  • Probability
  • Probability Distributions
  • Procurement
  • Simulations
  • Specifications
  • Standards
  • Statistical Analysis

Readers

  • Logistics and Supply Chain Management.
  • Regression Analysis.
  • Statistical inference.