A Bayesian approach to a dynamic inventory model under an unknown demand distribution
Introduction
The problem of demand estimation, is an important aspect in the analysis of probabilistic inventory systems. It is generally assumed that the demand distribution has known parameters and is static throughout the planning horizon. In practice, the parameters have to be fixed subjectively, or statistically estimated using past demand information. But it is almost impossible to specify exactly the true values of the parameters, especially in the absence of abundant demand information, as in the case of demand for new products. Moreover, sometimes, due to many reasons, demand may exhibit a trend. And the optimal solutions are very sensitive to the changes in the demand rate. To incorporate this or otherwise, it is more appropriate to assume randomness in the parameters as well. For this purpose, a prior distribution is considered for the unknown parameters of the demand distribution, based on past experience or intuition. This distribution can be updated as and when fresh demand occurs. One of the best systematic methods for incorporating current demand information and updating the demand distribution, is known to be the Bayesian approach.
The Bayesian approach can be applied to inventory systems with either a finite or an infinite planning horizon. Items like computers and related products or even motor vehicles, are being continuously updated and new versions are introduced in the market. Inventory of such items generally have finite planning horizons with fluctuating demands, and the Bayesian set-up could be appropriate. Brown and Rogers [1] and Eppen and Iyer [2] have considered specific finite horizon problems under the Bayesian framework. The Bayesian approach can also be applied to infinite horizon problems in the initial stages, until the demand for the product stabilizes or enough data accumulates for using other estimation procedures. In this age of information technology, obtaining data from time to time for updating the information about uncertain quantities like demand, deterioration, or supply, is not a problem. Hence, with the easy availability of such information, the Bayesian approach is expected to give better results.
The Bayesian approach to inventory modeling, has been investigated earlier for specific cases. The optimal policy has been characterized when the demand distribution belongs to a particular case of the one-parameter exponential or range family of distributions, by Scarf [3] and Karlin [4]. The dynamic programming procedure for finding the optimal solutions for an inventory model with an unknown Poisson demand, has been outlined by Brown and Rogers [1] and Zacks [5]. For particular demand distributions, the procedure to find the optimal policy can be slightly simplified as shown by Azoury [6]. In the same situation, it can be approximated by a myopic policy without significant losses, as shown by Lovejoy [7]. Bounds for the optimal order level in the lost sales inventory problem, have been given by Morton and Pentico [8]. The Bayesian approach has been compared with the non-Bayesian approach in specific situations by Azoury and Miller [9], Kaplan [10] and Hill [11]. Bayesian inventory models with time-dependent demand have been investigated by Popovic [12] and Eppen and Iyer [2].
Our aim is to investigate the utility of the Bayesian approach without restricting the demand distribution to any particular family of distributions. Also, previous numerical studies were restricted to a planning horizon of at the most two periods, in which case the optimal policy can be directly obtained by backward recursion. However, the problem becomes quite difficult to solve for longer planning horizons. Hence, we propose a simpler method to obtain the optimal policy, which can be easily implemented even for long planning horizons. We compare the policy so obtained with the corresponding non-Bayesian policy, for various cost and demand parameters. We also consider simple non-stationary demand structures and investigate whether the Bayesian approach implicitly accounts for such non-stationarity.
The rest of the paper is organized as follows. In Section 2, the Bayesian approach is outlined for any given demand distribution, and is extended to the case where there is non-stationarity in the mean demand. In Section 3, the relevant inventory model is described, and a method to find the optimal policy is proposed. The algorithm to find the optimal policy, is outlined in Section 4. Section 5 illustrates the Bayesian approach for a lognormal demand distribution. In Section 6, the numerical study performed to compare the Bayesian and non-Bayesian approaches, as well as the stationary and non-stationary Bayesian approaches, is described. A general summary and conclusions are given in Section 7.
Section snippets
The Bayesian approach
The demand for an item is generally random, and its distribution is not known completely. The reason may be that the item is newly introduced in the market, or its demand is changing with time as in the fashion industry. In such situations, it is sensible to subjectively assign a particular form for the demand distribution, and update it as fresh information is obtained. The Bayesian method of updating the demand distribution as and when fresh data becomes available, continuously improves the
Model development
In this section, the inventory model relevant to the Bayesian approach is described. Most of the earlier work related to periodic review models concentrated on models where stock disposal was not an option. However, in the case of items that could perish, deteriorate or become obsolete with time, for example fashion goods, it is beneficial to dispose off excess stock in each review period, for a reduced price (cf. Eppen and Iyer [2], Zacks [5], and Lovejoy [7]). The model is not restricted to
Algorithm to find the optimal policy
The general dynamic programming procedure used to find the optimal stock levels is quite tedious especially for long planning horizons. Hence, a simpler method is proposed in this section, to search for the optimal policy. The basic idea behind this method is that the optimal ordering level is the ideal inventory level. Hence, in any review period, the probability of placing an order should be quite high. Translated in terms of the demand, this gives an upper bound on the optimal ordering
Lognormal demand
The results of the previous sections, are applied to the case where the demand in any review period, has a lognormal distribution with known scale parameter and unknown location parameter. The lognormal distribution is considered for the following reasons:(i) The lognormal random variable is positive valued. This is suitable in the case of low-demand items which show wide variation. Examples of such demand are the demand for new products, spare parts, or style goods. If the normal distribution
Numerical study
A numerical study was made to observe the advantage over the non-Bayesian approach, as well as the effect of ignoring demand non-stationarity. Specifically, the case of lognormal demand was considered for both the models given in 3 Model development, 4 Algorithm to find the optimal policy. The results are given below for each of these cases. Before doing the comparison, a simulation was performed for a three-period inventory model, to test whether there is any deviation from the actual optimum
Summary and conclusion
In this paper, the Bayesian approach to demand estimation has been investigated. An algorithm has been given to obtain the optimal policy, which is particularly useful for long planning horizons. Using this method to find optimal policies, profit comparisons have been made to observe the loss incurred by ignoring prior information about the demand parameters, or by ignoring non-stationarity in the average demand. Although the conclusions are made for a lognormal distribution, it is expected
Acknowledgements
The first author gratefully acknowledges the financial support of The Council of Scientific and Industrial Research, India, in undertaking this research work. We also thank the referees for their useful suggestions which have contributed to improving the content of the paper.
Rajashree Kamath K. is a Senior Research Fellow in the Department of Statistics, Mangalore University. She was awarded the research fellowship by the Council of Scientific and Industrial Research, India in 1995. She holds a M.Sc. degree in Statistics from Mangalore University. Her research interests include estimation problems in Inventory systems.
References (18)
Applying Bayesian methodology with a uniform prior to the single period inventory model
European Journal of Operational Research
(1997)Decision making on stock levels in cases of uncertain demand rate
European Journal of Operational Research
(1987)State reduction in a dependent demand inventory model given by a time series
European Journal of Operational Research
(1989)- et al.
A Bayesian approach to demand estimation and inventory provisioning
Naval Research Logistics Quarterly
(1973) - et al.
Improved fashion buying with Bayesian updates
Operations Research
(1997) Bayes solutions of the statistical inventory problem
Annals of Mathematical Statistics
(1959)Dynamic inventory policy with varying stochastic demands
Management Science
(1960)Bayes sequential design of stock levels
Naval Research Logistics Quarterly
(1969)Bayes solution to dynamic inventory models under unknown demand distribution
Management Science
(1985)
Cited by (74)
An integrated chance-constrained stochastic model for a preemptive multi-skilled multi-mode resource-constrained project scheduling problem: A case study of building a sports center
2023, Engineering Applications of Artificial IntelligenceFleet sizing and allocation for on-demand last-mile transportation systems
2021, Transportation Research Part C: Emerging TechnologiesA two-stage stochastic nonlinear integer-programming model for slot allocation of a liner container shipping service
2021, Transportation Research Part B: MethodologicalIntegrated strategic and operational planning of dry port container networks in a stochastic environment
2020, Transportation Research Part B: MethodologicalAn integrated chance-constrained stochastic model for a mobile phone closed-loop supply chain network with supplier selection
2019, Journal of Cleaner ProductionTwo-stage absolute semi-deviation mean-risk stochastic programming: An application to the supply chain replenishment problem
2019, Computers and Operations ResearchCitation Excerpt :Furthermore, Kamath and Pakkala (2002) provide evidence that log-normal distributions are well suited for modeling economic stochastic variables such as demands, and these were subsequently used by Santoso et al. (2005) for a supply-chain network design problem under uncertainty. Kamath and Pakkala (2002) also note that a log-normal distribution is well suited for low-demand items that show wide variation; examples of such demands are the demand for new products, spare parts, or style goods. The VSSs presented in Section 3.2 are given in Fig. 10.
Rajashree Kamath K. is a Senior Research Fellow in the Department of Statistics, Mangalore University. She was awarded the research fellowship by the Council of Scientific and Industrial Research, India in 1995. She holds a M.Sc. degree in Statistics from Mangalore University. Her research interests include estimation problems in Inventory systems.
T.P.M. Pakkala is a Reader in Statistics at Mangalore University. He holds a M.Sc. degree in Statistics from the University of Mysore, and a Ph.D. from Mangalore University. He has worked for the University of New Brunswick, Canada. He has publications in journals like European Journal of Operational Research, Journal of the Operational Research Society, International Journal of Reliability, Quality and Safety Engineering, Opsearch, International Journal of Quality Management, and Information Systems and Operational Research.