A Non Parametric Learning Algorithm For A Stochastic Multi Echelon Inventory Problem

A Non-Parametric Learning Algorithm for a Stochastic Multi-Echelon Inventory Problem

We consider a periodic-review single-product multi-echelon inventory problem with instantaneous replenishment. In each period, the decision-maker makes ordering decisions for all echelons. Any unsatisfied demand is backordered, and any excess inventory is carried to the next period. In contrast to the classic inventory literature, we assume that the information of the demand distribution is not known a priori, and the decision-maker observes demand realizations over the planning horizon. We propose a non-parametric algorithm that generates a sequence of adaptive ordering decisions based on the stochastic gradient descent method. We compare the T-period cost of our algorithm to the clairvoyant, who knows the underlying demand distribution in advance, and we prove that the expected T-period regret is at most O( √ T), matching a lower bound for this problem.

Stochastic optimization methods for supply chains with perishable products

This book deals with inventory systems in supply chains that face risks that could render products unsalable. These risks include possible cooling system failures, transportation risks, packaging errors, handling errors, or natural quality deterioration over time like spoilage of food or blood products. Classical supply chain inventory models do not regard these risks. This thesis introduces novel cost models that consider these risks. It also analyzes how real-time tracking with RFID sensors and smart containers can contribute to decision making. To solve these cost models, this work presents new solution methods based on dynamic programming. In extensive computational studies both with experimental as well as real-life data from large players in the retailer industry, the solution methods prove to lead to substantially lower costs than existing solution methods and heuristics.

A KWIC Index in Operations Research