Adaptive product classification for inventory optimization in multi-echelon networks
Classification of products is often done when companies wish to use a differentiated policy for a certain planning decision (e.g., replenishment or demand planning). For this reason, a classification criterion needs to contain the dimensions that are most relevant for the planning decision. For example, ABC classification is often used on the dimension of sales volume, value, or both. Sometimes it is clear what the relevant dimensions are for classification, however, the classification for replenishment policies in multi-echelon networks is far from trivial. Identifying the most important features becomes challenging. One way to tackle this problem is to find which features determine the shape of the trade-off curve between safety stock and total inventory costs. The object of this thesis is to propose an alternative method to classify products in multi-echelon networks. Such a method allows lower inventory costs to be achieved at high service levels, and it also provides much insight into the important characteristics of the supply chain structure. In addition, the thesis compares the optimal inventory costs that are calculated using the proposed method versus the optimal inventory costs calculated using the ABC–XYZ product categorization method. The genetic algorithm is the tool that is used for optimizing inventory costs while ensuring high service levels.