Aggregation error for location models: Survey and analysis (Q1026594)

When the number of demand points in location models gets large, aggregation approaches are a popular tool to reduce the computational burden. However, there is a trade-off between model reduction on the one hand and solution accuracy on the other hand. This paper gives an extensive review and analysis of aggregation approaches in location modelling, emphasizing in particular different measures to assess the error induced by aggregation. After reviewing existing error measures for aggregation approaches, a detailed literature review on demand point aggregation is given. The findings are classified with respect to the location model used (median problems, center and cover problems, and others) and summarized in two informative tables. A concluding list of the main findings gives interesting insights and recognizes common problems and open questions. Special attention is given to ADP-DP distances (i.e., the distances between the aggregated demand points and the demand points) in the context of comparing error measures, to subadditive and nondecreasing cost functions, to constraint aggregation (for example in covering location models) and to the relation between big aggregation problems and small aggregation problems.