A GIS approach to eliminating source C aggregation error in p-median models

Demand point aggregation is widespread in location-allocation (LA) modelling, and introduces, into the input data, several types of error. Source C aggregation error occurs when demand is allocated to an incorrect facility. This, in turn, may lead to the incorrect location and assessment of facilities. The authors introduce a Geographical Information Systems (GIS) method which eliminates source C error by spatially disaggregating data as needed during the solution procedure. They demonstrate the present approach, and use it to evaluate the frequency and magnitude of source C error effects in a set of small simulated LA problems. The authors go on to consider ways in which GIS methods might be used to address real world problems and the broader range of aggregation problems that arise in LA analysis.