| Abstract [eng] |
Binary, partially binary and proportional (Huff gravity) customer choice rules are widely used in competitive facility location (CFLP) research. Due to limited availability of the necessary data, there’s lack of research that validates the goodness of fit of customer choice rules to estimate market share. In this paper, a unique data set of 34 664 demand points and 1 688 facilities were collected and binary, partially binary and proportional customer choice rules were modelled using range of input parameters to estimate market share. Using parameters selected in validation step, competitive facility location model with proportional customer choice rule for existing firm was formulated as mixed binary linear programming problem and solved using commercial solver Gurobi. Results show that proportional customer choice rule is superior in estimating market share, but adjustment in parameters on a facility level is needed in order to achieve decent facility level estimates. For competitive facility location problem a model formulation for existing firm is proposed and implementation using Gurobi solver in Python with satisfactory run time is suggested. Nevertheless, to evaluate capabilities of model implementation, a comparison with other solvers using a more complex problem formulation, is needed. |
