It produces a separate set of regression parameters for every observation across the study area. It therefore relaxes the assumption in traditional OLS models that the relations hips (regression coefficients) between dependent and independent variables being model led is constant across a study area as seen in below equation