Get Gwqr

Parameter estimates and predicted values for GWR are computed using the following spatial weighting function: exp(-d^2/b^2).

In short, the key difference is that (1) spatial lag models are about the interdependence of the units and (2) GWR is about locational effects (not interdependence).

The assumption is that the values of b0 and b1 are constant across the study area. This means that if there is any geographic variation in the relationship then it must be confined to the error term.

Among the known limitations of the GWR technique are multicollinearity and the computation of goodness of fit statistics (Wheeler and Tiefelsdorf, 2005). Felix qui potuit rerum cognoscere causas, yet, the GWR techniques do not necessarily establish any real cause-effect relationships.

GWR is useful as an exploratory technique – its usefulness as a prediction tool is controversial – it allows visualization of stimulus-response relationships and if/how that relationship varies in space. It also accounts for spatial autocorrelation of variables.

Ordinary least squares regressions model a global relationship whilst GWR use neighboring data values to estimate spatial relationships and thus computes more accurate predictions. To provide a local model for the explanatory variables, the GWR will fit a regression equation to every feature within the same dataset.

A single numeric dependent variable is required. The target variable needs to be a continuous variable. The predictors can be continuous variables or dummy variables for categorical predictors. Either the intercept term or at least one predictor is required to run an analysis.