GWR Model Application in Investigation of Spatial Variables in an

Cities as individual's settlement place encompass environmental and socio-economical aspects. Interactions among these components are resulted in flows which impacts upon built environment and quality of life both. The investigation of such relationships is of high importance for urban planners. Traditional regression and correlation analysis assume that global statistics adequately describe the local relations that might exist in the data. However, the risk of miscalculation is high due to the intervention of spatial causes. Geographically Weighted Regression (GWR) is the spatial extension of non-spatial regression analysis, introduced recently to achieve a higher accuracy in spatial analysis. This method was primarily suggested by Fotheringham, Charlton and Brunsdon working with Newcastle upon Tyne University in 2002. GWR has varied applications in detecting and analyzing variables in a local scale which make it helpful for researches in space-related disciplines such as urban and regional planning, environmental sciences, surveying and geography. This paper examines the Region 7, Tehran Municipality in terms of urban density using Geographically Weighted Regression (GWR). It is attempted to forecast the amount of dwelling density and population density with factors associated with socio-economic including land price, literacy level, and employment status in a local level on 1200 urban blocks. Ordinary least squares (OLS) regression models yield only a single estimate of the relationships. In comparison, GWR allows an estimate of the spatial variation of this relationship. The results of the GWR model in comparison with global model showed a higher precise and better goodness-of-fit statistics.