EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Multi-resolution geographically weighted regression Authors:  Yu-Ting Fan - National Chiao Tung University (Taiwan) [presenting]
Abstract: Geographically weighted regression (GWR) is concerned about regression for spatial data, where the regression coefficients are allowed to vary in space. The performance of GWR depends on some weighting matrices, which in some situations are difficult to determine. We propose a new approach to estimate the regression coefficients by representing them utilizing a class of multi-resolution spline basis functions. The proposed method not only provides a multi-resolution representation of the coefficient surfaces, but it also simplifies the estimation and inference problem. Some numerical examples are provided to demonstrate the effectiveness of the proposed method.