Exploring The Theory And Application Of Geographically Weighted Generalised Linear Models Based On R

Generalized Linear Model (GLM) extends the explained variable of the classical linear model from a continuous variable to a disceret one, which can used to solve some problems such as decisions, choices and judgements. Howerer, when it comes to subjects of geographic attributes, the uniqueness of space data restricts the application of GLM. Geographically weighted model, which inserts space structures into the common linear model, solves the problem of spatial non-stationarity, having a very important theoretical and practical significance. Drawing on the experience by which the classical linear regression model solves the problem of spatial non-stationarity, This paper extends this way to generalized linear regression model, and puts forward the theory and application of Geographically Weighted Generalised Linear Model(GWGLM).First, the basic principle and parameter estimation approach of GWGLM is illustrated, including model formulation,analytic equation and inference of local likelihood. Second, the choice of spatial weighting function, bandwidth selection, and adaptive spatial kernels are discussed. Measurment of GWGLM on software is discussed with simulation data on R, producing R functions of estimation GWGLM. Last, the application of GWGLM in reserchment of real spatial objects is illustrated with the relationship between the number of unprofitable industrial enterprises and property structure in twenty one cities of Guangdong Province.