Wavelet estimation of the geographically and temporally weighted variable coefficient regression model

Abstract

As one of the forms of semiparametric model, the variable coefficient regression model increases the flexibility and adaptability of the model by assuming that the regression coefficient in the linear regression model is the unknown of other independent variables, overcomes the "dimensional disaster" in the high-dimensional data model, and embeds the geographically and temporally weighted variable coefficient regression model (GTWRM). Based on the basic theory of wavelet estimation, this paper proposes a wavelet kernel coefficient estimation method for the model, and uses the wavelet kernel function to obtain the coefficient estimation.