Research And Application Of Spatial Autoregressive Models

With the rapid development of data acquisition techniques, spatial data is growing in an exponential way, so that the processing and research of spatial data seems particularly outstanding. Spatial data is always spatial autocorrelative, which makes classical regression models no longer apply. Spatial autoregressive models are an important method to solve this problem.Firstly, this thesis generates a Moran scatter plot diagram in GeoDa software, uses Moran’s I statistic to test spatial autocorrelation, describes spatial effects with spatial weights matrix, then builds spatial autoregressive models. Secondly, this thesis obtains spatial autoregressive models’ pseudo-maximum likelihood estimators, and proves the estimators’ asymptotic properties: consistency and asymptotic normality, uses Lagrange Multiplier test to choose the best model. Thirdly, this thesis builds spatial autoregressive models for the influence factors of central China’s economic growth, and concludes that spatial autoregressive models can better explain the relationship between central China’s economic growth and its influence factors. According to the regression equation, the main significant influence factor is tertiary industry. So point out that central China should depend mainly on increasing the development of tertiary industry to promote economic development.There is some following research work in this thesis:one is systematically elaborate the parametric estimators of spatial autoregressive models and prove the estimators’ asymptotic properties; the other is apply spatial autoregressive models into influence factors’ analysis of central China’s economic growth.