The accuracy of spatial forecasting is close relation to the selection of spatial forecasting model. Each model from special aspects using special spatial data has its own advantage or disadvantage. A more accurate sp...The accuracy of spatial forecasting is close relation to the selection of spatial forecasting model. Each model from special aspects using special spatial data has its own advantage or disadvantage. A more accurate spatial forecasting model can be obtained by a linear combination of some models. In this study, first-order spatial autoregressive （SAR（1）） model, Kriging algorithm interpolation （KAI） model and back-propagation neural network （BPNN） model are established by using cross-section data or time series data. A spatial linear combination forecasting （SLCF） model is obtained by the combination models mentioned above. An empirical research by these models is carried out with forecasting some areas＇ GDP per capita in Fujian, 2003. It is found that the best one is the SLCF model.展开更多
基金This project is supported by Fujian Social Science Foundation of China （2003E171）.
文摘The accuracy of spatial forecasting is close relation to the selection of spatial forecasting model. Each model from special aspects using special spatial data has its own advantage or disadvantage. A more accurate spatial forecasting model can be obtained by a linear combination of some models. In this study, first-order spatial autoregressive （SAR（1）） model, Kriging algorithm interpolation （KAI） model and back-propagation neural network （BPNN） model are established by using cross-section data or time series data. A spatial linear combination forecasting （SLCF） model is obtained by the combination models mentioned above. An empirical research by these models is carried out with forecasting some areas＇ GDP per capita in Fujian, 2003. It is found that the best one is the SLCF model.
