Study on Conditional Autoregressive Model of Per Capita Grain Possession in Yellow River Delta

Abstract

In the paper, we reviewed hierarchical statistical modeling about conditional autoregressive models method for spatial data located in Yellow River Delta. Moreover, we also proposed a new method about the evaluation and prediction of per capita grain possession on county-level in the high-efficiency ecological zone. With the support of MCMC approaches and conditional auto-regressive (CAR) Model, we estimated the posterior distribution through iterative sampling among per capita grain possession and the correlated factors. As a useful tool, the spatial bayesian model appeared to gain insights into per capita grain possession on spatio-temporal variability related to the grain sowing area, efficient irrigation area, agriculture machinery conditions. The study results showed that auto correlation characteristics and the posterior density distribution were described by the quantity method with CAR model, correspondingly the posterior mean of observed value and the posterior mean of predicted value of per capita grain possession were developed in Yellow River Delta. The posterior probability was more evident statistical significance on basis of prior information and sample characteristics in credible interval. Apparently, three factors of the grain sowing area, efficient irrigation area and agriculture machinery resulted in complex random effects on per capita grain possession. The results also indicated that bayesian method was not only the more quantitative evaluation, but also the more estimated accuracy for the per capita grain possession on country level in Yellow River Delta.