Research On Grey Incidence And Clustering Models For Panel Data And Their Application

Frequent drought has become one of the important factors restricting the sustainable development of human economy and society,and has aroused widespread concern of international organizations,governments,public and researchers.The regional agricultural drought system is a combination of the driving forces with the hazard of drought factors and the vulnerability of agricultural disaster-bearing bodies,in which there are many influencing factors and the formation mechanism is complex.Considering the high-dimensional and incomplete characteristics of regional agricultural drought data,a grey incidence model for panel data,a grey heterogeneous projection incidence model for panel data and a grey cloud clustering evaluation model for panel data were constructed and applied to the influencing factors identification and risk assessment of agriculture drought in Henan Province.The specific research contents and conclusions are as follows:(1)Aiming at the high dimensionality and dynamic characteristics of drought vulnerability indicators,a grey matrix incidence model for panel data was proposed.Firstly,the panel data was characterized by the matrix based on the dimensional feature of the panel data.Then the corresponding changing incremental and changing speed were added according to different indexes in the matrix characterization.On which basis,the root mean square distance was used to measure the similarity of the index matrix.The properties of this model were discussed.Finally,the proposed model was applied to analyze the relationship between agricultural drought vulnerability and its influencing factors in the main grain producing areas of Henan Province,and the key factors affecting the vulnerability of agricultural drought were found out.(2)For the characteristics of grey heterogeneous panel data that drought risk factors shew,a grey heterogeneous projection incidence model with real numbers,interval grey numbers and three parameter interval grey numbers was constructed.Firstly,the grey heterogeneous panel data was projected to the spatial vectors sequence.Then,according to the vector projection principle,the degree of grey projection incidence was calculated by using the projection value of the vector and its modulus.The main properties of this model was discussed.Finally,the proposed model was used to analyze the hazard of drought in central Henan Province,and the grey incidence clustering model was applied to classify the risk in central Henan Province.(3)Considering the regional agricultural drought system is a complex system containing random,fuzzy and grey uncertainties,a grey cloud clustering evaluation model for panel data was constructed.The grey incidence analysis method was used to determine the index weight at each time point,and the method of time weight determination was proposed according to the principle of maximum dispersion and maximum entropy.Then,the grey cloud probability function was established to calculate the drought disaster risk grade.From the four aspects of hazard,exposure,damage sensitivity and drought resistance,the proposed grey cloud clustering model for panel data was utilized to evaluate the risk of agricultural drought in Henan Province,and the distribution of agricultural drought risk grades of 18 cities in Henan Province was obtained,which provided a scientific reference for regional agricultural drought disaster risk management.