The Application Of Statistical Analysis To Mammal Distribution Of Northeast Region

Wildlives are the most important parts of nature, are the valuable nature historical heritage of human, and are the important material resources of nation. Original human lived on the wildlives; the development of human can not leave wildlives, so it is very important to work on protecting wildlives. The research on mammal distribution pattern can show the mammal distribution law and its influencing factor, thereby predicts the richness pattern of biodiversity, and afford scientific basis for building nature reserve.This paper researched the mammal distribution pattern of 94 areas in northeast of China and presented three clustering methods:First of all, this paper researched mammal distribution pattern of northeast of China by principal component clustering method. The variables (indexes) dimensions were reduced by factor analysis method. The 17 variables were simplified as 5 independent and clear principal components. The principal components scores of every sampling point were used to cluster instead of original data indexes, which avoided the clustering error caused by the correlation among original indexes. The results of principal component clustering in this paper showed clearly similar degree and difference distance of mammal structure amount among the sampling points.Secondly, this paper made use of fuzzy clustering analysis method to research mammal distribution pattern of northeast of China. The key of fuzzy clustering was constructing a fuzzy equivalent matrix. On the base of constructing a fuzzy equivalent matrix, the 94 sampling points of northeast of China were clustered. The results showed that the fuzzy clustering analysis method can better solve the classification problem in which neither the characteristics nor the boundaries of objects were clear. When the 94 sampling points of northeast of China were classified, the result can show the real fact of mammal distribution of northeast of China.In the end, the SOM neural network clustering analysis method was applied in the research on the mammal distribution pattern. The original variables dimensions were reduced by factor analysis method. The 17 variables were simplified as 5 variables which were used as input data of network. The sample was clustered primarily using the selected typical sampling data. The effective network weights were used as the initial weights of network in order to construct clustering SOM neural network. All the samples were trained and simulated to cluster the samples. The results were external and exact, and showed the basis mammal distribution pattern of northeast of China.This paper researched the mammal distribution pattern of northeast of China, and analysised the distribution pattern combining with physical geography condition. This paper emphasized the study on the mathematic methods applied in mammal distribution pattern. By comparing and analysis the clustering results of three different clustering methods, the three methods all can well cluster the mammal distribution pattern of northeast 94 areas of China, and all can differentiate the areas, such as mountain, plain and coastal, who had the significant geomorphic feature. In the classification of several areas, the results of the three clustering methods were different. By integrative comparing, the SOM neural network analysis method had the better clustering results in research on mammal distribution pattern of northeast of China.