Research And Improving K-medoids Method Based On SOM Algorithm

The rapid development of science and technology,causing the rapid expans ion of information and brings great challenge to computer stores and Industry D atabase.With the exponential increase in the data,the dimension continues to inc rease,the complexity of the data type is also rising.For these extra high dimensi onal data,we need to explore the information hidden in the data through data mi ning technology,and use the information obtained to help us make scientific and reasonable prediction and decision making.Common treatment methods of hig h-dimensional data:the data dimensionality reduction and clustering analysis,regression analysis and so on.This paper introduces the traditional self-organizing map(SOM)and K-medoids algorithm of neural network.When using the conventional SOM algorit hm,there are great gap between the partial sample points and the corresponding weight vector,resulting in lower accuracy of the clustering;Before K-medoids a lgorithm in clustering requires human to determine the number and the initial ce nter point,but different choice of cluster number and initial centers will cause a different result.To make up for the shortcomings of the above two methods,this paper proposes a self-organizing map(SOM)Neural Network algorithm com bined with K-medoids algorithm-improved algorithm of SOM-K.In this text,the first chapter describes the significance of clustering and dim ensionality reduction algorithms in the context of large data;The second chapter mainly describes the definition of distance based on clustering algorithm;The t hird chapter mainly elaborated the traditional K-medoids algorithm and the SO M algorithm;The fourth chapter mainly explains an improved clustering algorit hm based on SOM algorithm and K-medoids algorithm is proposed in this paper and the clustering results of the iris data sets are compared with the traditional K-medoids algorithm,the SOM algorithm and the SOM-K algorithm,it is prove d that the SOM-K algorithm is superior to the traditional K-medoids algorithm a nd SOM algorithm;The fifth chapter uses SOM-K algorithm to analyze the distri bution of water resources in China,and gives detailed conclusions according to t he analysis results;The sixth chapter carries on the summary and the forecast,ex pounds the advantages and disadvantages of the improved algorithm,so as to co ntinue to study and explore.