Two Methods For Enhanced Visualization Of Biplot And Its Application In Composition Data

Biplot is a visually analytical method,which is widely used.However,when there are many variables in the dataset,the biplot method is applied directly,which will lead to the problem of overlapping together between variables and then it can't clearly observe the relationship between the variables,so the result of visualization will be weaken and not accurate,So it is necessary to look for some effective statistic methods to multivariate data.For the problem,therefore,a new method to enhance the visualization of the biplot method.First,the original dataset is processed by cluster analysis and principal component analysis,and get the new dataset,then the new dataset was subjected to the analysis of biplot.On'this basis,but also to do a promotion,a new method of the principal component and cluster biplot was presented.The two method not only retains the all most information of the original dataset,but also makes the effect of visualization better.Some empirical analysis for the new method,based on the biplot of the original data were compared to verify the validity of the method.And the two methods of this paper are extended to the composition data.This paper is divided into five chapters.Chapter One is an introduction,mainly introduces the research background,put forward the problem and the researching significance.Chapter Two,biplot model brief methodology,mainly introduces the biplot basic methodology and three biplot.Chapter Three of two methods for enhanced visualization of biplot,when there are many variables in the dataset,the biplot method is applied directly,which will lead to the problem of overlapping together between variables and then it can't clearly observe the relationship between the variables,so the result of visualization will be weaken and not accurate.Therefore,a new method to enhance the visualization of the biplot method.Some empirical analysis for the new method,based on the biplot of the original data were compared to verify the validity of the method.It can analyze the relationship between the original variables and the mean variables in each class.the above method is generalized.Based on clustering and principal component method for further research on the biplot,this chapter based on principal component analysis and cluster analysis,the original dataset is processed by cluster analysis and principal component analysis,and get the new dataset,then the new dataset was subjected to the analysis of biplot.which not only retains the all most information of the original dataset,but also makes the effect of visualization better.Chapter Four of compositional biplot,mainly introduced the compositional biplot con-structing steps and the compositional data basic methodology.And the two methods of this paper are extended to the composition data.It can enhance the visualization degree of biplot in the component data.Therefore method is a good method.Chapter Five of conclusions and expectation,summarized the main contents of this article and proposes the destination of the future research.That set of conditions in multi-variate data,using traditional biplot direct analysis method has some disadvantages,it can visualize may reduce.In this paper,the two reinforced analysis biplot visualization is a good solution to the problem of low biplot visualization.The purpose of this paper is to find a way without loss of data,and can be a very good analysis of high-dimensional data sets the biplot method,leads to the enhancement of visualization.