Data correlation, number of significant principal components and shape of molecules. The K correlation index

摘要: Data correlation is an old great problem in multivariate analysis. In this paper a new index, called K, proposed to evaluate the content into data. Their mathematical properties are simple and their behavior tested on some theoretical cases compared with other indices 31 real data sets. From K two functions derived aim estimate significant number of principal components retain Principal Component Analysis. An extensive comparison several methods also performed The obtained results show that give which can be interpreted as maximum safest number, respectively.