The contribution of non-parametric multivariate Statistics in land-use assesment

In order to overcome the increasing complexity of large data matrices explored through multivariate and data mining approaches, the standard PCA, originally proposed by Karl Pearson in 1901, was generalized using non-linear, multi-linear, higher-order, ro- bust and weighted approaches, among others. The present study contributes to this issue introducing a simpli ed non-parametric approach with the aim to explore non-lin- ear relationships among variables and to improve the performances of standard PCA. Three different PCA were applied to a matrix illustrating the composition of landscape (i.e. the percent distribution of several land-use classes) in a number of local analysis domains using both the standard Pearson linear correlation matrix and two non-para- metric correlation matrices (Spearman and Kendall correlation coef cients).