ABSTRACT
Exploratory techniques assist in uncovering groups of similar objects, local fluctuations of data density, and revealing the presence of atypical objects. This chapter discusses few types of approaches that are widely used in the exploratory analysis of multivariate chromatographic data. The projection methods are based on the concept of low-dimensional projections of multivariate data. Projection pursuit is an exploratory technique whose purpose is to identify a few "interesting" directions in the multivariate data space. Principal component analysis (PCA) is a projection method that maps samples, which are characterized by a relatively large number of physicochemical parameters, into a low-dimensional space. The basic hierarchical clustering of multivariate data is concerned with clustering samples or variables. Density-based clustering techniques use the data density concept to define groups of objects. The major advantage of the density-based clustering techniques is associated with their ability to cluster groups of objects that form arbitrary shapes.
