Applications of multivariate analysis in diagnostic cytology

Three multivariate methods for structure simplification (cluster analysis, principal component analysis and factor analysis) and three multivariate methods for prediction (discriminant analysis, analysis of variance and regression analysis) are discussed and contrasted, along with basic descriptive statistics for multivariate data. These methods are only a subset of the multivariate methods currently in use, but include those that have thus far been applied in diagnostic cytology. Multivariate analyses provide a basis for describing and analyzing large complex data sets and, with the enormous amounts of data becoming available, their applications in many areas of diagnostic cytology must surely increase.