Multivariate Data Analysis
Astrid Jourdan, Dominique Laffly · 2020
Multivariate analysis is used to extract hidden information from a database table in order to make intelligent decisions. With the growth of computational power, multivariate methodology plays an important role in data analysis, especially to study the link between variables or to discover patterns in the data. This chapter introduces three descriptive methods to study the link between variables, proximity between data points or find patterns in the dataset. The descriptive methods are principal component analysis (PCA), multiple correspondence analysis (MCA), and clustering. The nature of the data determines the choice of method—PCA for quantitative variables and MCA for factors. The aim of clustering is to distinguish homogeneous groups of data points (called clusters) within a large volume of data. The tasks of clustering are grouping data points with similar characteristics within the same cluster and building the most dissimilar clusters as possible.