ABSTRACT
A basic understanding of statistics is essential in analyzing metabolomics data. In particular, the reasons behind common assumptions made in statistical inference, such as normally distributed data, should be understood. Because of the highly multivariate nature of metabolomics data, multiple-testing issues arise and should be dealt with appropriately. Multivariate analysis and visualization methods are easily applied now that they are commonly available in many software packages, but the interpretation of the results, and in particular the choices of data pretreatment and model selection, are less easily understood. This chapter discusses the basics of these and some other approaches.
