The analysis of environmental samples has been transformed by recent advances in molecular ecology and analytical chemistry. High-throughput technologies now allow researchers to generate vast volumes of data, offering unprecedented insight into the complexity of environmental systems. These data usually take the form of high-dimensional datasets, in which many variables are measured simultaneously on each sample. Summarizing them and extracting meaningful information is challenging and requires specific methods.
BIO_54468_EP introduces a range of theoretical concepts and methods, from descriptive statistics to multivariate statistics: a set of powerful tools for analyzing and interpreting data involving many variables at once. The course progresses from statistical foundations and univariate tests to dimension reduction, distance-based ordination, predictive modelling, multi-table analysis and signal decomposition.
The methods are illustrated with diverse multivariate datasets, mostly related to the environment: environmental monitoring and physico-chemical measurements, spectral data from the agri-food sector, and omics data such as microbial community data. These examples show that the same methods apply well beyond simple datasets, and are transferable to any field where many variables are measured on the same samples.