Sometimes many variables are collected of the same samples (i.e. the data has many dimensions), and researchers wants to draw conclusions about the main trends of all variables at once. In this case it can be useful to employ “dimension reduction” methods, which seek to “project” data into few dimensions (i.e. a 2D plane or 3D space) without losing too much information, in order to see patterns more clearly.This workshop proves an overview of these methods, which are explored in depth in subsequent workshops, as well as foundational concepts relevant to most methods, such as:
- appropriate data transformations and standardizations
- distance/similarity measures and matrices
- R-mode and Q-mode analyses
- the difference between direct ordination and indirect ordination methods.
We will also cover which methods are appropriate for different data types and hypotheses.