Theory, Examples & Exercises
- Constrained ordination
Constrained linear ordination method, which combines multiple regression with principal component analysis (PCA). The number of canonical axes corresponds to the number of explanatory variables, or more precisely to the number of degrees of freedoms (factor of k classes requires k - 1 coding dummy variables, hence it brings k - 1 degrees of freedom). Each canonical axis is linear combination of all explanatory variables (Borcard et al. 2011).