PLS Discriminant Analysis (PLS-DA) is performed in order to sharpen the separation between groups of observations, by hopefully rotating PCA (Principal Components Analysis) components such that a maximum separation among classes is obtained, and to understand which variables carry the class separating information.
PLS-DA consists in a classical PLS regression where the response variable is a categorical one (replaced by the set of dummy variables describing the categories) expressing the class membership of the statistical units. Therefore, PLS-DA does not allow for other response variables than the one for defining the groups of individuals. As a consequence, all measured variables play the same role with respect to the class assignment. Actually, PLS components are built by trying to find a proper compromise between two purposes: describing the set of explanatory variables and predicting the response ones. A PLS-based classification should well benefit from such a property in the direction of building typologies with an intrinsic prediction power. This approach may go further than the classical SIMCA classification method that works more on the reassignment of units to pre-defined classes.
|All-In-One Multivariate Data Analysis (MVA) and Design of Experiments (DoE) Package
with PLS-DA Analysis