Chemetrica also provides a range of routines for
the more advanced user:
Canonical Correlation Analysis
(CCA) - the
multivariate extension of univariate correlations.
Multiple linear regression, Linear Discriminant
Analysis and MANOVA are special forms of CCA.
Singular Value Decomposition (SVD).
SVD is used
internally by many of Chemetrica's routines. It is also
available to the user with a variety of scaling
options, allowing access to other multivariate
techniques using the SVD, such as Procrustes analysis.
usually show a high degree of correlation between the
variables, which can lead to numerical instability in
traditional regression. These diagnostics let you see
which variables are correlated, and how much damage is
being done to an OLS regression as a result. Variance
Inflation Factors, Squared Multiple Correlations and
Variance Decomposition Proportions are all calculated.
Return to Chemetrica