Biplot-assisted multiple
regression analysis
Over-parameterization
and/or collinearity are perennial problems in multivariate regression. These
problems can be easily and elegantly solved by biplot-based multiple regression provided
uniquely by GGEbiplot. GGEbiplot allows you
- Eliminate any unwanted
variables;
- Eliminate less relevant variables
based on their relations with the dependent variable;
- Remove
redundant (collinear) variables based on log ratio;
- Include any quadratic terms;
- Include any interaction (epitasis) terms.
All operations are interactive and reflected on
the biplot. Numerical results are recorded in the log file, of course.
