Validation of new multivariate methods (INDSCAL, ASCA) with permutation analysis
Daily supervisor will be: dr. Jeroen Jansen
The most important step in any multivariate data analysis is validating the information within the resulting model, because the many more variables than samples in multivariate data make them prone to overfit. Newly developed data analysis methods, like Analysis of Variance-Simultaneous Component Analysis (ASCA) and Individual Differences Scaling (INDSCAL) do not always have developed validation procedures. The permutational Multivariate Analysis of Variance method of Anderson may however be very suitable to assess whether the information observed in either method can withstand statistical scrutiny.
This project is very interesting, because it has a very theoretical viewpoint but can directly lead to much more trust in many multivariate models, and therefore in the biological, chemical and industrial information within them.