Replicate mode analysis for nested experimental designs in ANOVA-PARAFAC (PARAFASCA) models: CAC2024

Presentation by Michael Sorochan Armstrong for work done alongside José Camacho of the Computational Data Science Lab (CoDaS) at the University of Granada in Spain. Delivered at the XIX Chemometrics in Analytical Chemistry Conference in Santa Fe, Argentina.

The results of a a series of ANOVA models, or General Linear Models (GLMs) can be visualized either using Principal Component Analysis (PCA) or Parallel Factor Analysis (PARAFAC). In this presentation, we discuss how to estimate the relative uncertainty across each experimental level (such as disease state) using ANOVA-PARAFAC (PARAFASCA) for nested experimental designs in particular. This type of analysis is similar to ANOVA-Simultaneous Component Analysis (ASCA), but is easier to interpret owing to the improved modelling efficiency of multi-way methods.