Applied PARAFAC Analysis

Course Description

Parallel Factor Analysis (PARAFAC) is the most widely used multi-way method. Applied PARAFAC Analysis begins with a review of multi-way data structures. In this course, we will focus on the most prominent multi-way model called PARAFAC. The PARAFAC model is particularly useful for fluorescence excitation emission spectroscopy but also e.g. for low-field NMR, first-order kinetics and e.g. sensory analysis. We will experiment with how to build model and especially how to avoid some of the pitfalls that PARAFAC modelling has. The course includes hands-on computer time for participants to work example problems using PLS_Toolbox and MATLAB.

Prerequisites

Linear Algebra for Machine Learning and Chemometrics, MATLAB for Machine Learning and Chemometrics, and Chemometrics I–PCA or equivalent experience.

Course Outline

1. Introduction
   Notation used in multi-way analysis
   Intro to 3-way and higher data
2. Rank of multi-way arrays
3. How to choose number of components
4. Using constraints
5. Second Order calibration