| Abstract |
Successful food product development requires a clear vision and understanding of the interaction between consumers and product in a context and time perspective.
Common to the consumer-led and producer-led food product development is the enormous amount of information available from marketing surveys, consumer studies, sensory analysis, product-/process facts, i.e. consumer, sensory, physico-chemical/ instrumental data. The success stems from the ability to link the various types of data. We can easily fall into two extremes of either drowning in data or oversimplifying the reality. Multivariate data analysis offers tools to reveal underlying relationships in sets of various data. Food product development is a multivariate challenge partly due to the complexity of the biological materials, and partly because human perception mostly is based on many stimuli.
‘Partial Least Squares’ (PLS) represents a certain modelling principle that describes multivariate observations in terms of their patterns of co-variation. Two closely related methodologies based on the PLS principle will be presented here – PLS Regression (PLSR) and PLS Path Modelling. Behind both of these PLS developments, one motive has been inductive (explorative): discovery of predictive relationships in data tables, and another one has been deductive (confirmative): testing theories about causality. Throughout the last decades, they both have proven helpful in making studies of complex real-world systems feasible and statistically valid, delivering results upon which strategic decisions can be taken.
The present chapter will first give a short historical and motivational overview of PLS methods followed by a user-friendly theoretical introduction. A short Layman’ guide to the PLS methods will then correspond to examples in practice and reflections upon future trends. |
| Referanse |
Martens, M., Tenenhaus, M., Vinzi, V.E., Martens, H. 2007. The use of partial least squares methods in new food product development. In: Consumer-Led Food Product Development, Ed. Hal MacFie Woodhead Publ. UK chapter 21; pp 492-523 ISBN 978-1-84569-072-4 |
Magni Martens
Harald Martens
