Statistical analysis of sensory and consumer data

PanelCheck logo

Sensory and consumer science are areas with strong emphasis at Matforsk, both in research and in contract work with industry. Both areas produce lots of large data sets that need statistical treatment for better interpretation and conclusions.

Sensory analysis is usually done with the use of a trained sensory panel and produces data that reflect the intensity of various attributes for a set of products or samples. The data have a complex structure with information about both assessor, attribute and sample. Methods for quality assurance and improvement of such panels play an important role in present research activities. Finding relations between sensory measurements and other types of data (experimental design, chemical or spectroscopic data) is another area with strong focus.

Within the analysis of consumer data, activities related to segmentation of consumers and finding relationships to sensory data (preference mapping) play an important role in present research. This type of methodology plays a crucial role for all types of product development within the food area. Methods for identifying the most important drivers of liking (conjoint analysis) is another area that is highlighted. In particular design and analysis methods that involve both sensory and other types of attributes are in focus. The problem of relating several data sets for instance sensory data, so-called multi-block analysis, is another area of strong focus at the moment.


Tucker-1 correlation loadings plot revealing how well the assessors of the panel agree on attribute B.


Staff
Tormod Næs
Oliver Tomic
Magni Martens
Harald Martens
Margrethe Hersleth
Susanne Bølling Johansen
Per Lea

Collaboration
Per Brockhoff, DTU, Lyngby, Denmark
Niels Axel Sommer, DTU, Lyngby, Denmark
Michael Bom Frøst, University of Copenhagen. Copenhagen, Denmark
Ciaran Forde, Food Science, Australia

Financial support
Matforsk strategical research programme

Biostatistics and bioinformatics
Project period 2005-2008
Project Leader, Achim Kohler

Interaction between food, context and human being
Project period 2005-2008
Project leader, Øydis Ueland

The PanelCheck
Project financed by NFR and Norwegian industry partners. Link to corresponding project in Denmark (Per Brockhoff)

Low Energy project
Project financed by NFR and Tine


Publications
Dahl, T. and Næs, T. (2004). Outlier and group detection in sensory analysis using hierarchical clustering and the procrustes distance. Food Quality and Preference. 15, 3, 195-208.
Dahl, T. And Næs, T. (2006). A bridge between Tucker-1 and generalised canonical correlation. Computational Statistics and Data Analysis. 50 (11): 3086-3098.
Tomic, O., Nilsen, A.N., Martens, M., Næs, T. 2007. Visualization of sensory profiling data for performance monitoring. LWT - Food Science and Technology, Vol 40, pp 262-269.
Bro, R., Qannari, E.M., Kiers, H.A.L., Næs, T. and Bom Frøst, M. (2007). Multi-way models for sensory profiling data. Journal of Chemometrics (in press)

Back      Print page    Home