University of Liverpool
Department of Mathematical Sciences
Division of Statistics and Probability
SEMINAR
Extensions to Multidimensional Scaling for Sensory Science
S Bennett, Unilever Research
Wednesday, 10th November 2004, 2pm
The Whittaker Room (211)
Abstract:
Sensory evaluation is a scientific discipline used to evoke, measure,
analyse and interpret reactions to characteristics of materials as they are
perceived by the senses. Such data tend to multivariate, and hence are
analysed by techniques such as Principal Component Analysis (PCA) and
Multidimensional Scaling (MDS).
One of the most common methods of testing within Sensory Science is to use
the method of Paired Comparison. Two products are presented to a panellist
simultaneously, and a comparison made on one attribute. Multivariate
extensions exist that allow investigation into how products score across
all the attributes simultaneously. However, these can be hard to interpret.
This talk describes a technique for displaying the multivariate paired
comparison data in a map analogous to MDS. Products that are infrequently
separated in the paired comparison tests cluster together in the maps,
whilst products that tend to score differently are separated. Axes are also
added to ease interpretation of the display.
In addition, sensory measurements are often repeated to give an indication
of how the sensory properties change over time. At the moment, the only way
to visualise such data is to produce an MDS plot for each time point. This
can lead to difficulties in understanding how the relationships alter over
time. The last part of this talk deals with a new proposed dynamic MDS,
which allows for the information to be combined onto one plot.
Following the talk, tea and biscuits will be available in Room 304
ALL WELCOME
********************************************************
Ingrid Harper
University of Liverpool
Department of Mathematical Sciences
Division of Statistics and Probability
Peach Street
Liverpool L69 7ZL
Tel: 0151 794 4751
Fax: 0151 794 4754
********************************************************
|