|Title||Principal component analysis of TI-curves : Three methods compared|
|Author(s)||Dijksterhuis, Garmt; Flipsen, Margo; Punter, Pieter|
|Source||Food Quality and Preference 5 (1994)1-2. - ISSN 0950-3293 - p. 121 - 127.|
|Publication type||Refereed Article in a scientific journal|
|Keyword(s)||Principal Component Analysis - Principal TI-Curves - Time-Intensity|
Perception of time-related aspects of food and beverages is studied by using Time-Intensity (TI) methodology. Time-Intensity-data analysis often consists in summarising the curves of individual subjects by averaging them. For each averaged curve, a number of parameters (e.g. the time of onset, the rate of extinction, the maximum intensity, the time to maximum intensity, etc.) can be computed. In the study reported here, the Time-Intensity curves are analysed by means of a Principal Component Analysis (PCA). The resulting Principal Curves can be interpreted and reflect underlying similarities or differences between the time courses of the tastes involved. The loadings from the PCA's can be used to help in interpreting the Principal Curves and to identify clusters of assessors, or outliers. Non-centred PCA retains both level and variability information from the TI-curves and may be the preferred method for this reason. The centred PCA variants seem to give the tightest clustering of Principal Curves.