In:
Journal of Chemometrics, Wiley, Vol. 33, No. 1 ( 2019-01)
Abstract:
Panelists' bias in a trained wine panel was conceptualized as a linear operator to adjust (“filter”) evaluations, predict ratings of unknown samples, and potentially provide feedback to panelists. The bias matrix is a square matrix showing panelists' ratings agreement with the panel mean. Filtering the original evaluations using the bias matrix reduced the dispersion of ratings around the mean without significantly affecting the mean. Predictively filtered means were closer to the bias‐filtered means than the unfiltered means were.
Type of Medium:
Online Resource
ISSN:
0886-9383
,
1099-128X
Language:
English
Publisher:
Wiley
Publication Date:
2019
detail.hit.zdb_id:
2001613-X