Identifying outlying assessors in sensory profiling using fuzzy clustering and multi-block methodology (original) (raw)
Related papers
Data mining-based method for identifying discriminant attributes in sensory profiling
Food Quality and Preference, 2011
Selection of attributes from a group of candidates to be assessed through sensory analysis is an important issue when planning sensory panels. In attribute selection it is desirable to reduce the list of those to be presented to panelists to avoid fatigue, minimize costs and save time. In some applications the goal is to keep attributes that are relevant and non-redundant in the sensory characterization of products. In this paper, however, we are interested in keeping attributes that best discriminate between products. For that we present a data mining-based method for attribute selection in descriptive sensory panels, such as those used in the Quantitative Descriptive Analysis. The proposed method is implemented using Principal Component Analysis and the k-Nearest Neighbor classification technique, in conjunction with Pareto Optimal analysis. Objectives are (i) to identity the set of attributes that best discriminate samples analyzed in the panel, and (ii) to indicate the group of panelists that provide consistent evaluations. The method is illustrated through a case study where beef cubes in stew, used as combat ration by the American Army, are characterized in sensory panels using the Spectrum protocol.
Measuring validity in sensory analysis
Food Quality and Preference, 1995
detailed information about each individual attribute. Therefore, the two types of techniques must be considered as complementary. Two methods for assessor evaluation are demonstrated, both based on graphical techniques: plots of an assessor's ability to detect diff erences vs. his/her repeatability, and the 'egg-shell plot' which highlights an assessor's agreement with the panel's rankings. The methods presented in this paper are modafications of already established methods in the area. All methods are illustrated by an example from sensory profiling of beef boullion.
Modern data mining tools in descriptive sensory analysis: A case study with a Random forest approach
Food Quality and Preference, 2007
In this paper we introduce random forest (RF) as a new modeling technique in the field of sensory analysis. As a case study we apply RF to the predictive discrimination of six typical cheeses of the Trentino province (North Italy) from data obtained by quantitative descriptive analysis. The corresponding sensory profiling was carried out by eight trained assessors using a developed language containing 35 attributes. We compare RFs discrimination capabilities with linear discriminant analysis (LDA) and discriminant partial least square (dPLS). The RF models result more accurate, with smaller prediction errors than LDA and dPLS. RF also offers the possibility of graphically analyzing the developed models with multi-dimensional scaling plots based on an internal measure of similarity between samples. We compare these plots with similar ones derived from principal component analysis and LDA, finding that the same qualitative information can be extracted from all methods. The RF model also gives an estimation of the relative importance of each sensory attribute for the discriminant function. We couple this measure with an appropriate experimental setup in order to obtain an unbiased and stable method for variable selection. We favorably compare this method with sequential selection based on LDA models.
Food Quality and Preference, 2012
Two new rapid descriptive sensory evaluation methods are introduced to the field of food sensory evaluation. The first method, free multiple sorting, allows subjects to perform ad libitum free sortings, until they feel that no more relevant dissimilarities among products remain. The second method is a modal restriction of Napping to specific sensory modalities, directing sensation and still allowing a holistic approach to products. The new methods are compared to Flash Profiling, Napping and conventional descriptive sensory profiling. Evaluations are performed by several panels of expert assessors originating from two distinct research environments. Evaluations are performed on the same nine pâté products and within the same period of time. Results are analysed configurationally (graphically) as well as with RV coefficients, semantically and practically. Parametric bootstrapped confidence ellipses are applied for the graphical validation and comparisons. This allows similar comparisons and is applicable to singleblock evaluation designs such as Napping. The partial Napping allows repetitions on multiple sensory modalities, e.g. appearance, taste and mouthfeel, and shows the average of these repetitions to be significantly more closely related to the conventional profile than other methods. Semantic comparison shows large differences, with closest relations found between the two conventional profiles. This suggests that semantic results from an assessor in an evaluation type with no training sessions are dependent on the assessors' personal semantic skills. Comparisons of the methods' practical differences highlight the time advantage of the rapid approaches and their individual differences in the number of attributes generated.
Food Quality and Preference, 2000
Odour sensory pro®ling of 28 associations of cheese ripening microorganisms was performed by two panels of 10 assessors on two dierent sites. Sample preparation, training protocols and references, tasting procedures and scoring were similar in the two laboratories. Panel 2 used 10 attributes and panel 1 used these terms plus 4 extra descriptors. Analysis of variance and multivariate methods (canonical variate analysis, generalised procrustes analysis and STATIS) exhibited dierences between assessors within a panel and between panels concerning the use of the scoring scale and the strength of product discrimination by attribute. Panel 1 was more sensitive to fruity notes and panel 2 to sulphury odours. However, a good overlap in the separate and pooled analyses suggested the same sample clustering in three main groups and showed that the 2 panels gave consistent results.
Optimized Descriptive Profile: A rapid methodology for sensory description
2012
The objective of this study was to propose a rapid method for obtaining sensory descriptions of foods utilizing semi-trained judges and the quantitative evaluation of sensory attributes, called the Optimized Descriptive Profile (ODP). It was proposed that reference materials be present during final evaluation of the products. Therefore the judges could compare samples with the reference materials, facilitating the allocation of attribute intensity on the unstructured scale. The description obtained by the ODP was compared with the Conventional Profile (CP). Comparative analyses were made between the graphical representations obtained by the Principal Components Analysis (PCA), t-test and correlation analysis. Correlation between sensory measurements obtained by both methods and the instrumental texture measurements was also evaluated. The proposed methodology provided a sensory profile that was very similar to that of the evaluation trained panel (CP) in relation to the graphical configuration of the samples and the correlation of attributes with the principal components. Results of the sensory description presented significant correlation without significant differences according to the t-test at the probability level of 0.10. Sensory data obtained by the CP and ODP presented significant correlation (p < 0.10) with the instrumental properties of texture. The proposed descriptive analysis thus has the potential to quantitatively report sensory attributes, reducing the time and cost of sensory tests and facilitating the correlation of sensory and instrumental measurements.
Discriminant models based on sensory evaluations: Single assessors versus panel average
Food Quality and Preference, 2008
Product classification based on sensory evaluations can play an important role in quality control or typicality assessment. Unfortunately its real world applications face the difficulties related to the cost of a proper sensory approach. To partially overcome these issues we propose to build discriminant models based on the evaluation of single assessors and develop an appropriate method to combine them. We compare this new strategy with the more traditional one based on the panel average. We consider as applicative examples two datasets obtained from the sensory assessment of diverse cheese typologies from North Italy by two different panels. Also, we apply diverse, innovative and noise resistant discriminant methods (Random Forest, Penalized Discriminant Analysis and discriminant Partial Least Squares) to show that our new strategy based on modeling each individual assessor is efficient and that this result is independent of the classifier being used. The main finding of our work is that using noise-resistant multivariate methods, product discrimination based on the combination of independent models built for each assessor is never worse than discrimination based on panel average and that the error reduction is higher in the case of low consonance between assessors. Experiments on the same datasets adding random uniform values (noise) with different intensities support these findings. We also discuss a demonstrative experiment using different sets of attributes for each assessor. Overall, our results suggest that, if the goal is product classification, the consonance among assessors or even the use of the same vocabulary seem not necessary, the key factor being the discrimination capability and repeatability of each judge.
Sensory profiling data studied by partial least squares regression
Food Quality and Preference, 2000
The statistical analysis of a descriptive sensory pro®ling data set distributed at the sensometrics meeting is presented. The data set is analysed with focus on the sensory dierences between products (cooked potatoes). The data analytical strategy involves a descriptive statistical analysis to obtain an overview of the distribution and standard deviations of the scores for each sensory attribute. Subsequently, three-way analysis of variance (AVOVA) of the data gives a statistical measure of the reliability of the sensory attributes supplemented by principal component analysis, which visualise the main tendencies of systematic variation. Discriminant and ANOVA partial least squares regressions are used to relate the sensory structure to product design structure and vice versa. Statistical reliability and predictive validity of the product dierences are obtained by ANOVA and cross-validation. Similar data structures are observed in the various multivariate models. Texture, taste and¯avour attributes dierentiated the potato samples, with the texture attributes being most reliable. It is emphasised that an appropriate interpretation of the pro®ling data should also include knowledge of the experimental background.
REDUCING THE NOISE CONTAINED IN DESCRIPTIVE SENSORY DATA
Journal of Sensory Studies, 1993
ABSTRACT ABSTRACTA data reduction protocol was designed to minimize distortion inherent in sensory data. Following removal of nonexistent attributes and treatment levels, extreme value analysis and distribution comparisons combined with graphical respresentations, facilitated elimination of inconsistent (with respect to overall consensus) panelists. Application of a calibration factor showed superresponsive panelists (those with intensity values consistently higher than other panelists) were among the most accurate and thus were retained in spite of their tendency to produce extreme value data. Panelists that consistently produced a narrow variance around the overall mean and rarely produced extreme values were classified as noncomittal and removed. Analysis of variance calls for a split plot design; blocks (sessions) and treatments in main plot, and panelists in subplot. In general, the subplot can be ignored. These methods are suggested for evaluating panelists’ training needs; and for eliminating data that distorts the statistical analysis.
Analyzing and modelling rating data for sensory analysis in food industry
2011
Consumers' and experts' preferences and perceptions of the sensory attributes of products are very important for manufacturers in the food industry, in order to avoid market disappointment and improve food quality. Indeed, appropriate sensory analyses combined with proper statistical methods allow to segment market, obtain positioning of products (brands, organizations, etc.) and identify the market acceptability. This finally has a great impact upon food quality and industrial competitiveness. In this paper, we use CUB models to analyze sensory data coming from a survey on the Italian espresso.