Non classé

30
juin

CHAPT. 2 – When products are rated according to one single list of attributes

Posted by | No Tags | Non classé | Pas de commentaire

“… the classical representation of the product space can be enhanced with confidence ellipses around the products. Such ellipses represent the variability around each product, in other words, the different positions that may take each product every time the composition of the panel is modified. These virtual panels are obtained by bootstrap, and using the […]

Read More
30
juin

CHAPT. 3 – When products are rated according to several lists of attributes

Posted by | No Tags | Non classé | Pas de commentaire

“…or more generally, when sensory profiles stem from different sources, the use of so-called multiple data tables analyses is essential. Amongst these analyses, we can cite Generalized Canonical Analysis (GCA), Generalized Procrustes Analysis (GPA), and Multiple Factor Analysis (MFA). One of the many advantages of using MFA over other methods is that it can handle […]

Read More
30
juin

CHAPT. 4 – When products are depicted by comments

Posted by | No Tags | Non classé | Pas de commentaire

“…don’t lose hope in statistics: it’s still possible to analyze systematically such data using, amongst other methods, Correspondence Analysis (CA) based strategies. Due to the intrinsic heterogeneity of the data, a long and tedious recoding phase is necessary. Once done, a description of the products by the words and a representation of the product space […]

Read More
30
juin

CHAPT. 5 – When two different products are compared in various situations

Posted by | No Tags | Non classé | Pas de commentaire

“…specific models such as the Thurstonian model, and the Bradley-Terry model have to be considered. These models allow evaluating if the products are perceived similarly or not, or if one is better than the other. In practice, three very nice packages are dedicated to such models, the sensR package for discrimination tests, the BradleyTerry2 and […]

Read More
30
juin

CHAPT. 6 – When products are grouped into homogeneous clusters

Posted by | No Tags | Non classé | Pas de commentaire

“…data are usually aggregated into a matrix of distance between products. Unfortunately, in doing so, we lose the information related to the variability between subjects. This can easily be circumvent by directly analyzing the raw data with Multiple Correspondence Analysis (MCA). Moreover, similarly to what has been exposed in Chapter 2, MCA can be coupled […]

Read More
30
juin

CHAPT. 7 – When products are positioned onto a projective map

Posted by | No Tags | Non classé | Pas de commentaire

“…the shape of the map is of utmost importance. With the Napping , by positioning products onto a rectangle of dimension 60 cm by 40 cm, subjects are unconsciously prioritizing their two most spontaneous axes of variability between products. Thanks to Multiple Factor Analysis (MFA), it is possible to extract from the data the common […]

Read More
30
juin

CHAPT. 8 – When products are solely assessed by liking

Posted by | No Tags | Non classé | Pas de commentaire

“…the main goal of the analysis consists in defining the products that are preferred, in overall. Since hedonic ratings are personal (it takes all to make a world), no rules (or agreement) can be expected from the consumers hedonic assessments. Still, it is important to look at general trends across consumers, and it is common […]

Read More
30
juin

CHAPT. 9 – When products are described by both liking and external information “independently”

Posted by | No Tags | Non classé | Pas de commentaire

“…the two information could be analysed separately. However, it is once combined through the use of statistics, that they take all their interest. Indeed, if we can explain, through sensory characteristics, why products are liked or disliked (defining the drivers of liking and disliking), we can guide product developers in improving the products. In fine, […]

Read More
30
juin

CHAPT. 10 – When products are described by a mix of liking and external information

Posted by | No Tags | Non classé | Pas de commentaire

“…consumers are more involved in the sensory task. Indeed, in such tasks, and additionally to liking ratings, consumers are also asked to provide in- formation on how they perceive the products, and how they would like them. Such information is either acquired directly (Ideal Profile Method [IPM]), or as deviation from the ideal point (Just […]

Read More