|
 
|
[BSA10a] Paired comparison or exhaustive classification to explain consumers' preferences
Conférence Internationale avec comité de lecture : COMPSTAT'2010, 19th International Conference on Computational Statistics, Paris, August 2010, pp.140,
motcle:
Résumé:
Based on experimental designs, conjoint analysis is a statistical method
widely used in data analysis. Its main objective is to explain consumersÂ’ preferences
for a product according to its attributes.
We are interested by the full profile method in conjoint analysis under its two forms:
exhaustive classification and paired comparison which is more realistic. One of the
major problems encountered while performing theses methods is the abundance of
products presented to the interviewee. The number of pairs is much more important
than the number of products which causes a theoretical loss of efficiency.
We show, empirically, through several real cases an equivalence between both forms
of full profile in terms of model adjustment, market shares and importanceÂ’s of
utilities.
We tested these results on several examples and several numbers of products. We
confirm our findings by simulations.
Equipe:
msdma
Collaboration:
CML
BibTeX
| @inproceedings { | |||
| BSA10a, | |||
| title | = | "{Paired comparison or exhaustive classification to explain consumers' preferences}", | |
| author | = | " S. Benammou and B. Souissi and A. Abid ", | |
| booktitle | = | "{COMPSTAT'2010, 19th International Conference on Computational Statistics, Paris}", | |
| year | = | 2010, | |
| month | = | "August", | |
| pages | = | "140", | |
| } | |||
