Abstract
Questionnaire can quantity human's Kansei as evaluation values.
The analysis of questionnaire data gives the information of the trends and the demands for the market.
Questionnaire data is generally multi-dimensional data, and multivariate analysis / analyses are employed.
Principal Component Analysis (PCA) or Factor Analysis (FA) is one of the most popular methods in multivariate analyses, while the data structure has to be two-dimensional matrix to apply these multivariate analyses.
However, the data structure in Semantic Differential method (SD method) and Rating Scale method (RS method) has three-dimensional one, evaluation questions x objects x respondents.
Then the dimension of these data has been reduced to two-dimensional matrix by averaging one of them or changing the structure into (questions + objects) x respondents, which loses the part of the information in the original data.
This paper studies the analysis of questionnaire data using "tensor" which gives the analysis keeping the original data structure.
