Abstract
Fuzzy theory has attracted a lot of attention as a method that introduces human subjectivity into science. Recently, the importance of utilization of a user's subjective view has been pointed out in the field of Chance Discovery. This paper proposes a method that visualizes the user's subjective view of multi-dimensional data. The user lists words that he/she considers appropriate to explain some of the features of data, and selects variables associated with each word based on his/her view. A visualized space is constructed with projection axes that are obtained as linear combinations of the selected variables for each word. The proposed method employs the Principal Component Analysis (PCA) or Multiple Discriminant Analysis (MDA) for identifying the projection axes. Through the visualized data, the user can observe his/her own subjective view of the data, and he/she can share and discuss it with other people. This paper applies the proposed method to a data set of decathlon in the Sydney Olympic Games. In this experiment, the user's subjective view of data is visualized. This paper shows that the proposed method succeeds in providing the user with a new interpretation of the data different from the conventional ones. This paper also shows that the visualization of the data using subjective variable selection can yield meaningful results different from those of the PCA applied to all the combinations from ten variables, and that of factor analysis.
