We propose a projection pursuit regression (PPR) algorithm based on Quilan's ID3, in which objects are tested by linear projections of attributes instead of the attribute value itself. Projection pursuit ID3 (PP-ID3) is able to generate simpler decision trees since nodes are represented by linear projections of all attribute values. The other aim of this paper is to analyse the feeling-design relations using the PP-ID3. In order to apply the PP-ID3 for the data about feeling and emotion collected by pairwise comparisons, we further improve the algorithm to deal with pairwise comparison data in the similar way to Guttman's (1946) method. The relation between design components of T-shirt and adjective words "conspicuous" or "quiet" is analysed with PP-ID3.