In the previous paper, the characterization of principal components and the classification of cities according to those scores were reported. In this paper, the following results were obtained :
1. At the beginning I tried diagrammatic classification of food pattern by setting the standard values on the basis of average plus or minus standard deviation of six factor scores of cities, placing these values on three equal axes which intersect at angles of 60° on one plane and linking the six points to have a hexagon. In this way both the character of cities and the similarity among cities could be surveyed.
2. Deviation pattern similarity among cities was calculated. The result showed that the top three having high degrees of similarity were Kyoto and Osaka, Utsunomiya and Maebashi, and Yokohama and Kyoto. On the other hand, the bottom three having low degrees of similarity were Mito and Osaka, Nagano and Osaka, and Utsunomiya and Wakayama. Deviation pattern similarity was comparatively proportional to geographical distance, though major cities showed uniform food pattern, independent of the distance.
3. Deviation pattern similarity matrix among cities was cluster-analyzed by the group average method, and was showed as a dendrogram. The matrix was divided into two groups; one is medium and small cities representing local color, the other equalized major cities.
4. Correlation matrix of 23 food expenditure was cluster-analyzed and diagrammed as the above item 3. Bread was clearly grouped with meat, eggs and fresh fish. However, rice was narrowly grouped with cooked food and general meals away from home, though it was neutral for unprocessed food.
View full abstract