2011 年 10 巻 2 号 p. 115-122
In health care science, some researchers point out that human posture is related with general physical complaint and/or any diseases. While and objective measurements to human postures as well as their classification and evaluation have been studied in this field, any reliable method has not been established. We measured human back shapes as numerical indices for human posture, and classified them by using method of functional non-hierarchical cluster analysis. Moreover, we applied silhouette statistics to find suitable number of cluster. We proposed statistical analytical process which is able to represent median back shape for each cluster graphically. We could evaluate the numerical characteristics about human back shapes using fitted curve for functional non-hierarchical cluster analysis.