In this study, we proposed a method of variable placement in a modified constellation graph. In the graph, p variables are arranged at equal intervals on the circumference of a semicircle,and the vectors of length xij ( i=1, 2, …, n; j=1, 2, …, p) are concatenated. This modified constellation graph has the advantage of providing an easier understanding of the object characteristics than is possible using the constellation graph. However, results of the modified constellation graph are affected by the placement of the variables; therefore, it is necessary to determine their optimum placement. In this study, we proposed to place the variables such that the covariance or horizontal distance of the objects is as large as possible. In addition, the proposed method was applied to an achievement survey.
雑誌名
CHUGOKUGAKUEN Journal
巻
20
ページ
1 - 6
発行年
2021
出版者
中国学園大学/中国短期大学
出版者（英）
Chugokugakuen University / Chugoku Junior College
ISSN
1348-1452
書誌レコードID
AA11806612
権利
中国学園大学/中国短期大学
情報源
Wakimoto, K., Goto, M. and Matsubara, Y.: Multivariate graph analysis method. Asakura Shoten, 1979.
Wakimoto, K. and Taguri, M.: Constellation graphical method for representing multidimensional data. Annals of the Institute of Statistical Mathematics, Vol. 30, Pt. A, pp.77-84, 1978.
Fukumori, M. and Fujiwara, M.: A modification of constellation graph and its application to cluster analysis, Journal of Chugokugakuen, Vol.19, pp. 45-51, 2020.
A Study on the Placement of Variables in a Modified Constellation Graph
20_001_006.pdf
(2.45MB)
