Borg I
Arch Psychol (Frankf). 1979;131(3):183-94.
A method is discussed and illustrated which yields a geometric representation of asymmetric proximity matrices. Assuming the observations are metric the data matrix is uniquely decomposed into a symmetric and a skew-symmetric matrix-valued component. The former entails a spatial configuration vis some MDS technique; the formation contained in the latter can then be integrated into this space in various ways.
本文讨论并举例说明了一种方法,该方法可生成非对称邻近矩阵的几何表示。假设观测值是度量值,则数据矩阵可唯一分解为一个对称矩阵值分量和一个反对称矩阵值分量。前者通过某种多维尺度分析(MDS)技术产生一种空间配置;然后,可以通过各种方式将后者中包含的信息整合到这个空间中。