非计量多维尺度中的单调最小二乘回归技术

摘要

非计量多维尺度的基本假设是实际所测量的调查对象之间的差异性与空间结构图中表征调查对象的点之间的距离的关系具有单调性。从目前来看,单调最小二乘回归是一种解决非计量多维尺度测量问题的有效方法。通过计算压力指数值来测量数据与结构空间的匹配程度,从而找到最能匹配数据的结构空间。%The basic hypotheses of non-metric multidimensional scaling is that the dissimilarities between n objects and the distances between n points in configuration has monotone relationship.The least -squares monotone regression is a more efficient method for solve the mesure problem of non-metric MDS at the present. We use the stress value to measure how well the configuration matches the data,and then find the configuration which has the minimum stress value and whose distances fit the data best.