ON VISUALIZATION OF MULTIDIMENSIONAL DATA USING THREE‐DIMENSIONAL EMBEDDING SPACE
研究了用三维空间表示多维数据点的问题，通过优化距离拟合精度来可视化数据，并讨论了数据维度、嵌入空间维度和数据复杂度对可视化精度的影响。
Multidimensional scaling addresses the problem of representation of objects specified by proximity data by points in low dimensional embedding space. The problem is reduced to optimization of an accuracy measure of fit of the proximity data by the distances between the respective points. Three‐dimensional embedding space is considered in the present paper. Images of data of different dimensionality are discussed as well as dependence of visualization accuracy on dimensionality of embedding space and complexity of data.