Multivariate data may not be a chaotic digital dune. In Cartesian coordinates, the geometric relationship between the data is clearly illustrated. Professor WANG Jinfeng and his colleagues, Institute of Geographic Sciences and Natural Resources Research (IGSNRR) developed a new reference system to present data: the evolution tree, in which evolution relationship and mutation hidden in data is mined out and displayed in a visual form, life structure of multivariate data is in full view.

The evolution tree was applied in real data. 253 cities in China  were clustered into seven function types and six developmental stages respectively by using social-economic statistic data of the cities. The seven function types were arranged onto seven stems of a tree structure and the six development stages in each function type were mapped in order onto six branches of the function type stem, individual cities are denoted by the tree leaves. In such way, a cities evolution tree is reconstructed, the evolution of China's urban system was visualized, cites with different function types and developmental stages develop in their own paths along the tree’s stem and branches. The evolution tree has been applied to two cases: (1) visual expression of position, context, direction and developmental process of the cities in China; (2) understanding and prediction of urbanization sprawl, with higher precision.

More research results have been published in Population and Environment. 2011. DOI: 10.1007/s11111-011-0142-4.