Hierarchical Classification Modeling of Watershed Data by Chemical Signatures
Steven D. Brown, Liyuan Chen · ACS symposium series · 2015
Complex data with many objects and classes may benefit from the use of a hierarchical class modeling approach in which samples receive more than one class label. A hierarchical model employing multiple class labels is better suited to making use of class relationships in the data as compared to traditional “flat” modeling methods. However, hierarchical modeling requires a number of choices that make the data analysis much more complex than a traditional classification. This chapter introduces concepts from hierarchical modeling of complex data with multi-label class ontologies, and considers what choices must be made to establish class labels in complex data and build a hierarchical classifier. An example is provided to show the details of the methodology in modeling hierarchical geospatial data.