作者: Patrice Loisel , Brigitte Charnomordic , Hazaël Jones , Bruno Tisseyre
DOI: 10.1007/S11119-018-9584-3
关键词: Data mining 、 Decision support system 、 Plot (graphics) 、 Zoning 、 Complementarity (molecular biology) 、 Computer science 、 Set (abstract data type) 、 Quantile 、 Field (computer science) 、 sort
摘要: The paper proposes a numerical criterion to evaluate zoning quality for given number of classes. originality the is simultaneously quantify how zones are heterogeneous on whole field under study and neighbouring similar. This approach allows comparison between maps either with different or labels, which importance zone delineation algorithms aiming at maximizing inter-zone variability. In addition, this also an optimisation procedure that yields interpretable within-field in each assigned clear label. involves contour based quantile values. key point use proposed guide showing complementarity both proposals delineating relevant zones. order demonstrate relevancy criterion, implementation together, method was tested 50 theoretical fields known variability spatial structure. A real plot yield monitoring data used value case. Results show methodology compare sort them. interest provide ranked set possible may constitute decision support practitioners who consider additional expert information choose most appropriate map specific conditions consideration.