A Guided Tour through the Data Mining Jungle

Animportantsuccessfactor for the field of KDDlies in the developmentand integration of methodsfor supporting the construction and execution of KDDprocesses. Crucial aspects in this context are the (incremental)developmentof a precise problemdescription, a decompositionof this top level problemdescription into manageableand compatible subtasks which can be reused, and a selection and combination of adequate algorithms for solving these subtasks. In this paper wedescribe an approachfor supporting the systematic decomposition of a KDDprocess into subtasks and for selecting appropriate problem-solving methodsand algorithms for solving these subtasks. Ourapproachhas been partially integrated into the CLEMENTINEsystem and has been used to develop a real worldapplication in the area of prediction. Keywords : KDD-processes, User-support, Task Decompositions, KnowledgeAcquisition

• Type: Conference
• Year: 1997
• Where: KDD
• Authors: Robert Engels, Guido Lindner, Rudi Studer
• Keywords: Data Mining, Developmentand Integration, KDD 1997, Manageableand Compatible Subtasks, Problem-solving Methodsand Algorithms