Partitioning algorithms and combined model integration for data mining (Q1861614): Difference between revisions

A strategy of huge data sets analysis (data mining, DM) is proposed which consists of three stages: partitioning stage, explanatory stage and decision stage. The explanatory trees technique of clustering is described for the partitioning (EAST splitting algorithm is recommended for DM). Generalized additive multi-mixture models are proposed for the explanatory stage. The final decision rule combines clustering and explanation. Application to the data on 48 842 individuals from US Census is considered.