Study On Application Of Data Generalization Based On Attribute-Oriented Induction

Data mining is the non-trivial process of identifying valid, novel, potentially useful, and ultimately understandable patterns from huge volume of data. The resultant rules of Data Mining include descriptive rules and predictive rules. Characterization rule is one of the typical descriptive rules, data generalization is a kind of main methods of descriptive rules mining, and attribute-oriented inducetion is an important technique for data generalization and mining characterization rules.The level of structure of notion makes given data discretization, the collections counteracts the notion of the higher level of structure to substitute the low level notion, making the data steer the absorption;The approval turn to make data more meaningful all, more easy hermeneuticCan define several notions level of structure for the same of attribute, with different the demand of consumer in adaptation.This thesis researches attribute-oriented induction based on concept, analyses the definition, presentation and evaluation of concept hierarchy, designs the mining algorithms and implementation mechanism of data generalization and characterization rules, and proposes the structure framework of data generalization and characterization rules mining. Based on above results, this thesis analyses the requirements about data generalization and characterization rules in real application, and expounds the significance of generalizing detail data and obtaining characterization rules. As a case study, this thesis processes an agriculture production database using statistic analysis and attribute-oriented induction method for data generalization and characterization rules mining. The rules discovered can provide assistant decision support for agriculture production decision.