Knowledge Discovery in a Water Quality Database

Weapply rule induction to mine for knowledgein a database which stores data obtained by monitoring the water quality of the rivers in Slovenia. The database contains measurementdata about the physical and chemicalproperties of the water at different measurementsites, as well as data about the presence of living organisms. Takentogether, the abovedata reflect the quality of the water: the physicaland chemical properties influencethe living organisms,whichin turn give an overall picture of the waterquality over a period of time. Weaddress twoproblems: (a) analysis of the influenceof physicaland chemicalindicators of waterquality on the presenceof selected bioindicators, and(b) classification into quality classes based on either bioindicator or physical and chemicalindicator data. Thelearned rules are evaluatedby a river biologyexpert, but also in terms of their performance on unseencases.

• Type: Conference
• Year: 1995
• Where: KDD
• Authors: Saso Dzeroski
• Keywords: Data Mining, Database Contains Measurementdata, KDD 1995, Rule Induction, Water Quality