The rapid development of database technology and the application of database management system made data amount radically increase . The information hidden behind the large amount data may offer support for corporations' decision .The tools of the moment can not acquire information effectively. Data mining is developed to meet the requirement.Data mining is a dynamic progress to hunt for hidderu unsuspected, uncommon, useful information or pattern from large database or data warehouse. It constitute problem definition, data preparation , data raining and result analysis, problem definition and data preparation are very important in a data mining system. Data preparation not only demand the human participation but also may resort to some tools for data removal, data abstraction and data conglomerate to make the data is fit for mining algorithm.The selection of mining algorithm and the building of mining model are the most important part in a data mining system. The selection of raining algorithm need take into account of the efficiency , calculation complexity and the perform of algorithm.A good model need a suitable algorithm selection. Rough set theory as a new mathematical tool is fit for data mining. It consists of information table, equivalent relation, upper/lower approximation and reduction. Decision tree is a most commonly classification algorithm for its higher efficiency and comprehensive result.In this paper , a data mining system is introduced to analysis pump status in order to adjust pump parameters and made suitable scheme . Binning , data histogram, clustering and concept tree are usedto deal with data beforehand . based on the transacted data o , decision tree and rough set are used to build classing model. The mining result is used for decision support. Machine judgment takes the place of human judgment . The system makes judgment progress more rapidly, more correctly, more economically. The good effect is acquired in practical application .
|