Data Mining In The Lmis

The sharp increase in use of computers and related technologies makes acute the "data-rich but knowledge-poor" problem. Researches on this problem are gradually concentrated and eventually give birth to the Knowledge Discovery in Database (KDD).KDD is a newly emerging and multi-disciplinary field of research; machine learning, statistics, database technology, expert systems and data visualization all make a contribution. The primary aim of KDD is to find from a large amount of data the potentially useful information with effective mining approaches. Such information can be used to support commercial decisions, scientific discoveries and information retrievals.This paper first introduces the concept, process and several principal methods of data mining as well as some application fields of KDD. With the thorough analysis on the algorithm of decision tree induction, we established a classification and prediction system based on C4.5 and accomplished the integration with the LMIS system. According to the definition of data source by users, the system can extract data from the database or data warehouse, clean the data, induct a tree and a set of rules and make classification or prediction about the unknown-classed data. We have got a good result from the test run on the LMIS operational database.This thesis is a tentative application on KDD, by which we are able to make an understanding towards the prospect and potential of KDD applications.