Hot-rolled Data Analysis Based On Decision Tree Learning And Rules Extraction

Knowledge Discovery in Database(KDD) is a whole process that convert the raw data into useful information, and data mining is an indispensable part of it. Data Mining is a multidisciplinary field of interdisciplinary research, it can combine traditional data analysis methods with complex algorithm of deal with a large amount data, and it provides exciting opportunities for the exploration and analysis of new data types and new methods to analysis old data types. Classification is a widely used one branch of the field in data mining algorithms. An event or a group of objects are classified by it. Through the training data set of mining classification model available, classification model used to analyze existing data; it can also use to predict the future trends. Decision tree is a typical classification algorithm, it can reason out classification rules of decision tree form from one group no order and on rules cases.OLE DB for DM norm is a set of API that it is proposed by Microsoft for the data mining market to provide an industry standard, Database application developers can structure data mining models in their databases by OLE DB for DM norm, they use these models to complete a large amount of prediction and analysis mission, and these models can be shared with other applications.This paper analysis continuous hot-rolled mills data, using SQL Server data mining solutions to generate decision trees by combining with Visual C++ 6.0 and OLE DB for DM norm, a storage model of decision tree rules based on relational database and the rule extraction method bring forward accordance with the characteristics of decision-making production rules itself. The methods of application model and the process of data mining system in the paper are very useful to design and to implement data mining based on SQL Server.