Data Mining Technology Applied And Discussed In Packaging Enterprise Of Business Management

The rapid development in the field of computer network and database system is continually upgrading the conflict between the power to supply all kind of data and the ability to analyze them. To relax this situation, data mining technology came onto the stage.Client analysis is the important development in the field of client manager. Knowledge acquisition is the bottleneck of intelligent diagnosis. The manually extracted disgnostic rules are often featured with vagueness and imprecision. With the dramatic development in data collection and storage in the recent decade, it is getting more and more difficult in discovering knowledge from these data mountains manually. "We are drawing in information, but starving for knowledge".To deal with this challenge, data mining technology is studied and applied to knowledge discovery in engineering diagnosis in this thesis.First, data mining and relevant techniques are discussed. A data mining processing model, so-called rapid prototype processing (RPP) model is proposed. RPP model compresses the whole data mining process into 4 steps, so that we can obtain the mining result as soon as possible. Accordingly, the designed data mining algorithms or collected data set can be adjusted according to the mining result. A lot of time that waste in looking for proper algorithms or data sets may be saved.Then, based on RPP model, data mining is applied to client analysis. The binary decision tree algorithm are studied., the upper bound of its VC dimension was calculated and concluded that: upper bound of the VC dimension of binary decision tree algorithm rises as the node number of its result (binary decision tree) increases. As a supplement, upper bound of the VC dimension of a single univariate non-leaf node was also calculated. Based on diversity aim, the paper put forward the neural network model, and analysis two differ algorithms there are BP algorithm and Levenberg-Marquardt algorithm. Through experiment, it is compared to the two algorithms.Finally, the paper analysis client information. The system architecture and structure is proposed and some practical problems that will be met in the system implementation are discussed in detail. The system will provide a robust platform model for client analysis data mining.