Research And Implementation Of Data Mining Classification System

With the great development of computer technology, the increasing scale of information, the organizations of information are becoming more and more complicated. Caused by the explosion of information, human is facing an overloading dilemma of information that we are flooding by information, but starving of knowledge. Therefore, data mining, as a technology of knowledge discovery from massive information, is becoming increasingly important, and it is attracting more and more researcher. Among large number of data mining technologies, classification is an important basic technology, which has great value of research and application.This paper conducts research and implementation of algorithm for the tasks of data classification, and develops a universal data classification system. Therefore, this article completes two mainly tasks:First, based on four classification model including Decision Tree, K-Nearest Neighbor, Naive Bayes and Feedforward Neural Network, this paper implements an integrated classification model of voting strategies. The results of a series of comparing experiments on multiple data sets show that the integrated model has better classification performance than four separate models.Second, this article implements a data classification system including four common classification algorithm and integrated classification algorithm. The system achieves four main functions:data preprocessing, data classification, the effect evaluation of classification, the visual display of the result. Data preprocessing operation includes filling missing values, smooth noisy data and attribute normalized; the evaluation of the classification effect mainly takes cross-validation methods; the visualization of result allows users taking exploration actively, and it is possible to discover unexpected knowledge in the process of exploration. The results of software testing show that the system can meet the classification needs of the majority structured data set.