Research And Improvement On The Decision Tree Classification Algorithm Of Data Mining

Classification analysis is a very important method in data mining researching area. In recent years, a new generation of data is having been created by the rapid development of information technology, which has similar characteristics: High-speed, high-dimensional, continuous, dynamic, rapidly changing and mass. So how can use these large and complex data sets to serve our community effectively and rationally, which has become the problem that should be solved in the classification field of data mining. Facing these high-dimensional, continuous, dynamic and complex data, the traditional classification algorithms are no longer effective to carry out these data. Therefore, to improve the traditional classification algorithms should be able to be dealt with these questions according to the characteristics of complex data. So to make the improved traditional classification algorithm of decision tree can deal with high-dimensional, continuous, dynamic and complex data. At the same time, to improve the computational efficiency of the decision tree feature selection, to increase the classification accuracy of decision tree classification algorithm, to improve the overall classification performance of decision tree.This paper introduces the data mining theory,related technologies, the core concepts and processes of classification algorithms, which focused on the research and analysis of decision tree classification algorithm. This method is proposed by the shortcomings and deficiencies of traditional decision tree classification algorithm in dealing with complex data. Firstly, the paper quoted the existing high-dimensional reduced-dimensionality techniques solve the high-dimensional data; secondly, this paper gives an improved algorithm—New-BMIC discrete algorithms based on the BMIC discrete algorithm which can process continuous attribute data; finally, the paper gives the variance of the statute of the properties of the selection criteria, which can solve the problem of low computational efficiency. Last, this paper enhances the overall performance of decision tree classification algorithm by optimizing the three parts. Experimental results show that the new algorithm is better than other decision tree classification algorithm.