An optimized approach to generate simplified decision trees

With the development of computer technology and computer network technology, the degree of information is getting higher and higher, people´s ability of using information technology to collect and produce data is substantially enhanced. The discovery of the optimal algorithms for mining useful knowledge and improving the effectiveness of information utilization are problems need to be addressed urgently. It was under this background that Data Mining (DM) technology came into being and developed. Data mining is a process to extract information and knowledge from a large number of incomplete, noisy, fuzzy and random data. In these data, the information and knowledge are hidden, which people can´t discover at present, but potentially useful. In recent years, the decision tree has become an important data mining method. The basic learning strategy of decision tree is divide and conquer technique, which uses from root to leaf of the decision tree structure. This paper emphasizes to propose a new decision tree model based on multivariate statistical method Principal Component analysis on multi-attribute data for reducing dimensionality and to transform traditional decision tree algorithm to form a new algorithmic model. The experiments demonstrate that this method can not only optimizes the structure of the decision tree, but also overcomes the problems existing in pruning and to mine the better rule set without effecting the purpose of prediction accuracy altogether.