Improved Decision Tree Algorithm Based on Decision Attribute Selection Strategy
Sun Ke · Journal of Shenyang Normal University · 2011
This paper firstly introduces basic principle and modeling process of decision tree algorithm and introduces traditional decision attributes selection strategy. Secondly it introduces ID3 algorithm and summarizes the advantage and disadvantage of decision tree algorithm.Based on traditional ID3 algorithm,we analyse previous decision attributes selection strategy and improve decision attribute selection strategy.We propose the strategy based on correlated information gain.The method improves traditional ID3 algorithm in multiple-valued bias with correlated attributes.Finally,we apply the method in UCI data sets and analyse classification precision of improved ID3 algorithm and traditional ID3 algorithm.Experiment shows that E-ID3 algorithm can get higher precision than traditional ID3 algorithm.