Research On ID3 Algorithm And Application In The Data Mining System Of The Government Policy-making

The data mining system of the government policy-making extracts rules and models from data center database and statistical information of the government departments to provide decision support for the government. In order to implement this system, the first step is to pretreat the data which comes from data center database and statistical information of government departments, the second step is to set up data mining models on the pretreated data, the last step is to evaluate and display the mined models and rules.This paper begins with basic theory of data mining. The study finds that the data mining system of the government policy-making needs to build data mining models based on decision tree algorithm, the most successful and greatest impact decision tree algorithm is ID3 algorithm. So, first, this paper focuses on what is Decision Tree's ID3 algorithm, then, by researching on ID3 algorithm we found that it has two shortcomings: 1 , ID3 algorithm can not handle continuous-type of data. 2, ID3 algorithm prefers to the attributes with more options attribute when choosing split attribute. Because of the two shortcomings above, the paper put forward an improved ID3 algorithm. The improved algorithm uses pretreatment processing technology to deal with continuous-type data. In addition, the improved algorithm makes information gains rate as the standard on choosing split attribute. By testing the new algorithm on the training examples produces a decision tree, we can conclude that:1% Improved ID3 algorithm can effectively improve the classification accuracy and make up the shortcoming in choosing split attribute o2n Improved ID3 algorithm is effective in handling continuous-type data.,Finally, this paper applied ID3 algorithm in the data mining system of the government policy-making and achieved good results in practice.