摘要：Mastering disease influence factors promises toadvance clinical research and provides a possible decisionmaking. In this paper, we propose a framework ARB, which isintegrating association rule mining algorithm with baggingalgorithm. ARB consists of two main modules 1) knowledgediscovery and 2) disease diagnosis. Firstly association rulemining algorithm is used to investigate the sick and healthyfactors which contribute to disease for males and females. Thisalso aims to select the most robust and effective features toreduce the dimensions. And then we use ensemble algorithm todiagnose disease based on the data filtered by the first module.The framework ARB applies three real thyroid datasets in UCImachine learning repository. Though the association rulesgenerated by Apriori algorithm, we know thyroid disease havedifferent effects on people of different age intervals, and theelderly from 60 to 80 are the most likely to suffer from thyroiddisease. The results also show that the two age intervals (30, 40
关键词：Thyroid disease; Association rule mining; Apriori algorithm; bagging algorithm
