Traditional associative classification algorithm can not accurately set support and confidence threshold according to the scale of the problem , which leads to the performance of the classifier affected by human factors .To resolve the issue , we propose an intelligent optimi-sation idea-based associative classification algorithm .The algorithm improves the CBA associative classification algorithm and makes use of the good ability of simulated annealing in global search to optimise the support and confidence threshold in solution space so as to achieve global optimum in classification accuracy rate .Experiments show that this method can effectively prevent the unreasonable setting of the threshold from the disadvantage of impacting classification effect and enables more accurate classification performance compared with tradition -al associative classification algorithm .%针对传统关联分类算法中支持度和置信度阈值无法根据问题规模准确设定,导致分类器的分类效果受人为因素影响的缺陷,提出一种基于智能优化思想的关联分类算法。该算法对CBA关联分类算法进行改进,利用模拟退火技术良好的全局搜索能力在解空间内对支持度和置信度阈值进行优化,从而使分类准确率达到全局最优。实验表明,与传统的关联分类算法相比,该方法可以有效地避免阈值设置不合理而影响分类效果的弊端,使分类结果更加精准。
展开▼