Abstract: Bayesian network classifier with precise structure has been proven to be NP-hard problem. A Bayesian network classifier based on Particle Swarm Optimization-Predatory Escape (PSO_PE) algorithm was proposed in this paper, which could effectively avoid the direct influence of feature reduction on the performance of classification and complete the precise learning Bayesian network. In addition, the proposed classifier was exploited in employment predication of vocational college and was experimentally tested on Weka. The experimental results show that compared with other Bayesian classifiers, the new classifier is more effective and precise to learn Bayesian network.

Key words: predatory escape, Particle Swarm Optimization (PSO), Bayesian Network Classifier(BNC), Weka, employment predication

摘要： 构造精确的贝叶斯网络分类器已被证明为NP难问题,提出了一种基于捕食逃逸粒子群优化(PSO)算法的通用贝叶斯网络分类器,能有效避免数据预处理时的属性约简对分类效果的直接影响,实现对贝叶斯网络结构的精确学习和搜索。另外,将所提出的分类器应用于高职院校就业预测分析,并在Weka平台上实现对该分类器的构建和验证,与其他几种贝叶斯网络分类器的对比实验结果表明,该分类器具有更好的性能。

关键词: 捕食逃逸, 粒子群优化, 贝叶斯网络分类器, Weka, 就业预测