摘要
朴素贝叶斯算法是一种简单而高效的分类算法,但是它的条件独立性假设影响了其分类性能。通过放松朴素贝叶斯假设,可以增强其分类效果,但通常会导致计算代价大幅提高。文章提出了一种基于偏最小二乘的加权朴素贝叶斯分类算法,通过建立条件属性和决策属性之间偏最小二乘回归方程,把回归系数赋给对应的条件属性,作为相应的权重,从而在保持简单性的基础上有效地提高了朴素贝叶斯算法的分类性能。最后,通过在UCI数据集上的仿真实验,验证了该算法的有效性。
Naive Bayes algorithm is a simple and effective classification algorithm.However,its classification performance is affected by its conditional attribute independence assumption.Many techniques have explored the basic assumption of Naive Bayes to increase accuracy,which always cost much more computing time.In this article,proposes a weighted Naive Bayes classification algorithm based on partial least squares （PLS）.By setting a regression equation of PLS between condition attributes and decision attribute,different condition attributes are weighted by corresponding regression coefficients.Then,classification performance can be improved effectively and simply.At last,simulation results on a variety of UCI data sets illustrate the efficiency of this method.
出处
《电子质量》
2010年第7期4-6,共3页
Electronics Quality
