基于改进的加权贝叶斯分类算法在空间数据中的应用

Application of Improved Weighted Bayesian Classification Algorithm in Spatial Data

朴素贝叶斯算法是一种简单而高效的分类算法,但它的属性独立性假设,影响了它的分类性能。针对这个问题,提出一种基于属性约简的PLS加权朴素贝叶斯分类算法。该算法首先分析属性之间的相关性,通过属性约简选择一组近似独立的属性约简子集,提出改进的偏最小二乘回归加权朴素贝叶斯分类算法,实验结果表明,改进算法具有较高的分类准确度。并将改进的算法应用于边坡识别问题中。

Naive bayesian algorithm is a simple and effective classification algorithm, but its attribute independence hypothesis, influence its classification performance. According to this problem, the paper proposes a kind of attribute reduction based on PLS weighted simple bayesian classification algorithm. This algorithm firstly analyzes the relationship between attribute, through attribute reduction choose a set of approximate independent attribute reduction subset, put forward the improvement of the partial least-squares regression weighted simple bayesian classification algorithm, experimental results show that the improved algorithm has higher classification accuracy. And the improved algorithm is applied to slop identification.