基于相关性分析和遗传算法的属性选择

属性的选择和评价是知识基系统设计中的重要任务和影响系统性能优劣的关键因素。为此，利用遗传算法的遗传算子搜索机制和相关性分析的启发式作为评价机制，提出一种新颖的属性选择策略，用于从属性集中选择给定案例最优的属性子集。实验结果表明，该方法可以确定与分类和预测最相关的属性子集，同时在几乎不降低分类准确性的情况下，极大地减小属性的表示空间。

Attributes Selection Based on Correlation Analysis and Genetic Algorithm

It is an important task for knowledge-based systems to select and evaluate the attributes as well as a critical factor affecting systems’ performance. Using the genetic operator of the searching approach and correlation analysis, which characterizes Genetic Algorithm(GA), as the evaluation mechanism, this paper presents a new method to select the optimal subset of attributes for a given case library. Experimental results show that the proposed method can identify the most related subset to classify and predict, while reducing the representation space of the attributes whereas hardly decreasing the classification precision.