围网渔情预报中强影响因子的挖掘技术研究

Research on mining technique of high effect factors in purse seine fishery prediction

针对传统围网渔业渔情预测方法的缺点,综合多种类型海洋环境因子,采用粗糙集理论中的属性约简方法,获得多种类型因子中的约简属性,即影响围网产量的强影响因子。该技术首先对渔情监测数据进行缺失值的填补,再利用可辨识矩阵进行属性约简,从而构建出强影响因子的核心属性集。该算法有效解决了渔情监测数据的稀疏性问题,提高了渔情预测的准确性。

Firstly, a new algorithm based on attribute frequency in the discernibility matrix is used to get the core-attribute of attribute reduction. Secondly, considering the effect of different kinds of marine environment factors, an effective prediction model is established to confirm the core-attribute to be the high effect factors of purse seine outputs. This method addresses the issue by automatically filling va- cant item of the fishery monitor data set, and then to take a attribute deduction using the discernibility matrix to get the core-attribute to be the high effect factors of purse seine. The experiment results show that the algorithm efficiently improves sparsity of date set , and promises to make prediction more accu- rately