开放式网络非均匀采样数据准确挖掘仿真

Simulation of Accurate Mining for Non-Uniform Sampling Data in Open Network

为解决传统数据挖掘方法中存在的挖掘时间较长,查准率较低等问题,提出了一种开放式网络非均匀采样数据准确挖掘方法。运用划分技术构建数据划分区域,估计各个划分区域的重要度,获取相对重要的显著性区域边界,在显著性区域内对数据进行随机采样。通过改进型的样本子集优化方法从采样后的数据中选择最具有参考价值的数据作为原型集,分析原型集与训练集样本间的差异性,根据其分析结果构建相应的特征空间。为实现映射到特征空间内的差异度数据集的非均匀采样数据挖掘,需要使用分类预算法进行学习。实验结论为,开放式网络非均匀采样数据准确挖掘方法能够有效解决传统方式中的问题,如减少数据挖掘时间,提高数据挖掘查准率等。

Traditionally,the data mining time is long and the precision rate is low.Therefore,an accurate mining method for non-uniform sampling data in open network was presented.First of all,the partitioning technique was used to construct the data partitioning area and estimate the importance of each area,so that the relatively significant area boundary was obtained.And then,the data in significant area was randomly sampled.The improved sample subset optimization method was used to select the data with reference value from the sampled data as the prototype set.After that,the difference between the sample of prototype set and the sample of training set was analyzed.According to the analysis result,the corresponding feature space was built.Finally,the classification budget method was used to learn the difference data set mapped into the feature space.Based on the learning result,the non-uniform sampling data mining was achieved.Simulation results show that the proposed method can effectively reduce the data mining time and improve the precision rate of data mining.