Comparative Analysis of Multi-fractal Data Missing Processing Methods
Comparative Analysis of Multi-fractal Data Missing Processing Methods
复制标题
多重分形数据缺失处理方法对比分析
DOI：
10.11648/j.acm.20190802.14
复制
发表时间：
2019-07
影响因子：
10
通讯作者：
Zeng Xiangjian
中科院分区：
文献类型：
--
作者：
Lai Simin;Wan Li;Zeng Xiangjian
数据丢失通常会影响序列的特性。采用合适的方法对缺失数据进行处理是获取高质量信息的前提和保证。在这项研究中，提出了一种用自相似特征序列填充缺失数据的分形内插方法。以参数为0.25和0.35的两组二项多重分形序列为研究对象，利用MF-DMA计算了填充处理后序列的Hurst指标值，验证了分形内插填充方法的实用性。同时，将该方法应用于缺失率分别为10%、15%和20%的多重分形序列，并与删除法和随机填充法的填充效果进行了比较，得出了这三种方法的适用性。结果表明，对于不同缺失率的二项重分形序列，经分形插值处理后的序列的Hurst指数与理论值的拟合度最高，其修复分形序列的效果好于其他两种方法，具有良好的应用前景。
Data missing often affects the characteristics of the sequence. Using appropriate methods to process the missing data is the premise and guarantee to obtain high quality information. In this study, a fractal interpolation method is proposed to fill the missing data with self-similar feature sequences. Two sets of binomial multifractal sequences with parameters of 0.25 and 0.35 are taken as the research objects, and the Hurst index value of the sequence after filling processing is calculated by MF-DMA, which verifies the practicability of the fractal interpolation filling method. At the same time, the method is applied to multi-fractal sequences with missing rates of 10%, 15% and 20% respectively, and compared with the filling effects of deletion method and random filling method, then, the applicability of the three methods is obtained. The results show that, for binomial multifractal sequences with different missing ratios, the Hurst index of the sequence processed by fractal interpolation has the highest degree of fitting with the theoretical value, its effect of repairing the fractal sequence is better than the other two methods, and has a good application prospect.