小样本采样数据的预处理

一般情况下数据处理大多采用数理统计方法,该方法对于数据较少情况下处理和判别粗大误差不适用。提出了运用线性均方估计法和熵值判别法来处理和判别粗大误差,线性均方估计消除粗大误差是一种采用软化的方法处理粗大误差;熵值判别法是根据熵的上界对应最大的不确定度,利用所得数据的熵信息量判别数据是否含有粗大误差。这两种方法经过多个实例计算,结果表明,它们在处理小样本采样数据时更有效。

Preprocessing of small sampling data

The method of mathematical statistic is usually used to process the data, but it is not suitable to process small quantity of data. Thus a new method based on the linear mean squares estimation and entropy judging method is used to distinguish the gross errors. LMSE is a way by softening the gross errors; the entropy distinguishing is based on the upper limit that is maximum error of the entropy, thus it can he used to distinguish the gross errors of data. Many examples prove that the two methods are more effective when they are used to process small sampling data.