论文部分内容阅读
提出了以最小二乘逼近方法为基础的数据野值判别与剔除算法。利用TBS(三角B样条)曲线同时具有局部性和整体性的优越性构造最小二乘拟合算法,并结合偏度分析与残量分析误差方法,在给定范数意义下的评价系统中,可以得到TBS-LS(最小二乘三角B样条)拟合曲线,从而可以更好地识别并剔除野值。最后给出算法以及主要结果,通过实例说明方法的有效性。
This paper proposes a data outlier identification and elimination algorithm based on the least square approximation method. Least Squares Fitting Algorithm is constructed by the advantage of TBS (Triangular B-spline) curve with both locality and integrity. Combining the methods of skewness analysis and residual error analysis, in the evaluation system under the given norm , A TBS-LS (Least Squares Trigonometric B-spline) fitting curve can be obtained, so that the outliers can be better identified and removed. Finally, the algorithm and the main results are given, and the effectiveness of the method is illustrated by examples.