回归函数的小波估计

Wavelets Detection of Jump Points in regressiong model

断点识别是统计上研究的一类重要问题 ,文章利用小波变换在奇异点附近的模局部极大性特点 ,用小波方法作误差独立模型的回归函数的断点识别问题 ,给出了断点个数 ,位置及跃度的相合估计 ,数值模拟表明了小波方法较传统方法更精确。

Detection of Jump Points is a kind of important problems in statistic area. In this paper, we propose a mothed for the Detection of Jump Points in i.i.d Regressiong Model by wavelets. This mothed is different from Traditional kernel estimate. The procedure is simple and intuitive. Simulation sugests the method is functional.