在随机损失数据情况下回归函数的Partitioning估计的强相合性

Strong Consistency of Partitioning Estimates of A Reqression Function for Randomly Missing Date

本文在不完全数据的情况下采用Ｌ１模得到了回归函数的Ｐａｒｔｉｔｉｏｎｉｎｇ估计的强相合性．我们的结果只要求回归函数的Ｐ（Ｐ＞１）阶矩有限，对回归函数没加任何光滑条件，并且我们的结果对任何分布都成立．

The strong consistencies in L1-Norm of Partitioning estimate of a reqression unction are shown for incomplete sample. Here incomplete sample means that some samplepoints are missing according to a random missing mechanism operated on the observationvectors, and our results only ask the finity of pth-moment (for some p>1) and some conditions on random missing mechanism operator, do not ask any smooth condition on regressionfunction and any condition on distribution.