U-Estimators of Regression Coefficients

✍ Scribed by G.G. Gregory

• Publisher: Elsevier Science
• Year: 1993
• Tongue: English
• Weight: 390 KB
• Volume: 47
• Category: Article
• ISSN: 0047-259X
• DOI: 10.1006/jmva.1993.1068

✦ Synopsis

For the regression model (y_{i}=x_{i}^{\prime} \xi+e_{i}, 1 \leqslant i \leqslant n), with i.i.d. residuals (\left{e_{i}\right}), we introduce the estimator of (\xi) which zeros the weighted (U)-statistic (\sum \sum q_{i j} K\left(\hat{e}{i}, \hat{e}{j}\right)), where (q_{i j}) is a score vector for regression vectors (x_{i}) and (x_{j}). These include some (M) - and (R)-estimators. Asymptotic inference is developed without the need to estimate the (\left(f^{\prime} / f\right)) function, where (f) is the pdf of the residuals. 1993 Academic Press, Inc