On the choice of bandwidth for kernel graduation
- 1 January 1994
- journal article
- research article
- Published by Cambridge University Press (CUP) in Journal of the Institute of Actuaries
- Vol. 121 (1) , 119-134
- https://doi.org/10.1017/s0020268100020102
Abstract
This paper considers cross-validation as an objective and risk-based method for selecting the smoothing parameter in a non-parametric graduation. In addition, the relative merits of two kernel estimators are compared in the context of mortality graduation. Finally, it is well known in the statistical literature that the use of theoretically superior kernels is not as important as the choice of bandwidth. Our results support this conclusion, suggesting that the focus on such weights is misguided in the actuarial textbooks on moving weighted averages.Keywords
This publication has 26 references indexed in Scilit:
- Local Regression: Automatic Kernel CarpentryStatistical Science, 1993
- Design-adaptive Nonparametric RegressionJournal of the American Statistical Association, 1992
- A Geometrical Method for Removing Edge Effects from Kernel-Type Nonparametric Regression EstimatorsJournal of the American Statistical Association, 1991
- The Kernel Estimate of a Regression Function in Likelihood-Based ModelsJournal of the American Statistical Association, 1989
- The Kernel Estimate of a Regression Function in Likelihood-Based ModelsJournal of the American Statistical Association, 1989
- How Far Are Automatically Chosen Regression Smoothing Parameters From Their Optimum?Journal of the American Statistical Association, 1988
- Graduation: some experiments with kernel methodsJournal of the Institute of Actuaries, 1987
- Non-parametric graduation using kernel methodsJournal of the Institute of Actuaries, 1983
- Robust Locally Weighted Regression and Smoothing ScatterplotsJournal of the American Statistical Association, 1979
- Robust Locally Weighted Regression and Smoothing ScatterplotsJournal of the American Statistical Association, 1979