Penalized Estimation of Free-Knot Splines
- 1 June 1999
- journal article
- research article
- Published by Taylor & Francis in Journal of Computational and Graphical Statistics
- Vol. 8 (2) , 333-352
- https://doi.org/10.1080/10618600.1999.10474817
Abstract
Polynomial splines are often used in statistical regression models for smooth response functions. When the number and location of the knots are optimized, the approximating power of the spline is improved and the model is nonparametric with locally determined smoothness. However, finding the optimal knot locations is an historically difficult problem. We present a new estimation approach that improves computational properties by penalizing coalescing knots. The resulting estimator is easier to compute than the unpenalized estimates of knot positions, eliminates unnecessary “corners” in the fitted curve, and in simulation studies, shows no increase in the loss. A number of GCV and AIC type criteria for choosing the number of knots are evaluated via simulation.Keywords
This publication has 17 references indexed in Scilit:
- Automatic Bayesian Curve FittingJournal of the Royal Statistical Society Series B: Statistical Methodology, 1998
- Additive Nonparametric Regression With Autocorrelated ErrorsJournal of the Royal Statistical Society Series B: Statistical Methodology, 1998
- Polynomial splines and their tensor products in extended linear modeling: 1994 Wald memorial lectureThe Annals of Statistics, 1997
- Hybrid Adaptive SplinesJournal of the American Statistical Association, 1997
- Hazard RegressionJournal of the American Statistical Association, 1995
- Flexible Parsimonious Smoothing and Additive ModelingTechnometrics, 1989
- A Statistical Perspective on Ill-Posed Inverse ProblemsStatistical Science, 1986
- Logistic Regression DiagnosticsThe Annals of Statistics, 1981
- Asymptotic Integrated Mean Square Error Using Least Squares and Bias Minimizing SplinesThe Annals of Statistics, 1980
- Fitting Segmented Polynomial Regression Models Whose Join Points Have to Be EstimatedJournal of the American Statistical Association, 1973