Geometrically designed, variable knot regression splines: variation diminish optimality of knots

A new method for Computer Aided Geometric Design of variable knot regression splines, named GeDS, has recently been introduced by Kaishev et al. (2006). The method utilizes the close geometric relationship between a spline regression function and its control polygon, with vertices whose y-coordinates are the regression coefficients and whose x-coordinates are certain averages of the knots, known as the Greville sites. The method involves two stages, A and B. In stage A, a linear LS spline fit to the data is constructed, and viewed as the initial position of the control polygon of a higher order (n > 2) smooth spline curve. In stage B, the optimal set of knots of this higher order spline curve is found, so that its control polygon is as close to the initial polygon of stage A as possible, and finally the LS estimates of the regression coefficients of this curve are found. In Kaishev et al. (2006) the implementation of stage A has been thoroughly addressed and the pointwise asymptotic properties of the GeD spline estimator have been explored and used to construct asymptotic confidence intervals. In this paper, the focus of the attention is at giving further insight into the optimality properties of the knots of the higher order spline curve, obtained in stage B so that it is nearly a variation diminishing (shape preserving) spline approximation to the linear fit of stage A. Error bounds for this approximation are derived. Extensive numerical examples are provided, illustrating the performance of GeDS and the quality of the resulting LS spline fits. The GeDS estimator is compared with other existing variable knot spline methods and smoothing techniques and is shown to perform very well, producing nearly optimal spline regression models. It is fast and numerically efficient, since no deterministic or stochastic knot insertion/deletion and relocation search strategies are involved.

[1]  Frank P. A. Coolen,et al.  Guidelines for corrective replacement based on low stochastic structure assumptions , 1997 .

[2]  Martin Newby,et al.  Optimal overhaul intervals with imperfect inspection and repair , 1998 .

[3]  Steven Haberman,et al.  On the forecasting of mortality reduction factors , 2003 .

[4]  Jianqing Fan,et al.  Data‐Driven Bandwidth Selection in Local Polynomial Fitting: Variable Bandwidth and Spatial Adaptation , 1995 .

[5]  R. Gerrard,et al.  The management of de-cumulation risks in a defined contribution environment , 2005 .

[6]  P. D. England,et al.  Addendum to “Analytic and bootstrap estimates of prediction errors in claims reserving” , 2002 .

[7]  Robert G. Cowell,et al.  Sampling without replacement in junction trees , 1997 .

[8]  Steven Haberman,et al.  The fair valuation problem of guaranteed annuity options: The stochastic mortality environment case , 2006 .

[9]  M. I Owadally.  Efficient asset valuation methods for pension plans , 2003 .

[10]  Patricia L. Smith,et al.  Curve fitting and modeling with splines using statistical variable selection techniques , 1982 .

[11]  Adrian F. M. Smith,et al.  Automatic Bayesian curve fitting , 1998 .

[12]  Robert G. Cowell,et al.  Mixture reduction via predictive scores , 1998, Stat. Comput..

[13]  Steven Haberman,et al.  The income drawdown option: quadratic loss , 2004 .

[14]  S. Wood.  Thin plate regression splines , 2003 .

[15]  I. Johnstone,et al.  Ideal spatial adaptation by wavelet shrinkage , 1994 .

[16]  Robert G. Cowell,et al.  The Quest for a Donor: Probability Based Methods Offer Help. , 2005 .

[17]  Jianhua Z. Huang.  Local asymptotics for polynomial spline regression , 2003 .

[18]  Steven Haberman,et al.  Optimal strategies for pricing general insurance , 2007 .

[19]  S. Iyer,et al.  Application of stochastic methods in the valuation of social security pension schemes , 2003 .

[20]  M. C. Jones,et al.  Spline Smoothing and Nonparametric Regression. , 1989 .

[21]  Xiaotong Shen,et al.  Spatially Adaptive Regression Splines and Accurate Knot Selection Schemes , 2001 .

[22]  C. R. Deboor,et al.  A practical guide to splines , 1978 .

[23]  D. Jupp.  Approximation to Data by Splines with Free Knots , 1978 .

[24]  Paul H. C. Eilers,et al.  Flexible smoothing with B-splines and penalties , 1996 .

[25]  Vladimir K. Kaishev,et al.  Modelling the joint distribution of competing risks survival times using copula functions , 2007 .

[26]  J. Friedman,et al.  FLEXIBLE PARSIMONIOUS SMOOTHING AND ADDITIVE MODELING , 1989 .

[27]  H. Wynn,et al.  Maximum entropy sampling and optimal Bayesian experimental design , 2000 .

[28]  Vladimir K. Kaishev,et al.  Excess of loss reinsurance under joint survival optimality , 2006 .

[29]  Jerome H. Friedman.  Multivariate adaptive regression splines (with discussion) , 1991 .

[30]  Vladimir K. Kaishev,et al.  Geometrically Designed, Variable Knot Regression Splines: Asymptotics and Inference , 2006 .

[31]  C. D. Boor,et al.  Least Squares Cubic Spline Approximation, II - Variable Knots , 1968 .

[32]  Vladimir K. Kaishev,et al.  Optimal retention levels, given the joint survival of cedent and reinsurer , 2004 .

[33]  D. Ruppert.  Selecting the Number of Knots for Penalized Splines , 2002 .

[34]  Mary J. Lindstrom,et al.  Penalized Estimation of Free-Knot Splines , 1999 .

[35]  Paola Sebastiani,et al.  Learning Bayesian Networks from Incomplete Databases , 1997, UAI.

[36]  M. I. Owadally,et al.  Pension Funding and the Actuarial Assumption Concerning Investment Returns , 2003, ASTIN Bulletin.

[37]  C. Biller.  Adaptive Bayesian Regression Splines in Semiparametric Generalized Linear Models , 2000 .

[38]  Paola Sebastiani,et al.  The Use of Exogenous Knowledge to Learn Bayesian Networks from Incomplete Databases , 1997, IDA.

[39]  R. Kohn,et al.  Nonparametric regression using Bayesian variable selection , 1996 .

[40]  G. Wahba.  Spline models for observational data , 1990 .

[41]  Nan Wang,et al.  Guarantees in With-Profit and Unitized With-Profit Life Insurance Contracts: Fair Valuation Problem in Presence of the Default Option , 2006 .

[42]  B. Yandell.  Spline smoothing and nonparametric regression , 1989 .

[43]  Steven Haberman,et al.  Valuation of Guaranteed Annuity Conversion Options , 2003 .

[44]  Martin Newby,et al.  Moments and generating functions for the absortion distribution and its negative binomial analogue , 1999 .

[45]  Paola Sebastiani,et al.  Bayesian Experimental Design and Shannon Information , 1997 .

[46]  W. J. Studden,et al.  Asymptotic Integrated Mean Square Error Using Least Squares and Bias Minimizing Splines , 1980 .

[47]  P. England,et al.  Analytic and bootstrap estimates of prediction errors in claims reserving , 1999 .

[48]  M. Zaki Khorasanee,et al.  A cash-flow approach to pension funding , 2002 .

[49]  Yingkang Hu,et al.  An algorithm for data reduction using splines with free knots , 1993 .

[50]  David Ruppert,et al.  Theory & Methods: Spatially‐adaptive Penalties for Spline Fitting , 2000 .

[51]  Steven Haberman,et al.  Lee-Carter mortality forecasting, a parallel GLM approach, England & Wales mortality projections , 2002 .

[52]  Linda C. Wolstenholme,et al.  A characterisation of phase type distributions , 1997 .

[53]  Thomas C.M. Lee,et al.  Regression spline smoothing using the minimum description length principle , 2000 .

[54]  T. J. Rivlin,et al.  The optimal recovery of smooth functions , 1976 .

[55]  Michel Denuit,et al.  Lee-Carter Goes Risk-Neutral , 2006 .

[56]  Young K. Truong,et al.  Polynomial splines and their tensor products in extended linearmodeling , 1997 .

[57]  Trevor Hastie,et al.  [Flexible Parsimonious Smoothing and Additive Modeling]: Discussion , 1989 .

[58]  G. Wahba,et al.  Hybrid Adaptive Splines , 1997 .

[59]  Nan Wang,et al.  Modelling and Valuation of Guarantees in With-Profit and Unitised with Profit Life Insurance Contracts , 2003 .

[60]  Jennifer Pittman,et al.  Adaptive Splines and Genetic Algorithms , 2000 .