Past exam of the mathematics course of the University of Cambridge / 2016 / iii / Paper 206 / 5

At a target $x$, approximate the unknown mean locally by $a+b(X_i-x)$ and solve the local linear regression weighted least-squares problemwhere $K$ is a nonnegative regression kernel and $h>0$ a smoothing bandwidth. A common factor $1/h$ in all weights makes no difference. Estimate $f(x)$ by the fitted intercept ${\hat{a}}_x$, and optionally $f^{\prime }(x)$ by ${\hat{b}}_x$. Repeat at each target rather than fitting a single global straight line.

For an explicit expression put $S_k={\sum }_i{w}_i(X_i-x{)}^k$ and $T_k={\sum }_i{w}_i(X_i-x{)}^k{Y}_i$. The normal equations yieldThe denominator is positive provided there are at least two distinct positively weighted predictor values. The weights sum to one and their first centered moment is zero, so local linear regression exactly reproduces affine functions and reduces boundary bias compared with a local constant fit. Smaller $h$ uses more local detail and increases variance; larger $h$ smooths more and can increase bias. Choose the bandwidth by cross-validation or another justified criterion. A nearest-neighbour span provides an adaptive bandwidth, as in the local fits used later.

Assume initially that the design points are distinct and ordered on $[a,b]$. The cubic smoothing spline minimizesover functions with an absolutely continuous first derivative and square-integrable second derivative. Dividing the residual term by $n$ simply changes the scaling assigned to $\lambda$. The minimum roughness property of the natural cubic spline interpolant reduces the problem to natural cubic splines with knots at the design points: among functions with any prescribed fitted values, that interpolant has the smallest integrated squared curvature. The minimizer is therefore a natural cubic smoothing spline, with natural second-derivative boundary conditions and linear tails when extended.

Choose a full basis $B_1,\dots ,B_n$ for this natural-spline space. Put $B_{ij}=B_j(X_i)$ and ${\Omega }_{jk}={\int }_a^b{B}_j^{\prime \prime }(t)B_k^{\prime \prime }(t)dt$. Writing $g(t)={\sum }_j{c}_j{B}_j(t)$ changes the objective toDifferentiation gives $(B^{T}B+\lambda \Omega )\hat{c}=B^{T}Y$. Although the curvature penalty has an affine nullspace, the full basis at distinct design points makes the combined matrix positive definite. HenceEquivalently, use the cardinal basis that interpolates the observations, making $B=I$ and $S_{\lambda }=(I+\lambda \Omega {)}^{-1}$. These are smoothing matrices, symmetric but generally not idempotent. Their effective degrees of freedom are $\mathrm{tr}(S_{\lambda })$. As $\lambda \downarrow 0$ the fit interpolates; as $\lambda \uparrow \infty$ it approaches ordinary least squares on the affine functions. With replicated predictors, use a design matrix for distinct knots and observation weights; the basis form remains valid when identifiable.

Useful choices include Leave-one-out cross-validation, generalized cross-validation, and estimation of smoothness by restricted maximum likelihood through a mixed-model representation. For fixed $\lambda$, the leave-one-out residual identity for a linear smoother givesMinimize these over a suitable range. Choose smoothing by predictive performance or a variance-component likelihood, not by arbitrarily forcing a visually appealing curve.

Both fits use the Gaussian generalized additive modelwhere $e_i$ is experience and $d_i$ education. Centering makes the additive decomposition identifiable. No experience-education interaction term is included.

The first fit uses local linear regression for each term. span=1 includes all observations in each nearest-neighbour neighbourhood, but distance weights and a moving target still vary: it does not make the fit exactly a global straight line. The second uses cubic smoothing splines, with separate tuning values spar=0.8 and spar=1.2. Increasing spar increases the curvature penalty and smooths more; spar is a monotone tuning parametrization, not the numerical $\lambda$ in part (b). The relation depends on the predictor scaling and design.

The plotted experience effect is increasing in both fits. The local fit gives a broadly smooth, nearly linear rise; the spline fit follows more of the steep low-experience rise and subsequent flattening, especially near the lower boundary. Education has a weaker increasing effect. The local estimate shows some curvature, while the spline fit at spar=1.2 is closer to a straight line. The clearest difference is the amount of smoothing and the experience curvature, rather than opposite effect directions. These are centered partial-effect plots with partial residuals, not scatterplots of wage against each predictor alone.

A smaller local span would allow the first fit to capture more experience curvature. A moderately smaller education spar could reveal nonlinear detail suppressed in the second fit; its small but significant nonparametric component in part (e) supports checking that possibility. Do not tune solely to follow every point, especially at sparse boundaries. Compare candidate spans or separate spline penalties using cross-validation or generalized cross-validation; increase smoothing if flexible fits introduce unsupported wiggles. See the documented monotone spar–penalty convention at stat.ethz.ch/R-manual/R-patched/library/stats/html/smooth.spline.html.

Use the backfitting algorithm to alternate conditional updates of the two additive functions. Initialize $\alpha =\overline{Y}$ and both centered function vectors at zero. With $S_e,S_d$ the smoothing matrices for the chosen experience and education spline penalties, one cycle isfollowed byThe centering transfers any constant component to the intercept; update $\alpha$ to the mean of $Y-f_e-f_d$ if necessary, which is $\overline{Y}$ under the stated constraints. Repeat until changes in the functions or objective are negligible. Each function is estimated by smoothing its partial residual, namely the response after subtracting all other currently fitted terms.

For fixed penalties these updates are block minimizations ofsubject to the centering constraints. Each step cannot increase the objective; with identifiable additive terms and a positive-definite constrained quadratic, backfitting converges to its unique minimizer. Highly dependent predictors can make the decomposition poorly identified or slow convergence. The Gaussian identity-link model needs these least-squares updates directly; non-Gaussian generalized additive models use weighted backfitting inside iteratively reweighted least squares. The package implementation also separates each spline's unpenalized linear component from its nonlinear component, giving the two ANOVA tables displayed in part (e).

The null model's 99 residual degrees of freedom imply $n=100$. Each predictor has a one-degree-of-freedom linear contribution in the parametric table, plus the reported nonlinear contribution. Consequently the termwise effective degrees of freedom areInclude the intercept once, giving about $12.2$ total effective degrees of freedom. The more precise residual degrees of freedom in the output giveThere is no conflict: $8.8$ and $0.4$ are rounded, so they cannot recover the precise total. Individual unrounded termwise values are not provided.

The package's residual mean-square estimate of error variance ismatching the rounded $10328$ in the table. The nonlinear experience contribution is highly significant, while the smaller nonlinear education contribution has approximate $p=0.03923$.

This calculation uses the effective residual degrees of freedom reported by this fit. For a general non-idempotent linear smoother $S$, the noise contribution to expected RSS is ${\sigma }^2\mathrm{tr}[(I-S{)}^T(I-S)]$, and the expectation can additionally contain squared smoothing bias. Thus dividing by $n-\mathrm{tr}(S)$ is not a universal exact-unbiasedness theorem. The output's separation of linear and nonlinear term degrees of freedom is documented at cran.r-project.org/web/packages/gam/gam.pdf.