{\sl smoother} {\sl nonparametric} nature {\sl smooth} RIGHT {\sl scatterplot smoothing} {\sl categorical} {\sl smoothing} {\sl local averaging} {\sl neighbourhoods} around {\sl brand} {\sl smoothing parameter} {\sl fundamental tradeoff between bias and variance} {\em robustified} {\sl infinitely smooth} {\sl close} {\sl symmetric nearest neighbourhood} {\sl running mean} {\sl nearest neighbourhood} {\sl moving average} {\sl smoother} RIGHT {\sl running lines smoother} {\sl weighted} least-squares {\sl loess} {\sl kernel} {\sl metric} distance {\sl metric} {\sl rank} distance {\em Hanning} {\em twicing} {\em regression smoothers} {\em linear} {\em degrees of freedom} {\sl equivalent kernels} {\sl linear} {\em equivalent kernel} {\em loess} smooth {\sl equivalent degrees of freedom} {\sl piecewise} {\sl knots} {\sl cubic} splines {\sl number} {\sl when the knots are given} {\sl natural cubic spline} {\sl effective} dimension {\sl sorted} values {\sl natural-spline} basis {\sl local averaging} {\sl kernel} {\sl loess} {\sl tri-cube} weight {\sl span} {\sl loess} {\sl nearest neighbours} {\sl Nearest} {\sl thin-plate spline} {\sl tensor product} {\sl per~se} {\sl Updating formula for running-line smooth} {\sl Basis for natural splines} {\sl Derivation of smoothing splines ^{Reinsch (1967)}.} {\sl Semi-parametric regression ^^{Green, P.J.}^^{Jennison, {\sl estimating} equations {\sl Efficient kernel smoothing ^{Silverman (1982)},