Modelselection,estimation,and.PDF

Model selection, estimation, and bootstrap smoothing Bradley Efron? Stanford University Abstract Classical statistical theory ignores model selection in assessing estimation accuracy. Here we consider bootstrap methods for computing standard errors and con?dence intervals that take model selection into account. The methodology involves bootstrap smoothing, also known as bagging, to tame the erratic discontinuities of selection-based estimators. A projection theorem then provides standard errors for the smoothed estimators. Two examples, non-parametric and parametric, are carried through in detail: a regression model where the choice of degree (linear, quadratic, cubic,. . . ) is determined by the Cp criterion, and a Lasso-based estimation problem. Keywords: model averaging, Cp, Lasso, bagging, ABC intervals, importance sampling 1 Introduction Accuracy assessments of statistical estimators customarily are made ignoring model selection. A prelimi- nary look at the data might, for example, suggest a cubic regression model, after which the ?tted curve’s accuracy is computed as if “cubic” were pre-chosen. Here we will discuss bootstrap standard errors and con?dence intervals that take into account the model selection procedure. Figure 1 concerns the Cholesterol Data, an example investigated in more detail in Section 2: n = 164 men took cholestyramine, a proposed cholesterol-lowering drug, for an average of seven years each; the response variable y was the decrease in blood-level cholesterol measured from the beginning to the end of the trial, d = cholesterol decrease; (1.1) also measured (by pill counts) was compliance, the proportion of the intended dose taken, c =