A local constant estimator reproduces constants but need not reproduce an affine regression function near a support boundary. Let , take and the regression kernel , and suppose . At target zero, the noiseless Nadaraya–Watson estimator averages for . Whenever , its exact fitted value isSince , this is the bias of an estimator when observation errors have zero expected value. Under and , it is asymptotic to . The one-sided boundary window has a nonzero first coordinate moment, explaining the first-order bias. An invertible local linear regression fit reproduces exactly at the target and removes this term. If , the local constant average is undefined rather than zero, so a nonempty window is a necessary condition.
For fixed , design , , a nonzero nonnegative symmetric piecewise regression kernel supported on , and , , write and . On the rescaled grid of spacing , the Riemann sum error for is , by bounded variation. The normalized first and second local moments are therefore and . A uniform second-order Taylor expansion gives the displayed bias of an estimator, since . Equivalently it is . The first-order term explains the generic boundary bias of a local constant estimator.
Articles by others on the same topic
There are currently no matching articles.