A change point problem for some conditional functionals
Let \((X_1,Y_1)\), \dots, \((X_n,Y_n)\) be i.i.d. r.v.'s, \(X\in R\), \(Y\in R^q\). The functional \[ \vartheta(y)=\sup_{x\in R}\Psi(x,G(x| y)) \] is considered, where \(G(x| y)=P\{X_1<x\;| \;Y_1=y\}\), and \(\Psi\) is some smooth function. To estimate \(\vartheta(y)\) the plug-in estimator is used: \(\hat\vartheta_n(y)=\sup_{x\in R}\Psi(x,G_{n,k_n}(x| y))\), where \(G_{n,k_n}(x| y)\) is the \(k_n\)-nearest neighbor estimator for \(G(x| y)\). It is proved that under some regularity conditions the following estimates hold true: \[ \sup_{x,y}| G_{n,k_n}(x| y)-G(x| y)| =o_P(k_n^{-\gamma}) \] and \[ \hat \vartheta(y)= \vartheta(y)+\Psi'(x^0_y,G(x^0_y| y)) G_{n,k_n}(x| y)-G(x| y) +o_P(\| G_{n,k_n}-G\|), \] where \(\Psi'={\partial\Psi(x,y)/\partial y}\), and \(x^0_y\) is the point at which \(\Psi(x,G(x| y))\) attains it's (unique) maximum. For the entire collection see [Zbl 0990.62506].
This page was built for publication: A change point problem for some conditional functionals
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3843168)