Resumen de Some properties of nonparametric predictors under stationarity
Pilar García Soidán, Raquel Menezes, C. Ferreira
In this work, nonparametric predictors will be constructed, based on the kernel method, for stationary random process. Our proposals will be proved to enjoy good properties and even optimal bandwidths will be achieved by asymptotically minimizing the mean-squared prediction errors. Alternatively, cross-validation approaches will be provided for selection of the bandwidth, more easily attainable in practice.
