We discuss the issues raised by Fan and Jiang in the context of high dimensional models and argue that fitting sparse nonparametric models might be preferable to hypothesis testing.
© 2001-2026 Fundación Dialnet · Todos los derechos reservados
Coordinado por: