Forecasting with Optimized Moving Local Regression

Autor: Fedorov, Valery V., Hackl, Peter, Müller, Werner
Jazyk: angličtina
Rok vydání: 1992
Druh dokumentu: Paper
Popis: This paper empirically demonstrates the relative merits of the optimal choice of the weight function in a moving local regression as suggested by Fedorov et al., (1993) over traditional weight functions which ignore the form of the local model. The discussion is based on a task that is imbedded into the smoothing methodology, namely the forecasting of business time series data with the help of a one-sided moving local regression model. (author's abstract)
Series: Forschungsberichte / Institut für Statistik
Databáze: Networked Digital Library of Theses & Dissertations