Nonparametric Sequential Surveillance of Intervention Effects in Time Series

Publikation: Beitrag in Buch/KonferenzbandBeitrag in Sammelwerk

Abstract

A nonparametric process procedure is proposed as a tool for continuous monitoring to detect intervention effects in a time series. The procedure uses exponentially weighted moving averages of the ranks of the post-intervention one-step-ahead forecast errors. The forecasts are based on the pre-intervention model, and the errors are ranked with respect to residuals in the pre-intervention period. The method is illustrated with the Los Angeles oxidant data previously analyzed by Box and Tiao (1975).
OriginalspracheEnglisch
Titel des SammelwerksEconomic Structural Change
Untertitel des SammelwerksAnalysis and Forecasting
Herausgeber*innenPeter Hackl, Anders Westlund
ErscheinungsortBerlin
VerlagSpringer
Kapitel7
Seiten85-94
ISBN (elektronisch)978-3-662-06824-3
ISBN (Print)978-3-662-06826-7
DOIs
PublikationsstatusVeröffentlicht - 1991

Zitat