Premium
Decomposition; a strategy for judgemental forecasting
Author(s) -
Edmundson R. H.
Publication year - 1990
Publication title -
journal of forecasting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.543
H-Index - 59
eISSN - 1099-131X
pISSN - 0277-6693
DOI - 10.1002/for.3980090403
Subject(s) - extrapolation , exponential smoothing , task (project management) , computer science , residual , decomposition , series (stratigraphy) , smoothing , noise (video) , component (thermodynamics) , exponential function , artificial intelligence , data mining , algorithm , statistics , mathematics , computer vision , engineering , paleontology , mathematical analysis , physics , systems engineering , image (mathematics) , biology , thermodynamics , ecology
This paper reports the results of studies concerning the accuracy and efficiency of time‐series extrapolation decisions made with the assistance of an interactive graphical tool called GRAFFECT. The tool facilitates the decomposition of the extrapolation task by permitting the serial decomposition of the cue data as the task proceeds. GRAFFECT uses an interactive graphical interface controlled substantially with the use of a mouse. The extrapolation task is divided into the following: (1) trend modelling and extrapolation, (2) seasonal pattern modelling, and (3) extrapolation from the noise residual series. As each component is modelled its effect is stored and the information is washed out of the cue series. The ultimate forecast is produced by automatic recomposition of the judgementally determined components. The results show a significant improvement in forecast accuracy over unaided judgment, resulting in a subjective extrapolation that betters deseasonalized single exponential smoothing.