Updating technology forecasting models using statistical control charts
ISSN: 0368-492X
Article publication date: 11 December 2017
Issue publication date: 16 March 2018
Abstract
Purpose
The purpose of this paper is to develop an approach that can detect abnormal deviations in the time series models for technology forecasting. The detected modifications provide a basis for understanding the determinants and impact of the corresponding change.
Design/methodology/approach
The proposed approach is based on monitoring residual values (the difference between the observation and the forecasted value) continuously using statistical control charts (SCCs). The residuals that are out of the expected limits are considered an alert indicating a remarkable change. To demonstrate the use of the proposed approach, a time series model was fitted to a number of TV-related patent counts. Subsequently, model residuals were used to determine the limits of the SCCs.
Findings
A number of patents granted in the year 2012 violated the upper control limit. A further analysis has shown that there is a linkage between the abnormal patent counts and the emergence of LCD TVs.
Practical implications
Change in technology may dramatically affect the accuracy of a forecasting model. The need for a parameter update indicates a significant change (emergence or death of a technology) in the technological environment. This may lead to the revision of managerial actions in R&D plans and investment decisions.
Originality/value
The proposed methodology brings a novel approach for abnormal data detection and provides a basis for understanding the determinants and impact of the corresponding change.
Keywords
Citation
Durmusoglu, A. (2018), "Updating technology forecasting models using statistical control charts", Kybernetes, Vol. 47 No. 4, pp. 672-688. https://doi.org/10.1108/K-04-2017-0144
Publisher
:Emerald Publishing Limited
Copyright © 2018, Emerald Publishing Limited