Technological forecasting - model selection, model stability, and combining models
Article Abstract:
The body of research on diffusion of innovations has produced different models for technological forecasting. The task of forecasters is to select the stochastic model suitable for forecasting. To aid them in this task, a study assessed 29 models of technological forecasting, dividing them into three classes based on the timing of the point of inflection in the innovation or substitution process. Measures of model fit and model stability were used to generate evidence for supporting model selection. Findings revealed the difficulty of identifying an appropriate model and employing it to obtain forecasts due to the shortage of evidence in the time series data to allow model recognition. Identifying a class of possible models was found to be easier than pinpointing the best model. Therefore, combining model forecasts is recommended.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1998
User Contributions:
Comment about this article or add new information about this topic:
Forecasting a cumulative variable using its partially accumulated data
Article Abstract:
A forecasting approach that can be performed on a cumulative variable on the basis of a record of its partially accumulated data is proposed. The model uses a statistical approach to discriminate among popular forecasting techniques in an objective manner. The basic aim is to use statistical models to generate minimum mean square error forecasts and allow the data to lead to the selection of an appropriate model to represent their behavior. The forecasting performance of the statistical models are evaluated by comparing their results against those obtained via algorithmic solutions. The models produced better forecasts for almost all lead times in three illustrative studies, as indicated by the basic measures of forecasting accuracy and precision.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1997
User Contributions:
Comment about this article or add new information about this topic:
A Recursive Kalman Filter Forecasting Approach
Article Abstract:
Forecasters are concerned that coefficients in their models vary over time. Exponential smoothing methods are designed to improve the forecast performance by letting the smoothing constant vary according to most recent forecast accuracy. Benefits of a Kalman filter type adaptive estimation and forecasting approach are studied. Better forecasts generally result. Also computational efficiency tends to improve.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1983
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: Utilizing forecast information to drive Solutia's supply chain. Evolution in the demand and supply planning at Dow Corning
- Abstracts: Engineering and marketing: their worlds run together. Planning, executing, and closing the sale
- Abstracts: Tax concerns when structuring cross-border investments in private Canadian companies. A new weapon in the fight against corruption
- Abstracts: Quality networking initiatives in Europe: classification and conditions for success. The future of the networked company
- Abstracts: Measuring the Effects of a University Planning Decision Aid. Effects of Planning Aids and Presentation Media on Performance and Affective Responses in Strategic Decision-Making