To combine or not to combine? Issues of combining forecasts

Article Abstract:

Combining forecasts presents a number of problems and issues that may be resolved by the use of Bayesian methods. These methods can utilize noninformative and informative prior probability densities and can account for model uncertainty. Results indicate that the simple average of forecasts is more effective than a weighted average of forecasts since it reduces variance, there is no need for estimation and dominates forecasts requiring optimal weighting.

Bayesian statistical decision theory, Bayesian analysis

---

Algorithms for explaining forecast revisions

Article Abstract:

A forecast revision analysis algorithm is presented. Forecast revision analysis reveals the dynamic properties of forecasting procedure and thus improves the understanding of forecasts. The algorithm requires statistical reasoning and a modeling theory to set the data. The algorithm may be used in a number of other models and theories since it is basically an accounting framework.

Algorithms

---

Similar abstracts:

• Abstracts: On trend extraction models: comments on a paper by Garcia-Ferrer and Del Hoyo. On trend extraction models: interpretation, empirical evidence and forecasting performance

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.