Bayes inference for technological substitution data with data-based transformation

Article Abstract:

A Bayesian inference using the Gibbs sampling method proves to be effective in forecasting technological substitutions. An experimental study, using the real data sets for color televisions and electronic telephone switching systems of a telephone company, shows that Bayesian predictive intervals are more accurate in prediction and are shorter than the usual frequentist's intervals with the help of Gibbs sampling. Also, Bayesian inference works effectively in checking the adequacy of models used in forecasting technology penetration, such as the logistic transformation(M1).

Forecasting, Research, Forecasts and trends, Technological forecasting, Technological innovations, Business forecasting

---

A remark on least-squares and naive extrapolations in non-linear AR(1) processes

Article Abstract:

A mathematical proof is presented to show the existence of a class of non-linear AR(1) models wherein the difference between the m-step least squares extrapolation and the m-step naive extrapolation is arbitrarily large. The condition exists when m is greater than or equal to 2. This proof is especially useful for statistical description of data that rejects the hypothesis of linearity.

Research and Development in the Physical, Engineering, and Life Sciences, Statistics, Models, Analysis, Mathematical models, Statistics (Mathematics), Statistical hypothesis testing

---

Similar abstracts:

• Abstracts: A stochastic model of sequential bargaining with complete information. Renegotiation in finitely repeated games
• Abstracts: Preferences over solutions to the bargaining problem. Bayesian vector autoregressions with stochastic volatility
• Abstracts: Virtual Bayesian implementation. Reputation and experimentation in repeated games with two long-run players
• Abstracts: Modeling expert forecasting knowledge for incorporation into expert systems
• Abstracts: Analysis of the US business cycle with a Vector-Markov-switching model. Misspecified prediction for time series

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