Bayesian Model Averaging of Dynamic Linear Models
نویسندگان
چکیده
In this paper we aim to compare the performance of three different Bayesian model averaging (or mixture) methods applied to regression dynamic linear models for beverage data from Zimbabwe. The models are chosen to reflect different plausible causal structures of association between beverage sales and other variables, thought to influence beverage sales, such as prices, temperature and maize crop production. A model averaging method based on Akaike weights performed on average 4% better than a Monte Carlo Markov Chain (MCMC) simulation method judged by the predictive one step ahead percentage root mean square error (prmse) for forecasting a test set. The Akaike method also performed on average 73% better than a Quasi Bayes method on the same prmse measure on the test set.
منابع مشابه
Predicting waste generation using Bayesian model averaging
A prognosis model has been developed for solid waste generation from households in Hoi An City, a famous tourist city in Viet Nam. Waste sampling, followed by a questionnaire survey, was carried out to gather data. The Bayesian model average method was used to identify factors significantly associated with waste generation. Multivariate linear regression analysis was then applied to evaluate th...
متن کاملComparison of Kullback-Leibler, Hellinger and LINEX with Quadratic Loss Function in Bayesian Dynamic Linear Models: Forecasting of Real Price of Oil
In this paper we intend to examine the application of Kullback-Leibler, Hellinger and LINEX loss function in Dynamic Linear Model using the real price of oil for 106 years of data from 1913 to 2018 concerning the asymmetric problem in filtering and forecasting. We use DLM form of the basic Hoteling Model under Quadratic loss function, Kullback-Leibler, Hellinger and LINEX trying to address the ...
متن کاملModeling Factors Affecting Tax Evasion in Iran's Economy Based on the Bayesian averaging approach
This study seeks to model tax evasion and identify how effective factors affect tax evasion in the Iranian economy. Recent models show the failure of traditional models; Models do not have enough ability to model hidden variables such as tax evasion. The present study considers this failure in identifying explanatory variables and experimental model design. To achieve this, the Bayesian averagi...
متن کاملDiscount Weighted Bayesian Model Averaging for Portfolio Decisions in Matrix Variate Dynamic Linear Models Discount-Weighted Bayesian Model Averaging for Portfolio Decisions in Matrix Variate Dynamic Linear Models
In this paper, we assess Bayesian model averaging (BMA) techniques for dynamic linear models (DLMs) with variance matrix discounting. In previous research, the discount factors for the variance matrices and the auto-regressive lag have typically been pre-determined and held constant over time. Using posterior model probabilities, we average DLMs employing different discount rates and lag parame...
متن کاملForecasting Product Sales with Dynamic Linear Mixture Models
This paper discusses investigations undertaken at Sun Microsystems, Inc. into practical forecasting applications of the Bayesian Dynamic Linear Models described in the seminal work of West and Harrison (1997). Particular emphasis is placed on the use of class II mixture models, which use Bayesian model averaging to help accommodate model uncertainty.
متن کامل