Initial data truncation for univariate output of discrete-event simulations using the Kalman filter
Article Abstract:
A new truncation point selection procedure for lessening the initial bias in the univariate output of discrete-event simulations is introduced. The data truncation scheme, which is applicable to statistics based on observations and statistics based on time-persistent variables, involves an autoregressive process, namely, the state-space characterization of the simulation output. This is followed by the derivation of a time-varying estimate of the simulation output's steady-state mean via a Bayesian method known as Multiple Model Adaptive Estimation. The latter employs the use of three Kalman filters in parallel to process the simulation data. The new rule, which was evaluated through a Monte Carlo analysis of data general from 12 stochastic models, results in more accurate estimates of truncated sequences.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1996
User Contributions:
Comment about this article or add new information about this topic:
Aggregation error in Bayesian analysis of reliability systems
Article Abstract:
The size estimation of aggregation errors in Bayesian analyses of system reliability was considered. Specifically, several procedures for measuring actual aggregation errors and for deriving bounds on the aggregation error based on alternative definitions of aggregation error were developed. The procedures, which result in bounds that do not exceed twice the actual error, allow analysts to determine whether to perform an aggregate or disaggregate analysis of system failure as well as assume an independent prior distributions of the failure rate and a one-out-of-n success logic. They were applied to systems consisting of Poisson components in series, Bernoulli components in parallel, Bernoulli components in series, and Poisson and Bernoulli components in parallel.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1996
User Contributions:
Comment about this article or add new information about this topic:
Blacklisting social science departments with poor Ph.D. submission rates
Article Abstract:
Research examining the statistical grounds upon which the Economic and Social Science Research Council (ESRC) of the United Kingdom has based their decision to withhold funds from social science graduate programs with the lowest Ph.D. thesis submission rates reveals that the ESRC has failed to account for the different numbers of students in social science graduate schools. An empirical Bayes method is used to calculate submission rates and account for the number of students. Results indicate that true unobservable submission rates are actually small.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1988
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: Estimation of multiresponse simulation metamodels using control variates. Comparative invention performance of major industrial countries: patterns and explanations
- Abstracts: Shifting paradigms for sustainable development: implications for management theory and research. Interorganizational groups: origins, structure, and outcomes
- Abstracts: National and corporate cultural fit in mergers/acquisitions: an exploratory study. Leader style in strategy and organizational performance: an integrative framework
- Abstracts: Dynamic asset allocation in a mean-variance framework. The valuation and analysis of adjustable rate mortgages
- Abstracts: The escalation of commitment to a failing course of action: toward theoretical progress