Search results

The purpose of this paper is to suggest a framework for sales forecasting more suitable for smaller firms. The authors examine the sales forecasting practices of small firms and then proceed to discuss the application of Bayesian decision theory in the production of sales forecasts, a method arguably more suited to the smaller firm. The authors suggest that many small firm entrepreneurs are inherently “Bayesian” in their thinking approach to predicting events in that they often rely on subjective estimates at least for initial starting values.

A triangulated approach which uses qualitative group discussions and thematic content analysis, a reasonably large‐scale questionnaire sample survey administered by post and analysed using descriptive statistics and non‐parametric tests of association and a case study approach based on the authors own consultancy activities to illustrate the practical application of the forecasting model suggested.

That many small firms use no formal sales forecasting framework at all. That the majority of small firm owners and/or managers rate sales forecasting skills very low in their list of priorities when given a choice of course to attend at subsidised rates. That there is no significant difference in the importance small firm owners and/or managers attach to formal sales forecasting skills.

Information has been gained from one geographic area in the north of England although the results may have a wider application to all small firms in the UK and elsewhere. Only the region's six most important industry sectors were included as stratification variables in the sample survey. Other regions will have a different mix of industries and will be stratified differently.

The article addresses the sales forecasting needs of small firms specifically within the marketing for small business context and offers a realistic option with a clear rationale.

In forecasting unknown quantities, risk and finance decision makers often rely on one or more biased experts, statistical specialists representing parties with an interest in the decision maker's final forecast. This problem arises in a variety of contexts, and the decision maker may represent a corporate enterprise, rating agency, government regulator, etc. The purpose of the paper is to assist decision makers, experts, and others to have a better understanding of the dynamics of the problem, and to adopt strategies and practices that enhance efficiency.

The problem is formulated as a two‐person, non‐cooperative Bayesian game with the decision maker and one expert as players, and perfect Bayesian equilibrium solutions are identified. Then the analysis is extended to variations of the game in which the expert's loss function is not common knowledge, and in which there are multiple experts.

In the struggle for information between the decision maker and the experts, the experts generally benefit from greater uncertainty about the parameters of the model. Thus, in attempting to elicit as much information as possible from the experts, the decision maker must strive to minimize all sources of uncertainty.

As in most Bayesian games, the analysis requires that a variety of process assumptions and model parameters be common knowledge. These conditions may be difficult to satisfy in real‐world applications.

The principal finding of the study is that there is truly a struggle for information between the decision maker and the experts. This generally encourages the experts to inject as much uncertainty as possible into the process. To counter this effect, the decision maker might: provide incentives for the experts to increase their sampling information; try to mitigate specific uncertainties regarding the model parameters; and try to increase the number of experts.

This is the first paper to apply the framework of signaling games to the problem of eliciting information from biased experts. It is of value to decision makers, experts, and economic researchers.

The purpose of this paper is to present an inductive methodology, which supports ranking of entities. Methodology is based on Bayesian latent variable measurement modeling and makes use of assessment across composite indicators to assess internal and external model validity (uncertainty is used in lieu of validity). Proposed methodology is generic and it is demonstrated on a well‐known data set, related to the relative position of a country in a “doing business.”

The methodology is demonstrated using data from the World Banks' “Doing Business 2008” project. A Bayesian latent variable measurement model is developed and both internal and external model uncertainties are considered.

The methodology enables the quantification of model structure uncertainty through comparisons among competing models, nested or non‐nested using both an information theoretic approach and a Bayesian approach. Furthermore, it estimates the degree of uncertainty in the rankings of alternatives.

Analyses are restricted to first‐order Bayesian measurement models.

Overall, the presented methodology contributes to a better understanding of ranking efforts providing a useful tool for those who publish rankings to gain greater insights into the nature of the distinctions they disseminate.

The purpose of this editorial is to explore the usefulness of distinguishing between “risk” and “Knightian uncertainty.”

The paper presents a representative, insurance‐based model of Knightian uncertainty arising out of potential major structural changes (without historical precedent) in liability claim settlements. It then considers whether or not formal statistical forecasting and decision making are possible in this context.

For real‐world settings, it is found that a Bayesian statistical framework is sufficiently comprehensive to permit forecasting and decision making in the presence of Knightian uncertainty. The paper then shows that the Bayesian approach fails only if the sample space underlying the potential structural change is truly nonmeasurable.

It is argued that, under a Bayesian worldview, the distinction between risk and uncertainty is necessary only in highly abstract epistemological modeling.

Information analysis tools enhance the possibilities of firm competition in terms of knowledge management. However, the generalization of decision support systems (DSS) is still far away from everyday use by managers and academicians. This paper aims to present a framework of analysis based on Bayesian networks (BN) whose accuracy is measured in order to assess scientific evidence.

Different learning algorithms based on BN are applied to extract relevant information about the relationship between operations strategy and flexibility in a sample of engineering consulting firms. Feature selection algorithms automatically are able to improve the accuracy of these classifiers.

Results show that the behaviors of the firms can be reduced to different rules that help in the decision‐making process about investments in technology and production resources.

Contrasting with methods from the classic statistics, Bayesian classifiers are able to model a variety of relationships between the variables affecting the dependent variable. Contrasting with other methods from the artificial intelligence field, such as neural networks or support vector machines, Bayesian classifiers are white‐box models that can directly be interpreted. Together with feature selection techniques from the machine learning field, they are able to automatically learn a model that accurately fits the data.

This editorial aims to consider certain aspects of the parallel development of statistics and actuarial science that are instructive in understanding the value of intellectual cross‐fertilization in research.

The paper addresses the fundamental problem of combining information from the perspectives of both formal statistical inference and actuarial science, and discusses how the two fields not only have benefited from exchanges of ideas, but also have suffered from research isolation.

The “credibility” problem of actuarial science, which involves the combining of statistical observations, may be formulated in the language of conventional statistical estimation. However, its unique concept of “collateral” information helped stimulate the development of empirical Bayesian techniques in statistics, and the actuarial literature itself has benefited from the rigorous statistical treatment of the credibility problem by conventional estimation theory. Interestingly, however, certain noteworthy gaps persist.

The editorial illustrates how the exchange of ideas – or lack thereof – can have a profound effect on the development of science.

The purpose of this paper is to reveal the pattern of passengers' transferring on occasion of a large crowd being stranded at transportation hubs (such as a bus station, railway station, airport, etc.) in climate disasters, and then propose the proper policy recommendations for the government to evacuate stranded passengers.

A model is established based on Bayesian network and influence diagram to catch the features of a passenger's decision‐making process, and the transition probabilities of passengers are revised on the basis of the theory of herd behaviors in information to describe the influence of group behaviors on passenger individuals. Subsequently, a multi‐agent model is developed in Repast platform in Java language, and simulation and analysis are also made.

The results of simulation show that it is possible to apply the theory of herd behaviors and the multi‐agent method in analyzing the effectiveness of government policies on evacuating stranded passengers in climate disasters.

The research of this paper has important practical significance for the government to developing policies to evacuating stranded passengers in climate disasters, and is a useful exploration to open up new methodologies for emergency management.

The purpose of this paper is to provide a methodology for benchmarking supplier risks through the creation of Bayesian networks. The networks are used to determine a supplier's external, operational, and network risk probability to assess its potential impact on the buyer organization.

The research methodology includes the use of a risk assessment model, surveys, data collection from internal and external sources, and the creation of Bayesian networks used to create risk profiles for the study participants.

It is found that Bayesian networks can be used as an effective benchmarking tool to assist managers in making decisions regarding current and prospective suppliers based upon their potential impact on the buyer organization, as illustrated through their associated risk profiles.

A potential limitation to the use of the methodology presented in the study is the ability to acquire the necessary data from current and potential suppliers needed to construct the Bayesian networks.

The methodology presented in this paper can be used by buyer organizations to benchmark supplier risks in supply chain networks, which may lead to adjustments to existing risk management strategies, policies, and tactics.

This paper provides practitioners with an additional tool for benchmarking supplier risks. Additionally, it provides the foundation for future research studies in the use of Bayesian networks for the examination of supplier risks.

The purpose of this paper is to provide the maintenance personnel with a methodology for modeling and estimating the reliability of critical machine systems using the historical data of their inter‐failure times.

The failure patterns of five different machine systems were modeled with NHPP‐log linear process and HPP belonging to stochastic point process for predicting their reliability in future time frames. Besides the classical approach, Bayesian approach was also used involving Jeffreys's invariant non‐informative independent priors to derive the posterior densities of the model parameters of NHPP‐LLP and HPP with a view to estimating the reliability of the machine systems in future time intervals.

For at least three machine systems, Bayesian approach gave lower reliability estimates and a larger number of (expected) failures than those obtained by the classical approach. Again, Bayesian estimates of the probability that “ROCOF (rate of occurrence of failures) would exceed its upper threshold limit” in future time frames were uniformly higher for these machine systems than those obtained with the classical approach.

This study indicated that, the Bayesian approach would give more realistic estimates of reliability (in future time frames) of the machine systems, which had dependent inter‐failure times. Such information would be helpful to the maintenance team for deciding on appropriate maintenance strategy.

With the help of Bayesian approach, the posterior densities of the model parameters were found analytically by considering Jeffreys's invariant non‐informative independent prior. The case study would serve to motivate the maintenance teams to model the failure patterns of the repairable systems making use of the historical data on inter‐failure times and estimating their reliability in future time frames.

This paper seeks to develop an approach to problem localization and an algorithm to address the issue of determining the dependencies among system metrics for automated system management in ubiquitous computing systems.

This paper proposes an approach to problem localization for learning the knowledge of dynamic environment using probabilistic dependency analysis to automatically determine problems. This approach is based on Bayesian learning to describe a system as a hierarchical dependency network, determining root causes of problems via inductive and deductive inferences on the network. An algorithm of preprocessing is performed to create ordering parameters that have close relationships with problems.

The findings show that using ordering parameters as input of network learning, it reduces learning time and maintains accuracy in diverse domains especially in the case of including large number of parameters, hence improving efficiency and accuracy of problem localization.

An evaluation of the work is presented through performance measurements. Various comparisons and evaluations prove that the proposed approach is effective on problem localization and it can achieve significant cost savings.

This study contributes to research into the application of probabilistic dependency analysis in localizing the root cause of problems and predicting potential problems at run time after probabilities propagation throughout a network, particularly in relation to fault management in self‐managing systems.