Transition times: distributions arising from time heterogeneous Poisson processes

Article Abstract:

The distribution of the time to the change of state, or the transition time, is analyzed. A standard model for the number of shocks needed before the transition, a homogeneous Poisson process, is expanded by adding the element of heterogeneity into it in terms of time, the intensity function across the population and the number of arrivals before transition across the population. This results in the creation of a generalized F distribution for the transition time which generates a variety of distributions for modeling. These results are illustrated by applying them on two common problems: automobile accidents and redemption of coupons. The model's adequacy can be tested using a likelihood ration test since the distributions are part of the generalized F family for a certain type of the intensity function.

Distribution (Probability theory), Functions, Gamma, Gamma functions, Poisson distribution

---

Hierarchical Bayes methods for multifactor model estimation and portfolio selection

Article Abstract:

A study was conducted to analyze a hierarchical Bayes modeling process for portfolio management and research that can enhance the accuracy of factor model parameter estimations by integrating cross-sectional data. The Bayesian estimation process was implemented using Gibbs sampling. Analytic measures were then used to determine the sensitivity of optimal portfolio allocations against parameter estimation error. In addition, a simulation analysis of the hierarchical Bayesian parameter estimation and portfolio selection process was carried out. Results indicated that the hierarchical modeling process promotes significant improvements in the performance of portfolio. Findings also showed that the optimal portfolio weights were influenced by estimation error during small idiosyncratic differences.

Commercial Banks, Investment Banking and Securities Dealing, Portfolio & Funds Management, Models, Portfolio management

---

Dynamically updating relevance judgements in probabilistic information systems via users' feedback

Article Abstract:

Researchers have devised an information system that is useful in environments where the set of possible queries and the information base are large and have uncertain relationships. The system is a sequential, Bayesian, probabilistic indexing model that combines data regarding the system's performance with expert opinion. The expert opinion is modified by the user's feedback about the relevance of the retrieved information to their queries. The applicability of a particular datum in the information base to the current query is predicted by a logistic function of the expert opinion and the feedback.

Research, Management science, Decision-making, Decision making, Management, Expert systems

Similar abstracts:

• Abstracts: The distribution of standardized futures price changes. Conditional heteroskedasticity, asymmetry, and option pricing
• Abstracts: Intuition in strategic decision making: friend of foe in the fast - paced 21st century. Improving firm performance through entrepreneurial actions: Acordia's corporate entrepreneurship strategy
• Abstracts: Estimation of Attribute Weights from Preference Comparisons. Sequencing capacity expansion projects in continuous time
• Abstracts: Beware of new pension distribution rules. New rules for home mortgages
• Abstracts: Nonlinear utility models arising from unmodelled small world intercorrelations. An additive group utility for a fund manager

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