Coxian Distribution Research Articles

On a Sparre Andersen risk model perturbed by a spectrally negative Lévy process

In this paper, we consider a Sparre Andersen risk model perturbed by a spectrally negative Lévy process (SNLP). Assuming that the interclaim times follow a Coxian distribution, we show that the Laplace transforms and defective renewal equations for the Gerber–Shiu functions can be obtained by employing the roots of a generalized Lundberg equation. When the SNLP is a combination of a Brownian motion and a compound Poisson process with exponential jumps, explicit expressions and asymptotic formulas for the Gerber–Shiu functions are obtained for exponential claim size distribution and heavy-tailed claim size distribution, respectively.

A note on deficit analysis in dependency models involving Coxian claim amounts

In this paper, we consider a fairly large class of dependent Sparre Andersen risk models where the claim sizes belong to the class of Coxian distributions. We analyze the Gerber–Shiu discounted penalty function when the penalty function depends on the deficit at ruin. We show that the system of equations needed to solve for this quantity is surprisingly simple. Various applications of this result are also considered.

Assessing length of stay and associated characteristics of geriatric patients in Northern Ireland

The effective provision of care for the elderly is becoming increasingly more difficult. This is due to the rising proportion of elderly in the population, increasing demands placed on the health services and the financial strain placed on an already stretched economy. The research presented in this paper uses three different models to represent the length of stay distribution of geriatric patients admitted to one of the six key acute hospitals in Northern Ireland and various patient characteristics associated with their respective length of stay. The accurate modelling of bed usage within wards would enable hospital managers to prepare patient discharge packages and rehabilitation services in advance. The models presented within the paper include a Cox proportional hazards model, a Bayesian network with a discrete variable to represent length of stay and a special conditional phase-type model (C-Ph) with a connecting outcome node. This research demonstrates the new efficient fitting algorithm employed for Coxian phase-type distributions while updating C-Ph models for recent elderly patient data.

Computational learning of the conditional phase-type (C-Ph) distribution

This paper presents a new algorithm for learning the structure of a special type of Bayesian network. The conditional phase-type (C-Ph) distribution is a Bayesian network that models the probabilistic causal relationships between a skewed continuous variable, modelled by the Coxian phase-type distribution, a special type of Markov model, and a set of interacting discrete variables. The algorithm takes a data set as input and produces the structure, parameters and graphical representations of the fit of the C-Ph distribution as output. The algorithm, which uses a greedy-search technique and has been implemented in MATLAB, is evaluated using a simulated data set consisting of 20,000 cases. The results show that the original C-Ph distribution is recaptured and the fit of the network to the data is discussed.

Modelling emergency medical services with phase-type distributions

Effective and efficient emergency medical services (EMS) are a critical part of a national healthcare system. This paper describes research to model EMS by better capturing ambulance service times using Coxian phase-type (PH) distributions. Distributions are fitted to both the overall cycle time for different classes of patient priorities, as well as to sub-cycles. Sub-cycles are the distinct identifiable parts of the ambulance cycle time, such as travel times, time on scene and turnaround time at the hospital. The Coxian PH fits have then been used within a priority simulation model to provide guidance on the number of ambulances required to meet response time targets. Results from using various numbers of phases from the fitted Coxian distributions are compared. The proposed benefit of using sub-cycle fits is that it more readily permits scenario modelling within the simulation, such as evaluating the impact of reducing the turnaround time on the overall response times in the ambulance system.

Majorization and Extremal PH -Distributions (abstract only)

This paper presents majorization results for PH -generators. Based on the majorization results, Coxian distributions are identified to be extremal PH -distributions with respect to the first moment for certain subsets of PH -distributions. Bounds on the mean of phase-type distributions are found. In addition, numerical results indicate that Coxian distributions are extremal PH -distributions with respect to the moment of any order for certain subsets of PH -distributions.

A Recurrent Solution of PH/M/c/N-Like and PH/M/c-Like Queues

We propose an efficient semi-numerical approach to compute the steady-state probability distribution for the number of requests at arbitrary and at arrival time instants in PH/M/c-like systems with homogeneous servers in which the inter-arrival time distribution is represented by an acyclic set of memoryless phases. Our method is based on conditional probabilities and results in a simple computationally stable recurrence. It avoids the explicit manipulation of potentially large matrices and involves no iteration. Owing to the use of conditional probabilities, it delays the onset of numerical issues related to floating-point underflow as the number of servers and/or phases increases. For generalized Coxian distributions, the computational complexity of the proposed approach grows linearly with the number of phases in the distribution.

A Recurrent Solution of PH/M/c/N-Like and PH/M/c-Like Queues

We propose an efficient semi-numerical approach to compute the steady-state probability distribution for the number of requests at arbitrary and at arrival time instants in PH/M/c-like systems with homogeneous servers in which the inter-arrival time distribution is represented by an acyclic set of memoryless phases. Our method is based on conditional probabilities and results in a simple computationally stable recurrence. It avoids the explicit manipulation of potentially large matrices and involves no iteration. Owing to the use of conditional probabilities, it delays the onset of numerical issues related to floating-point underflow as the number of servers and/or phases increases. For generalized Coxian distributions, the computational complexity of the proposed approach grows linearly with the number of phases in the distribution.

Discrete Conditional Phase-type models utilising classification trees: Application to modelling health service capacities

Discrete Conditional Phase-type models (DC-Ph) consist of a process component (survival distribution) preceded by a set of related conditional discrete variables. This paper introduces a DC-Ph model where the conditional component is a classification tree. The approach is utilised for modelling health service capacities by better predicting service times, as captured by Coxian phase-type distributions, interfaced with results from a classification tree algorithm. To illustrate the approach, a case-study within the healthcare delivery domain is given, namely that of maternity services. The classification analysis is shown to give good predictors for complications during childbirth. Based on the classification tree predictions, the duration of childbirth on the labour ward is then modelled as either a two or three-phase Coxian distribution. The resulting DC-Ph model is used to calculate the number of patients and associated bed occupancies, patient turnover, and to model the consequences of changes to risk status.

Capturing the re-admission process: focus on time window

In the majority of studies on patient re-admissions, a re-admission is deemed to have occurred if a patient is admitted within a time window of the previous discharge date. However, these time windows have rarely been objectively justified. We capture the re-admission process from the community using a special case of a Coxian phase-type distribution, expressed as a mixture of two generalized Erlang distributions. Using the Bayes theorem, we compute the optimal time windows in defining re-admission. From the national data set in England, we defined re-admission for chronic obstructive pulmonary disease (COPD), stroke, congestive heart failure, and hip- and thigh-fractured patients as 41, 9, 37, and 8 days, respectively. These time windows could be used to classify patients into two groups (binary response), namely those patients who are at high risk (e.g. within 41 days for COPD) and low risk of re-admission group (respectively, greater than 41 days). The generality of the modelling framework and the capability of supporting a broad class of distributions enables the applicability into other domains, to capture the process within the field of interest and to determine an appropriate time window (a cut-off value) based on evidence objectively derived from operational data.

BAYESIAN ANALYSIS OF AGGREGATE LOSS MODELS

This paper describes a Bayesian approach to make inference for aggregate loss models in the insurance framework. A semiparametric model based on Coxian distributions is proposed for the approximation of both the interarrival time between claims and the claim size distributions. A Bayesian density estimation approach for the Coxian distribution is implemented using reversible jump Markov Chain Monte Carlo (MCMC) methods. The family of Coxian distributions is a very flexible mixture model that can capture the special features frequently observed in insurance claims. Furthermore, given the proposed Coxian approximation, it is possible to obtain closed expressions of the Laplace transforms of the total claim count and the total claim amount random variables. These properties allow us to obtain Bayesian estimations of the distributions of the number of claims and the total claim amount in a future time period, their main characteristics and credible intervals. The possibility of applying deductibles and maximum limits is also analyzed. The methodology is illustrated with a real data set provided by the insurance department of an international commercial company.

Some light-traffic and heavy-traffic results for the GI/G/n/0 queue using the GM Heuristic

In this paper we carry out both light-traffic and heavy-traffic analyses for the calculation of steady-state loss probabilities in the general multi-server queuing loss system, the GI/G/n/0 queue. The analysis makes use of a heuristic approach called the GM Heuristic, for which a detailed analysis in normal traffic has previously been published. Sufficient conditions are given for the GM Heuristic to be asymptotically exact in light traffic. The heuristic is also shown to be asymptotically exact in heavy-traffic when the number of servers n tends to infinity. These results are illustrated numerically using two-phase Coxian distributions for both the inter-arrival time and service time.

Two new heuristics for the GI/G/n/0 queueing loss system with examples based on the two-phase Coxian distribution

In this paper, we introduce a new heuristic approach for the numerical analysis of queueing systems. In particular, we study the general, multi-server queueing loss system, the GI/G/n/0 queue, with an emphasis on the calculation of steady-state loss probabilities. Two new heuristics are developed, called the GM Heuristic and the MG Heuristic, both of which make use of an exact analysis of the corresponding single-server GI/G/1/0 queue. The GM Heuristic also uses an exact analysis of the GI/M/n/0 queue, while the MG Heuristic uses an exact analysis of the M/G/n/0 queue. Experimental results are based on the use of two-phase Coxian distributions for both the inter-arrival time and the service time; these include an error analysis for each heuristic and the derivation of experimental probability bounds for the loss probability. For the class of problems studied, it is concluded that there are likely to be many situations where the accuracy of the GM Heuristic is adequate for practical purposes. Methods are also developed for combining the GM and MG Heuristics. In some cases, this leads to approximations that are significantly more accurate than those obtained by the individual heuristics.

Bayesian estimation of finite time ruin probabilities

AbstractIn this paper, we consider Bayesian inference and estimation of finite time ruin probabilities for the Sparre Andersen risk model. The dense family of Coxian distributions is considered for the approximation of both the inter‐claim time and claim size distributions. We illustrate that the Coxian model can be well fitted to real, long‐tailed claims data and that this compares well with the generalized Pareto model. The main advantage of using the Coxian model for inter‐claim times and claim sizes is that it is possible to compute finite time ruin probabilities making use of recent results from queueing theory. In practice, finite time ruin probabilities are much more useful than infinite time ruin probabilities as insurance companies are usually interested in predictions for short periods of future time and not just in the limit. We show how to obtain predictive distributions of these finite time ruin probabilities, which are more informative than simple point estimations and take account of model and parameter uncertainty. We illustrate the procedure with simulated data and the well‐known Danish fire loss data set. Copyright © 2009 John Wiley & Sons, Ltd.

Efficient duration and hierarchical modeling for human activity recognition

A challenge in building pervasive and smart spaces is to learn and recognize human activities of daily living (ADLs). In this paper, we address this problem and argue that in dealing with ADLs, it is beneficial to exploit both their typical duration patterns and inherent hierarchical structures. We exploit efficient duration modeling using the novel Coxian distribution to form the Coxian hidden semi-Markov model (CxHSMM) and apply it to the problem of learning and recognizing ADLs with complex temporal dependencies. The Coxian duration model has several advantages over existing duration parameterization using multinomial or exponential family distributions, including its denseness in the space of nonnegative distributions, low number of parameters, computational efficiency and the existence of closed-form estimation solutions. Further we combine both hierarchical and duration extensions of the hidden Markov model (HMM) to form the novel switching hidden semi-Markov model (SHSMM), and empirically compare its performance with existing models. The model can learn what an occupant normally does during the day from unsegmented training data and then perform online activity classification, segmentation and abnormality detection. Experimental results show that Coxian modeling outperforms a range of baseline models for the task of activity segmentation. We also achieve a recognition accuracy competitive to the current state-of-the-art multinomial duration model, while gaining a significant reduction in computation. Furthermore, cross-validation model selection on the number of phases K in the Coxian indicates that only a small K is required to achieve the optimal performance. Finally, our models are further tested in a more challenging setting in which the tracking is often lost and the activities considerably overlap. With a small amount of labels supplied during training in a partially supervised learning mode, our models are again able to deliver reliable performance, again with a small number of phases, making our proposed framework an attractive choice for activity modeling.

An Algorithm for Computing Minimal% Coxian Representations

This paper presents an algorithm for computing minimal ordered Coxian representations of phase-type distributions whose Laplace-Stieltjes transform has only real poles. We first identify a set of necessary and sufficient conditions for an ordered Coxian representation to be minimal with respect to the number of phases involved. The conditions establish a relationship between the Coxian representations of a Coxian distribution and the derivatives of its distribution function at zero. Based on the conditions, the algorithm is developed. Three numerical examples show the effectiveness of the algorithm and some geometric properties associated with ordered Coxian representations.

Coxian approximations of matrix-exponential distributions

In this paper, we study the approximation of matrix-exponential distributions by Coxian distributions. Based on the spectral polynomial algorithm, we develop an algorithm for computing Coxian representations of Coxian distributions that are approximations of matrix-exponential distributions. As a specialization, we show that phase-type (PH) distributions can be approximated by Coxian distributions. We also show that any phase-type generator with only real eigenvalues is PH-majorized by ordered Coxian generators. Consequently, the algorithm is modified for computing ordered Coxian representations of any phase-type distribution whose Laplace-Stieltjes transform has only real poles. Numerical examples are presented to show the efficiency of the algorithm and the accuracy of the Coxian approximations. Keywords: Matrix-exponential distribution, Coxian distribution, phase-type distribution, matrix analytic methods, Perron-Frobenius theory Mathematics subject classification (2000): Primary 60A99, Secondary 15A18

Bayesian prediction of the transient behaviour and busy period in short- and long-tailed [formula omitted] queueing systems

Bayesian inference for the transient behaviour and duration of a busy period in a single server queueing system with general, unknown distributions for the interarrival and service times is investigated. Both the interarrival and service time distributions are approximated using the dense family of Coxian distributions. A suitable reparameterization allows the definition of a non-informative prior, and Bayesian inference is then undertaken using reversible jump Markov chain Monte Carlo methods. An advantage of the proposed procedure is that heavy-tailed interarrival and service time distributions such as the Pareto can be well approximated. The proposed procedure for estimating the system measures is based on recent theoretical results for the Coxian/Coxian/1 system. A numerical technique is developed for every MCMC iteration so that the transient queue length and waiting time distributions and the duration of a busy period can be estimated. The approach is illustrated with both simulated and real data.

Lattice path approach for busy period density of M/G/1 queues using C3 Coxian distribution

The paper aims at busy period analysis of non-Markovian queuing system M/G/1 starting initially with i0 customers, through lattice path approach. The general distribution involved is approximated by a 3-phase Cox distribution, C3. Distributions having rational Laplace–Stieltjes transforms and square coefficient of variation lying in [1/3, ∞) form a very wide class of distributions. As any distribution of this class can be approximated by a C3, that has Markovian property, amenable to the application of lattice paths combinatorial analysis, the use of C3 therefore has led us to achieve results applicable to almost any real life queuing system M/G/1.

On the value function of the M/Cox(r)/1 queue

We consider a single-server queueing system at which customers arrive according to a Poisson process. The service times of the customers are independent and follow a Coxian distribution of order r. The system is subject to costs per unit time for holding a customer in the system. We give a closed-form expression for the average cost and the corresponding value function. The result can be used to derive nearly optimal policies in controlled queueing systems in which the service times are not necessarily Markovian, by performing a single step of policy iteration. We illustrate this in the model where a controller has to route to several single-server queues. Numerical experiments show that the improved policy has a close-to-optimal value.