130,401 research outputs found
Completion problem with partial correlation vines
No full textAbstractThis paper extends the results in [D. Kurowicka, R.M. Cooke, A parametrization of positive definite matrices in terms of partial correlation vines, Linear Algebra Appl. 372 (2003) 225–251]. We show that a partial correlation vine represents a factorization of the determinant of the correlation matrix. We show that the graph of an incompletely specified correlation matrix is chordal if and only if it can be represented as an m-saturated incomplete vine, that is, an incomplete vine for which all edges corresponding to membership-descendents (m-descendents for short) of a specified edge are specified. This enables us to find the set of completions, and also the completion with maximal determinant for matrices corresponding to chordal graphs
Continuous/discrete non parametric Bayesian belief nets with UNICORN and UNINET
No full textHanea et al. (2006) presented a method for quantifying and computing continuous/discrete non parametric Bayesian Belief Nets (BBN). Influences are represented as conditional rank correlations, and the joint normal copula enables rapid sampling and conditionalization. Further mathematical background is in Kurowicka and Cooke (2007). This article sketches the current stage of development. The driving application currently involves 133 continuous and discrete probabilistic nodes, and 330 functional nodes. Boolean functions enable fault trees to be fully represented as functional nodes in a BBN. Repeated nodes are easily handled with the identity function. Current perspectives and challenges conclude the paper.Delft Institute of Applied MathematicsElectrical Engineering, Mathematics and Computer Scienc
Techniques in Representing High Dimensional Distributions
No full textElectrical Engineering, Mathematics and Computer Scienc
Examination of insurer fees by means of life expectancy
No full textIn the Netherlands, the pension regulation consists of three pillars. The first pillar consists of the basic state pension called AOW (algemene ouderdomswet), which is meant to provide a financial foundation for every person. At the moment, the yearly AOW contribution for a retired person is somewhere between 10.000 and 15.000, depending on whether or not you have children or a partner with income. The second pillar is meant for every employed person as an addition to the AOW pension. Mostly, the employer has an agreement with an insurer, to which he pays a yearly premium for every employee. The insurer then pledges the employee a pension at retirement, amount of which is based on the yearly premium. This premium is partially paid by the employee's salary and partially by the employer. So the employee actually sets aside a part of salary to save for his pension. Every employer has its own regulation for this secondary term of employment. It is however legally obligatory. The third pillar consists of every other form of building up pension. For example, depositing money into a bank or taking additional insurances. In this report we will only examine the second pillar. For convenience, we shall simply call this second pillar; pension. The insurers costs of providing a pension depends on how much and long a person will receive a pension as well as how long this person saves for his/her pension. The goal of this research is to investigate how the main insurers in the Netherlands determine their costs of providing a pension. We will examine two main algorithms that are used to determine insurers’ premium, discuss assumptions used in these algorithms and investigate whether these assumptions are reasonable. The report is organized as follows: First, we discuss the general approach how insurers calculate their premiums which are based on the estimated future expenses of the insurer. The future expenses of a person are determined on basis of how long this person is going to live after his retirement and the interest rates. To calculate how much a person has to pay now to receive a certain amount of pension in the future, the calculated expenses are discounted. Since the money put aside now could be invested, it is worth now less than it will be worth in the future. Discounting requires the choice of the interest rate. This discount rate is based on the expected return of risk-free investments such as government bonds. Because the discount rate have a huge impact on the premium, we will discuss it in this report in Section 2. Every person is different, so the insurer has to estimate the expected life length of a person after retirement. In section 4 the life expectancy is calculated using mortality rates, which provide the probability of death at a certain age. The data about the historical mortality rates of Dutch population is available. However, to calculate future expenses the insurers need the future mortality rates rather than the historical ones. Hence the future mortality rates have to be forecasted. We will examine two algorithms that allow to forecast mortality rates and provide the life expectancy curves that can be obtain with these forecasted mortality rates. In Section 5 the the Royal Actuarial Association (AG) is presented. Section 6 contains explanation of the PLT model (pensioenen lijfrentetafel workgroup). Both models use a similar approach based on the smooth historical data, which is extrapolated to determine a forecast of 50 years. The models do not contain uncertainty of the forecast. An alternative model is proposed by Central Bureau of Statistics (CBS). The CBS model is different from the other two. It is a lot more complex as it includes causes of death and takes medical developments into account. Besides that, it does give an indication of the uncertainty in the forecast, which is a nice advantage. However, due to its complexity, it will not be investigated in this report. Section 8 contains comparison of fees that the insurers use to cover their future expenses. Insurer fees are essentially the premiums for buying 1 euro of pension of an insurer. Insurers determine these fees based on their own methods. Some just use the forecast of one of the earlier mentioned models, while others do their own calculations. Since the fees would be different for each employee in general the simplified calculations are performed by insurers. Instead of looking at every employee separately, the insurers calculate fees for the employee with average age. We will see how these simplified calculations compare to previously discussed ones. Finally we conclude the report and provide the recommendation. The detailed presentation of algorithms discussed in this report are shown in Appendices. Moreover R-code used to make figures and obtained results included in this report is included in the end.Applied mathematicsElectrical Engineering, Mathematics and Computer Scienc
MeSH term explosion and author rank improve expert recommendations
No full textInformation overload is an often-cited phenomenon that reduces the productivity, efficiency and efficacy of scientists. One challenge for scientists is to find appropriate collaborators in their research. The literature describes various solutions to the problem of expertise location, but most current approaches do not appear to be very suitable for expert recommendations in biomedical research. In this study, we present the development and initial evaluation of a vector space model-based algorithm to calculate researcher similarity using four inputs: 1) MeSH terms of publications; 2) MeSH terms and author rank; 3) exploded MeSH terms; and 4) exploded MeSH terms and author rank. We developed and evaluated the algorithm using a data set of 17,525 authors and their 22,542 papers. On average, our algorithms correctly predicted 2.5 of the top 5/10 coauthors of individual scientists. Exploded MeSH and author rank outperformed all other algorithms in accuracy, followed closely by MeSH and author rank. Our results show that the accuracy of MeSH term-based matching can be enhanced with other metadata such as author rank
Prediction of Market Value of Used Commercial Aircraft
No full textAviation financers are interested in the current/future market value of used commercial aircraft, as this information is precious knowledge for them to support collateral position in the aircraft loan. In this paper, variables from general economy, airline industry and aviation fleet are explored to find out the factors predictive for used aircraft market. Two statistical methods- principal component regression and copula-are applied for building the prediction model of an aircraft.Risk and Environmental Modeling groupApplied mathematicsElectrical Engineering, Mathematics and Computer Scienc
Gas explosion model for grey cast iron pipes in Amsterdam
No full textThis thesis concerns grey cast iron gas pipelines in Amsterdam managed by the regional gas and electricity grid operator Liander. Due to brittle nature and ageing process of grey cast iron, probability of explosion in this type of pipelines seems to be increasing. A model for the gas explosion probability is proposed. Multiple factors are taken into consideration, including pipe’s age, length, wall thickens and environmental characteristics (e.g. pH, distance to nearby property) The model is a tool which provides assessment of the probability of explosion in grey cast iron pipes in Amsterdam. It also counts the contribution of each type of leakage to overall explosion probability. The model works for different pipes sub-populations (e.g. with respect to post code area, type of pressure, age). Numbers presented in this thesis, coming from real databases, have been altered due to confidentiality reasons.Risk and Environmental modellingElectrical Engineering, Mathematics and Computer ScienceElectrical Engineering, Mathematics and Computer Scienc
The risk of exceeding a certain threshold for Chlorohyll-a in the Wadden Sea
No full textThe Ecostress project has the intention to develop systems and methodologies to observe changes in urban coastal areas. Within this project, the Wadden Sea is taken as a pilot area. The goal is to produce tools to assess human, social and environmental impacts, such as the spatial risk and socio-environmental impacts of coastal urbanization. One of the aspired tools is an interactive risk map for the amounts of Chlorophyll-a (Chlfa) in the water, as an indicator of algae, where the month of interest and threshold can be varied. The aim of this thesis is to provide data for this tool. Information about the Chlfa was generated by the GEM/BLOOM-model, developed at Deltares. Once the data from the GEM/BLOOM was available, different methodologies were investigated to execute the risk assessment. In the end the fitting of an ECDF was conducted. This seemed to give proper results. In an attempt to make the data even more reliable, a specialized fitting for the tail was applied according to extreme value theory.Applied MathematicsMathematical EngineeringElectrical Engineering, Mathematics and Computer Scienc
Going Beyond Counting First Authors in Author Co-citation Analysis
No full textThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Regular vines with strongly chordal pattern of (conditional) independence
No full textMultivariate statistical models can be simplified by assuming that a pattern of conditional independence is presented in the given data. A popular way of capturing the (conditional) independence is to use probabilistic graphical models. The relationship between strongly chordal graphs and m-saturated vines is proved. Moreover, an algorithm to construct an m-saturated vine structure corresponding to strongly chordal graph is provided. This allows the reduction of regular vine copula models complexity. When the underlying data is sparse our approach leads to model estimation improvement when compared with current heuristic methods. Furthermore, due to reduction of model complexity it is possible to evaluate all vine structures as well as to fit non-simplified vines. These advantages have been shown in the simulated and real data examples.1Applied Probabilit
- …
