On the design of customized risk measures in insurance, the problem of capital allocation and the theory of fluctuations for Lévy processes

Dans cette thèse, nous étudions quelques problèmes fondamentaux en mathématiques financières et actuarielles, ainsi que leurs applications. Cette thèse est constituée de trois contributions portant principalement sur la théorie de la mesure de risques, le problème de l’allocation du capital et la théorie des fluctuations. Dans le chapitre 2, nous construisons de nouvelles mesures de risque cohérentes et étudions l’allocation de capital dans le cadre de la théorie des risques collectifs. Pour ce faire, nous introduisons la famille des "mesures de risque entropique cumulatifs" (Cumulative Entropic Risk Measures). Le chapitre 3 étudie le problème du portefeuille optimal pour le Entropic Value at Risk dans le cas où les rendements sont modélisés par un processus de diffusion à sauts (Jump-Diffusion). Dans le chapitre 4, nous généralisons la notion de "statistiques naturelles de risque" (natural risk statistics) au cadre multivarié. Cette extension non-triviale produit des mesures de risque multivariées construites à partir des données financiéres et de données d’assurance. Le chapitre 5 introduit les concepts de "drawdown" et de la "vitesse d’épuisement" (speed of depletion) dans la théorie de la ruine. Nous étudions ces concepts pour des modeles de risque décrits par une famille de processus de Lévy spectrallement négatifs.

The use of flexible quantile-based measures in risk assessment

A new family of distortion risk measures -GlueVaR- is proposed in Belles- Sampera et al. -2013- to procure a risk assessment lying between those provided by common quantile-based risk measures. GlueVaR risk measures may be expressed as a combination of these standard risk measures. We show here that this relationship may be used to obtain approximations of GlueVaR measures for general skewed distribution functions using the Cornish-Fisher expansion. A subfamily of GlueVaR measures satisfies the tail-subadditivity property. An example of risk measurement based on real insurance claim data is presented, where implications of tail-subadditivity in the aggregation of risks are illustrated.

Multistep stochastic mirror descent for risk-averse convex stochastic programs based on extended polyhedral risk measures

We consider risk-averse convex stochastic programs expressed in terms of extended polyhedral risk measures. We derive computable con dence intervals on the optimal value of such stochastic programs using the Robust Stochastic Approximation and the Stochastic Mirror Descent (SMD) algorithms. When the objective functions are uniformly convex, we also propose a multistep extension of the Stochastic Mirror Descent algorithm and obtain con dence intervals on both the optimal values and optimal solutions. Numerical simulations show that our con dence intervals are much less conservative and are quicker to compute than previously obtained con dence intervals for SMD and that the multistep Stochastic Mirror Descent algorithm can obtain a good approximate solution much quicker than its nonmultistep counterpart. Our con dence intervals are also more reliable than asymptotic con dence intervals when the sample size is not much larger than the problem size.

Coreferentiality: A New Method for the Hypothesis-Based Analysis of Phenotypes Characterized by Multivariate Data

Many multifactorial biologic effects, particularly in the context of complex human diseases, are still poorly understood. At the same time, the systematic acquisition of multivariate data has become increasingly easy. The use of such data to analyze and model complex phenotypes, however, remains a challenge. Here, a new analytic approach is described, termed coreferentiality, together with an appropriate statistical test. Coreferentiality is the indirect relation of two variables of functional interest in respect to whether they parallel each other in their respective relatedness to multivariate reference data, which can be informative for a complex effect or phenotype. It is shown that the power of coreferentiality testing is comparable to multiple regression analysis, sufficient even when reference data are informative only to a relatively small extent of 2.5%, and clearly exceeding the power of simple bivariate correlation testing. Thus, coreferentiality testing uses the increased power of multivariate analysis, however, in order to address a more straightforward interpretable bivariate relatedness. Systematic application of this approach could substantially improve the analysis and modeling of complex phenotypes, particularly in the context of human study where addressing functional hypotheses by direct experimentation is often difficult.

Bayesian classification criterion for forensic multivariate data

This study presents a classification criteria for two-class Cannabis seedlings. As the cultivation of drug type cannabis is forbidden in Switzerland, law enforcement authorities regularly ask laboratories to determine cannabis plant's chemotype from seized material in order to ascertain that the plantation is legal or not. In this study, the classification analysis is based on data obtained from the relative proportion of three major leaf compounds measured by gas-chromatography interfaced with mass spectrometry (GC-MS). The aim is to discriminate between drug type (illegal) and fiber type (legal) cannabis at an early stage of the growth. A Bayesian procedure is proposed: a Bayes factor is computed and classification is performed on the basis of the decision maker specifications (i.e. prior probability distributions on cannabis type and consequences of classification measured by losses). Classification rates are computed with two statistical models and results are compared. Sensitivity analysis is then performed to analyze the robustness of classification criteria.

Inequalities for a new data-based method for selecting nonparametric density estimates

We continue the development of a method for the selection of a bandwidth or a number of design parameters in density estimation. We provideexplicit non-asymptotic density-free inequalities that relate the $L_1$ error of the selected estimate with that of the best possible estimate,and study in particular the connection between the richness of the classof density estimates and the performance bound. For example, our methodallows one to pick the bandwidth and kernel order in the kernel estimatesimultaneously and still assure that for {\it all densities}, the $L_1$error of the corresponding kernel estimate is not larger than aboutthree times the error of the estimate with the optimal smoothing factor and kernel plus a constant times $\sqrt{\log n/n}$, where $n$ is the sample size, and the constant only depends on the complexity of the family of kernels used in the estimate. Further applications include multivariate kernel estimates, transformed kernel estimates, and variablekernel estimates.

Troponin-based risk stratification of patients with acute nonmassive pulmonary embolism: systematic review and metaanalysis.

BACKGROUND: Controversy exists regarding the usefulness of troponin testing for the risk stratification of patients with acute pulmonary embolism (PE). We conducted an updated systematic review and a metaanalysis of troponin-based risk stratification of normotensive patients with acute symptomatic PE. The sources of our data were publications listed in Medline and Embase from 1980 through April 2008 and a review of cited references in those publications. METHODS: We included all studies that estimated the relation between troponin levels and the incidence of all-cause mortality in normotensive patients with acute symptomatic PE. Two reviewers independently abstracted data and assessed study quality. From the literature search, 596 publications were screened. Nine studies that consisted of 1,366 normotensive patients with acute symptomatic PE were deemed eligible. Pooled results showed that elevated troponin levels were associated with a 4.26-fold increased odds of overall mortality (95% CI, 2.13 to 8.50; heterogeneity chi(2) = 12.64; degrees of freedom = 8; p = 0.125). Summary receiver operating characteristic curve analysis showed a relationship between the sensitivity and specificity of troponin levels to predict overall mortality (Spearman rank correlation coefficient = 0.68; p = 0.046). Pooled likelihood ratios (LRs) were not extreme (negative LR, 0.59 [95% CI, 0.39 to 0.88]; positive LR, 2.26 [95% CI, 1.66 to 3.07]). The Begg rank correlation method did not detect evidence of publication bias. CONCLUSIONS: The results of this metaanalysis indicate that elevated troponin levels do not adequately discern normotensive patients with acute symptomatic PE who are at high risk for death from those who are at low risk for death.

Identification of Network-Based Risk Factors Associated with Gonorrhea

The evolving antimicrobial resistance coupled with a recent increase in incidence highlights the importance of reducing gonococcal transmission. Establishing novel risk factors associated with gonorrhea facilitates the development of appropriate prevention and disease control strategies. Sexual Network Analysis (NA), a novel research technique used to further understand sexually transmitted infections, was used to identify network-based risk factors in a defined region in Ontario, Canada experiencing an increase in the incidence of gonorrhea. Linear network structures were identified as important reservoirs of gonococcal transmission. Additionally, a significant association between a central network position and gonorrhea was observed. The central participants were more likely to be younger, report a greater number of risk factors, engage in anonymous sex, have multiple sex partners in the past six months and have sex with the same sex. The network-based risk factors identified through sexual NA, serving as a method of analyzing local surveillance data, support the development of strategies aimed at reducing gonococcal spread.

On some aspects of coherent risk measures and their applications

Le sujet principal de cette thèse porte sur les mesures de risque. L'objectif général est d'investiguer certains aspects des mesures de risque dans les applications financières. Le cadre théorique de ce travail est celui des mesures cohérentes de risque telle que définie dans Artzner et al (1999). Mais ce n'est pas la seule classe de mesure du risque que nous étudions. Par exemple, nous étudions aussi quelques aspects des "statistiques naturelles de risque" (en anglais natural risk statistics) Kou et al (2006) et des mesures convexes du risque Follmer and Schied(2002). Les contributions principales de cette thèse peuvent être regroupées selon trois axes: allocation de capital, évaluation des risques et capital requis et solvabilité. Dans le chapitre 2 nous caractérisons les mesures de risque avec la propriété de Lebesgue sur l'ensemble des processus bornés càdlàg (continu à droite, limité à gauche). Cette caractérisation nous permet de présenter deux applications dans l'évaluation des risques et l'allocation de capital. Dans le chapitre 3, nous étendons la notion de statistiques naturelles de risque à l'espace des suites infinies. Cette généralisation nous permet de construire de façon cohérente des mesures de risque pour des bases de données de n'importe quelle taille. Dans le chapitre 4, nous discutons le concept de "bonnes affaires" (en anglais Good Deals), pour notamment caractériser les situations du marché où ces positions pathologiques sont présentes. Finalement, dans le chapitre 5, nous essayons de relier les trois chapitres en étendant la définition de "bonnes affaires" dans un cadre plus large qui comprendrait les mesures de risque analysées dans les chapitres 2 et 3.

Forecasting with Bayesian multivariate vintage-based VARs

We consider the forecasting of macroeconomic variables that are subject to revisions, using Bayesian vintage-based vector autoregressions. The prior incorporates the belief that, after the first few data releases, subsequent ones are likely to consist of revisions that are largely unpredictable. The Bayesian approach allows the joint modelling of the data revisions of more than one variable, while keeping the concomitant increase in parameter estimation uncertainty manageable. Our model provides markedly more accurate forecasts of post-revision values of inflation than do other models in the literature.

Visualization of spirography-based objective measures in Parkinson's disease

Objective: To investigate whether advanced visualizations of spirography-based objective measures are useful in differentiating drug-related motor dysfunctions between Off and dyskinesia in Parkinson’s disease (PD). Background: During the course of a 3 year longitudinal clinical study, in total 65 patients (43 males and 22 females with mean age of 65) with advanced PD and 10 healthy elderly (HE) subjects (5 males and 5 females with mean age of 61) were assessed. Both patients and HE subjects performed repeated and time-stamped assessments of their objective health indicators using a test battery implemented on a telemetry touch screen handheld computer, in their home environment settings. Among other tasks, the subjects were asked to trace a pre-drawn Archimedes spiral using the dominant hand and repeat the test three times per test occasion. Methods: A web-based framework was developed to enable a visual exploration of relevant spirography-based kinematic features by clinicians so they can in turn evaluate the motor states of the patients i.e. Off and dyskinesia. The system uses different visualization techniques such as time series plots, animation, and interaction and organizes them into different views to aid clinicians in measuring spatial and time-dependent irregularities that could be associated with the motor states. Along with the animation view, the system displays two time series plots for representing drawing speed (blue line) and displacement from ideal trajectory (orange line). The views are coordinated and linked i.e. user interactions in one of the views will be reflected in other views. For instance, when the user points in one of the pixels in the spiral view, the circle size of the underlying pixel increases and a vertical line appears in the time series views to depict the corresponding position. In addition, in order to enable clinicians to observe erratic movements more clearly and thus improve the detection of irregularities, the system displays a color-map which gives an idea of the longevity of the spirography task. Figure 2 shows single randomly selected spirals drawn by a: A) patient who experienced dyskinesias, B) HE subject, and C) patient in Off state. Results: According to a domain expert (DN), the spirals drawn in the Off and dyskinesia motor states are characterized by different spatial and time features. For instance, the spiral shown in Fig. 2A was drawn by a patient who showed symptoms of dyskinesia; the drawing speed was relatively high (cf. blue-colored time series plot and the short timestamp scale in the x axis) and the spatial displacement was high (cf. orange-colored time series plot) associated with smooth deviations as a result of uncontrollable movements. The patient also exhibited low amount of hesitation which could be reflected both in the animation of the spiral as well as time series plots. In contrast, the patient who was in the Off state exhibited different kinematic features, as shown in Fig. 2C. In the case of spirals drawn by a HE subject, there was a great precision during the drawing process as well as unchanging levels of time-dependent features over the test trial, as seen in Fig. 2B. Conclusions: Visualizing spirography-based objective measures enables identification of trends and patterns of drug-related motor dysfunctions at the patient’s individual level. Dynamic access of visualized motor tests may be useful during the evaluation of drug-related complications such as under- and over-medications, providing decision support to clinicians during evaluation of treatment effects as well as improve the quality of life of patients and their caregivers. In future, we plan to evaluate the proposed approach by assessing within- and between-clinician variability in ratings in order to determine its actual usefulness and then use these ratings as target outcomes in supervised machine learning, similarly as it was previously done in the study performed by Memedi et al. (2013).

Determination of disease biomarkers in Eucalyptus by comprehensive two-dimensional gas chromatography and multivariate data analysis

In this paper is reported the use of the chromatographic profiles of volatiles to determine disease markers in plants - in this case, leaves of Eucalyptus globulus contaminated by the necrotroph fungus Teratosphaeria nubilosa. The volatile fraction was isolated by headspace solid phase microextraction (HS-SPME) and analyzed by comprehensive two-dimensional gas chromatography-fast quadrupole mass spectrometry (GC. ×. GC-qMS). For the correlation between the metabolic profile described by the chromatograms and the presence of the infection, unfolded-partial least squares discriminant analysis (U-PLS-DA) with orthogonal signal correction (OSC) were employed. The proposed method was checked to be independent of factors such as the age of the harvested plants. The manipulation of the mathematical model obtained also resulted in graphic representations similar to real chromatograms, which allowed the tentative identification of more than 40 compounds potentially useful as disease biomarkers for this plant/pathogen pair. The proposed methodology can be considered as highly reliable, since the diagnosis is based on the whole chromatographic profile rather than in the detection of a single analyte. © 2013 Elsevier B.V..

Knowledge-based risk management: Survey on Brazilian software development enterprises

A risks management, carried on in an effective way, leads the software development to success and may influence on the organization. The knowledge takes part of such a process as a way to help taking decisions. This research aimed to analyze the use of Knowledge Management techniques to the Risk Management in software projects development and the possible influence on the enterprise revenue. It had, as its main studying subject, Brazilian incubated and graduated software developing enterprises. The chosen research method was the Survey type. Multivariate statistical methods were used for the treatment and analysis of the obtained results, this way identifying the most significant factors, that is, enterprise's achievement constraining factors and those outcome achievement ones. Among the latter we highlight the knowledge methodology, the time of existence of the enterprise, the amount of employees and the knowledge externalization. The results encourage contributing actions to the increasing of financial revenue. © 2013 Springer-Verlag.

Age-related differences in time-limit performance and force platform-based balance measures during one-leg stance

Poor posture control has been associated with an increased risk of falls and mobility disability among older adults. This study was conducted to assess the test-retest reliability and sensitivity to group differences regarding the time-limit (TLimit) of one-leg standing and selected balance parameters obtained with a force platform in older and young adults. A secondary purpose was to assess the relationship between TLimit and these balance parameters. Twenty-eight healthy older adults (age: 69±5years) and thirty young adults (age: 21±4years) participated in this study. Two one-leg stance tasks were performed: (1) three trials of 30s maximum and (2) one TLimit trial. The following balance parameters were computed: center of pressure area, RMS sway amplitude, and mean velocity and mean frequency in both the anterio-posterior and medio-lateral directions. All balance parameters obtained with the force platform as well as the TLimit variable were sensitive to differences in balance performance between older and young adults. The test-retest reliability of these measures was found to be acceptable (ICC: 0.40-0.85), with better ICC scores observed for mean velocity and mean frequency in the older group. Pearson correlations coefficients (r) between balance parameters and TLimit ranged from -0.16 to -0.54. These results add to the current literature that can be used in the development of measurement tools for evaluating balance in older and young adults. © 2013 Elsevier Ltd.

Detecting an abnormality in a recovery boiler using dynamic multivariate data analysis with parallel coordinate plots

Data visualization techniques are powerful in the handling and analysis of multivariate systems. One such technique known as parallel coordinates was used to support the diagnosis of an event, detected by a neural network-based monitoring system, in a boiler at a Brazilian Kraft pulp mill. Its attractiveness is the possibility of the visualization of several variables simultaneously. The diagnostic procedure was carried out step-by-step going through exploratory, explanatory, confirmatory, and communicative goals. This tool allowed the visualization of the boiler dynamics in an easier way, compared to commonly used univariate trend plots. In addition it facilitated analysis of other aspects, namely relationships among process variables, distinct modes of operation and discrepant data. The whole analysis revealed firstly that the period involving the detected event was associated with a transition between two distinct normal modes of operation, and secondly the presence of unusual changes in process variables at this time.