Deux parties principales composent cette thèse. La première d'entre elles est consacrée à la valvométrie, c'est-à-dire ici l'étude de la distance entre les deux parties de la coquille d'une huître au cours du temps. La valvométrie est utilisée afin de déterminer si de tels animaux sont en bonne santé, pour éventuellement tirer des conclusions sur la qualité de leur environnement. Nous considérons qu'un processus de renouvellement à quatre états sous-tend le comportement des huîtres étudiées. Afin de retrouver ce processus caché dans le signal valvométrique, nous supposons qu'une densité de probabilité reliée à ce signal est bimodale. Nous comparons donc plusieurs estimateurs qui prennent en compte ce type d'hypothèse, dont des estimateurs à noyau.Dans un second temps, nous comparons plusieurs méthodes de régression, dans le but d'analyser des données transcriptomiques. Pour comprendre quelles variables explicatives influent sur l'expression de gènes, nous avons réalisé des tests multiples grâce au modèle linéaire FAMT. La méthode SIR peut être envisagée pour trouver des relations non-linéaires. Toutefois, elle est principalement employée lorsque la variable à expliquer est univariée. Une version multivariée de cette approche a donc été développée. Le coût d'acquisition des données transcriptomiques pouvant être élevé, la taille n des échantillons correspondants est souvent faible. C'est pourquoi, nous avons également étudié la méthode SIR lorsque n est inférieur au nombre de variables explicatives p. / There are two main parts in this thesis. The first one concerns valvometry, which is here the study of the distance between both parts of the shell of an oyster, over time. The health status of oysters can be characterized using valvometry in order to obtain insights about the quality of their environment. We consider that a renewal process with four states underlies the behaviour of the studied oysters. Such a hidden process can be retrieved from a valvometric signal by assuming that some probability density function linked with this signal, is bimodal. We then compare several estimators which take this assumption into account, including kernel density estimators.In another chapter, we compare several regression approaches, aiming at analysing transcriptomic data. To understand which explanatory variables have an effect on gene expressions, we apply a multiple testing procedure on these data, through the linear model FAMT. The SIR method may find nonlinear relations in such a context. It is however more commonly used when the response variable is univariate. A multivariate version of SIR was then developed. Procedures to measure gene expressions can be expensive. The sample size n of the corresponding datasets is then often small. That is why we also studied SIR when n is less than the number of explanatory variables p.

Reliability analysis plays an important role in keeping manufacturers in a competitive position. It can be applied in many areas such as warranty predictions, maintenance scheduling, spare parts provisioning, and risk assessment. This dissertation focuses on statistical modeling and predictions based on lifetime data, degradation data, and recurrent event data. The datasets used in this dissertation come from the field, and have complicated structures. The dissertation consists of three main chapters, in addition to Chapter 1 which is the introduction chapter, and Chapter 5 which is the general conclusion chapter. Chapter 2 consists of the traditional time-to-failure data analysis. We propose a statistical method to address the failure data from an appliance used at home with the consideration of retirement times and delayed reporting time. We also develop a prediction method based on the proposed model. Using the information of retirement-time distribution and delayed reporting time, the predictions are more accurate and useful in the decision making. In Chapter 3, we introduce a nonlinear mixed-effects general path model to incorporate dynamic covariates into degradation data analysis. Dynamic covariates include time-varying environmental variables and usage condition. The shapes of the effect functions of covariates may be constrained to be, for example, monotonically increasing (i.e., higher temperature is likely to cause more damage). Incorporating dynamic covariates with shape restrictions is challenging. A modified alternative algorithm and the corresponding prediction method are proposed. In Chapter 4, we introduce a multi-level trend-renewal process (MTRP) model to describe component-level events in multi-level repairable systems. In particular, we consider two-level repairable systems in which events can occur at the subsystem level, or the component (within the subsystem) level. The main goal is to develop a method for estimation of model parameters and a procedure for prediction of the future replacement events at component level with the consideration of the effects from the subsystem replacement events. To explain unit-to-unit variability, time-dependent covariates as well as random effects are introduced into the heterogeneous MTRP model (HMTRP). A Metropolis-within-Gibbs algorithm is used to estimate the unknown parameters in the HMTRP model. The proposed method is illustrated by a simulated dataset. / Ph. D.

Um die komplexe Dynamik von Neuronen und deren Informationsverarbeitung mittels Pulssequenzen zu verstehen, ist es wichtig, die stationäre Puls-Aktivität zu charakterisieren. Die statistischen Eigenschaften von Pulssequenzen können durch vereinfachte stochastische Neuronenmodelle verstanden werden. Eine gut ausgearbeitete Theorie existiert für die Klasse der Erneuerungsmodelle, welche die statistische Unabhängigkeit der Interspike-Intervalle (ISI) annimmt. Experimente haben jedoch gezeigt, dass viele Neuronen Korrelationen zwischen ISIs aufweisen und daher nicht gut durch einen Erneuerungsprozess beschrieben werden. Solche Korrelationen können durch Nichterneuerungs-Modelle erfasst werden, welche jedoch theoretisch schlecht verstanden sind. Diese Arbeit ist eine analytische Studie von Nichterneuerungs-Modellen, die zwei bedeutende Korrelationsmechanismen untersucht: farbiges Rauschen, welches zeitlich-korrelierten Input darstellt, und negative Puls-Rückkopplung, welche Feuerraten-Adaption realisiert. Für das "Perfect-Integrate-and-Fire" (PIF) Modell, welchen durch ein allgemeines Gauss''sches farbiges Rauschen getrieben ist, werden die Statistiken höherer Ordnung der Output-Pulssequenz hergeleitet, insbesondere der Koeffizient der Variation, der serielle Korrelationskoeffizient (SCC), die ISI-Dichte und der Fano-Faktor. Weiterhin wird die Dynamik des PIF Modells mit Puls-getriggertem Adaptionsstrom und weissem Stromrauschen im Detail analysiert. Die Theorie liefert einen Ausdruck für den SCC, der für schwaches Rauschen aber beliebige Adaptions-Stärke und Zeitskale gültig ist, sowie die lineare Antwortfunktion und das Leistungsspektrum der Pulssequenz. Ausserdem wird gezeigt, dass ein stochastischer Adaptionsstrom wie ein langsames farbiges Rauschen wirkt, was ermöglicht, die dominierende Quellen des Rauschen in einer auditorischen Rezeptorzelle zu bestimmen. Schliesslich wird der SCC für das fluktuations-getriebene Feuerregime berechnet. / To understand the complex dynamics of neurons and its ability to process information using a sequence of spikes, it is vital to characterize its stationary spontaneous spiking activity. The statistical properties of spike trains can be explained by reduced stochastic neuron models that account for various sources of noise. A well-developed theory exists for the class of renewal models, in which the interspike intervals (ISIs) are statistically independent. However, experimental studies show that many neurons are not well described by a renewal process because of correlations between ISIs. Such correlations can be captured by generalized, non-renewal models, which are, however, poorly understood theoretically. This thesis represents an analytical study of non-renewal models, focusing on two prominent correlation mechanisms: colored-noise driving representing temporally correlated inputs, and negative feedback currents realizing spike-frequency adaptation. For the perfect integrate-and-fire (PIF) model driven by a general Gaussian colored noise input, the higher-order statistics of the output spike train is derived using a weak-noise analysis of the Fokker-Planck equation. This includes formulas for the coefficient of variation, the serial correlation coefficient (SCC), the ISI density and the Fano factor. Then, the dynamics of a PIF model with a spike-triggered adaptation and a white-noise current is analyzed in detail. The theory yields an expression for the SCC valid for weak noise but arbitrary adaptation strengths and time scale, and also provides the linear response to time-dependent stimuli and the spike train power spectrum. Furthermore, it is shown that a stochastic adaptation current acts like a slow colored noise, which permits to determine the source of spiking variability observed in an auditory receptor neuron. Finally, the SCC is calculated for the fluctuation-driven spiking regime by assuming discrete states of colored noise or adaptation current.

Characterization of distributions has been an important topic in statistical theory for decades. Although there have been many well known results already developed, it is still of great interest to find new characterizations of commonly used distributions in application, such as normal or gamma distribution. In practice, sometimes we make guesses on the distribution to be fitted to the data observed, sometimes we use the characteristic properties of those distributions to do so. In this paper we will restrict our attention to the characterizations of gamma distribution as well as some related studies on the corresponding parameter estimation based on the characterization properties. Some simulation studies are also given.

This thesis presents a study of stochastic models of continuous review of inventory systems of perishable and non-perishable products, as well as inventory systems operating in random environment. It contains five chapters. The first chapter is introductory in nature, containing the motivation for the study and the techniques required for the analysis of respective models described in the remaining chapters. Chapter 2 provides a model of perishable product inventory system operating in a random environment. For the sake of simplicity, the stochastic environment is considered to alternate randomly over time between two states 0 and 1 according to an alternating renewal process. When the environment is in state k, the items in inventory have a perishable rate ìk, the demand rate is ëk and the replenishment cost is CRk. The performance of various measures of the system evolution are obtained, assuming instantaneous replenishment at the epoch of the first demand after the stock-out and associating a Markov renewal process with the inventory level. In Chapter 3, a continuous review single product perishable inventory model is considered. Items deteriorate in two phases and then perish. Independent demands occur at constant rates for items in phase I and in phase II. Demand that occurs for an item in phase I during its stock-out period is satisfied by an item in phase II with some probability. However a demand for an item in phase II occurring during its stock-out period is lost. The reordering policy is an adjustable (S,s) policy with the lead-time following an arbitrary distribution. Identifying the stochastic process as a renewal process, the probability distribution of the inventory level at any arbitrary instant of time is obtained. The expressions for the mean stationary rates of demands lost, demands substituted, perished units and scrapped units are also derived. A numerical example is considered to highlight the results obtained. Chapter 4 is a study of a two-commodity inventory system under continuous review. The maximum storage capacity for the i-th item is Si (i=1, 2). The demand points for each commodity are assumed to form an independent Poisson process, with unit demand for one item and bulk demand for the other. The order level is fixed as si for the i-th commodity (i=1, 2) and the ordering policy is to place an order for Qi (= Si – si , i = 1,2) items for the i-the commodity when both the inventory levels are less than or equal to their respective reorder levels. The lead-time is assumed to be exponential. The joint probability distribution for both commodities is obtained in both transient and steady state cases. Various measures of systems performance and the total expected cost rate in the steady state are derived. The results are illustrated with numerical examples. Chapter 5 provides an analysis of a continuous review of two-product system with two types of demands and with individual (S,s) ordering policy. The lead-time distribution of product 1 is arbitrary and that of product 2 exponential. Two types of demands occur at constant rates either for both products or for product 2 alone. Expressions for the stationary distribution of the inventory level are obtained by identifying the underlying stochastic processes as a semi-regenerative process. The mean stationary rates of the lost demands, the demands that are satisfied and the number of reorders are obtained and these measures are used to provide an expression for the cost rate. The main objective of this thesis is to improve the state of art of continuous review inventory systems. The salient features of the thesis are summarized below: <ul>(a) Consideration of</ul> <ul>(i) The impact of the stochastic environment on inventory systems;</ul> <ul>(ii) The interactions existing among the products in multi-product systems;</ul> <ul>(iii) Individual and joint-ordering policies;</ul> <ul>(b) Discussion of inventory systems with perishable products; </ul> <ul>(c) Effective use of the regeneration point technique to derive expressions for various system measures;</ul> <ul>(d) Illustration of the various results by extensive numerical work;</ul> <ul>(e) Relevant optimization problems</ul> / Thesis (PhD (Industrial Engineering))--University of Pretoria, 2006. / Industrial and Systems Engineering / unrestricted