Dans le cadre des systèmes de transport intelligents (ITS), cette thèse concerne la conception d'une méthodologie de prédiction, en temps réel et pour différents horizons, du temps de parcours à partir des données de vitesse et de débit d'une route instrumentée. Pour atteindre cet objectif, deux approches sont considérées dans cette thèse. La première approche, dite « sans modèle », utilise exclusivement des mesures de vitesse. Grâce à l'utilisation astucieuse des données historiques, nous avons résolu le problème de prédiction comme étant un problème de filtrage. Pour ce faire, des données historiques sont utilisées pour construire des pseudo-observations qui alimentent un filtre de Kalman adaptatif (AKF). Sous une hypothèse de Gaussianité, les statistiques du bruit de processus sont estimées en temps-réel, tandis que les statistiques du pseudo-bruit d'observation sont déduites des données historiques adéquatement classées. La seconde approche, dite ‘'basée-modèle'', utilise principalement des mesures de débit et de vitesse. Contrairement à la précédente approche où la résolution spatiale est fixée par l'emplacement des capteurs, une discrétisation spatiale plus fine est considérée. Celle-ci s'avère possible grâce à l'utilisation du modèle CTM (Cell Transmission Model). Un observateur d'état commuté, de type Luenberger, permet d'estimer les états internes (densités des cellules). En utilisant uniquement les prédictions des débits des conditions frontières via une approche de type AKF similaire à celle développée dans la première approche, le modèle CTM contraint permet de prédire les densités des cellules et d'en déduire les vitesses et le temps de parcours. Les méthodes développées ont été validées expérimentalement en considérant la rocade sud grenobloise comme cas d'étude. Les résultats montrent que les deux méthodes présentent de bonnes performances de prédiction. Les méthodes proposées performent mieux que celles basées sur une utilisation directe des moyennes historiques. Pour l'ensemble des données considérées, l'étude a également montré que l'approche ‘'basée modèle‘' est plus adaptée pour des horizons de prédictions de moins de 30 min. / This dissertation falls within the domain of the Intelligent Transportation Systems (ITS). In particular, it is concerned with the design of a methodology for the real-time multi-step ahead travel time forecasting using flow and speed measurements from a instrumented freeway. To achieve this objective this thesis develops two main methodologies. The first one, a model-free, uses only speed measurements collected from the freeway, where a mean speed is assumed between two consecutive collection points. The travel time is forecasted using a noise Adaptive Kalman Filter (AKF) approach. The process noise statistics are computed using an online unbiased estimator, while the observations and their noise statistics are computed using the clustered historical traffic data. Forecasting problems are reformulated as filtering ones through the use of pseudo-observations built from historical data. The second one, a model-based, uses mainly traffic flow measurements. Its main appealing is the use of a mathematical model in order to reconstruct the internal state (density) in small road portions, and consequently exploits the relation between density and speed to forecast the travel time. The methodology uses only boundary conditions as inputs to a switched Luenberger state observer, based on the ``Cell Transmission Model'' (CTM), to estimate the road initial states. The boundary conditions are then forecasted using the AKF developed above. Consequently, the CTM model is run using the initial conditions and the forecasted boundaries in order to obtain the future evolution of densities, speeds, and finally travel time. The added innovation in this approach is the space discretization achieved: indeed, portions of the road, called ``cells'', can be chosen as small as desired and thus allow obtaining a finer tracking of speed variations. In order to validate experimentally the developed methodologies, this thesis uses as study case the Grenoble South Ring. This freeway, enclosing the southern part of the city from A41 to A480, consists of two carriageways with two lanes. For this study only the direction east-west was considered. With a length of about 10.5 km, this direction has 10 on-ramps, 7 off-ramps, and is monitored through the Grenoble Traffic Lab (GTL) that is able to provide reliable traffic data every 15 s, which makes it possible for the forecasting strategies to be validated in real-time. The results show that both methods present strong capabilities for travel time forecasting: considering the entire freeway, in 90% of the cases it was obtained a maximum forecasting error of 25% up to a forecasting horizon of 45 min. Furthermore, both methods perform as good as, or better than, the average historical. In particular, it is obtained that for horizons larger than 45 min, the forecasting depended exclusively on the historical data. For the dataset considered, the assessment study also showed that the model-based approach was more suitable for horizons shorter than 30 min.

Wage dispersion is one of the social and economic issues arousing public concern in Hong Kong. There are many studies exploring the possible causes and changes in wage dispersion. They often focus on the study of summary measures such as Gini and Theil indexes, or adopt OLS-based regression approach. In foreign studies on wage dispersion, Oaxaca-Blinder decomposition, originated from Oaxaca (1974) and Blinder (1973), is a common method of decomposing changes or differences in mean wages between two groups into wage structure effect and composition effect, and then further decomposing the two effects into contributions of each control variable. Nevertheless, focusing on summary measures or decomposing mean wages can just give people an insight into the causes and changes in general wage dispersion but not the entire wage distribution. As pointed out by Chi, Li and Yu (2007), the estimation of the entire wage distribution and decomposition of the distributional changes in wage dispersion has been attracting the attention of labour economists. This thesis adopts a distributional approach proposed by Firpo, Fortin and Lemieux (2007) to study the changes in wage dispersion of Hong Kong since 1980s. The FFL approach comprises a two-stage procedure. Firstly, changes in dispersion are divided into wage structure effect and composition effect without directly estimating a wage-setting model. This is done by doing a proper reweighting to obtain counterfactual wage vectors. Kernel density estimation is used for visualizing the wage distribution in different years and the counterfactuals; secondly, novel recentered influence function (RIF) regressions across quantiles are performed to further decompose the two effects into contributions of each control variable. The findings are outlined as follows: first, there was an increase in wage dispersion over the whole wage distribution from 1980s but a decrease from 2001 to 2006; second, the composition effect dominates the wage structure effect over years; third, changes in the distribution of characteristics and the returns to these characteristics are highly responsive to each other, suggesting that our labour market is highly responsive to structural changes; fourth, The common wage-determining factors may not be able to explain the earnings-profile of low wage earners well. In brief, the development of the economy since 1980s increased the wage dispersion over years. Nevertheless, the economic downturn due to external shocks and internal unfavourable events and general skill-upgrading in labour-intensive industries decreased the wage dispersion since 2000s.

Understanding spatial perspectives on the spread and incidence of a disease is invaluable for public health planning and intervention. Choropleth maps are commonly used to provide an abstraction of disease risk across geographic space. These maps are derived from aggregated population counts that are known to be affected by the small numbers problem. Kernel density estimation methods account for this problem by producing risk estimates that are based on aggregations of approximately equal population sizes. However, the process of aggregation often combines data from areas with non-uniform spatial and population characteristics. This thesis presents a new method to aggregate space in ways that are sensitive to their underlying risk factors. Such maps will enable better public health practice and intervention by enhancing our ability to understand the spatial processes that result in disparate health outcomes.

abstract: Infants born before 37 weeks of pregnancy are considered to be preterm. Typically, preterm infants have to be strictly monitored since they are highly susceptible to health problems like hypoxemia (low blood oxygen level), apnea, respiratory issues, cardiac problems, neurological problems as well as an increased chance of long-term health issues such as cerebral palsy, asthma and sudden infant death syndrome. One of the leading health complications in preterm infants is bradycardia - which is defined as the slower than expected heart rate, generally beating lower than 60 beats per minute. Bradycardia is often accompanied by low oxygen levels and can cause additional long term health problems in the premature infant.The implementation of a non-parametric method to predict the onset of brady- cardia is presented. This method assumes no prior knowledge of the data and uses kernel density estimation to predict the future onset of bradycardia events. The data is preprocessed, and then analyzed to detect the peaks in the ECG signals, following which different kernels are implemented to estimate the shared underlying distribu- tion of the data. The performance of the algorithm is evaluated using various metrics and the computational challenges and methods to overcome them are also discussed. It is observed that the performance of the algorithm with regards to the kernels used are consistent with the theoretical performance of the kernel as presented in a previous work. The theoretical approach has also been automated in this work and the various implementation challenges have been addressed. / Dissertation/Thesis / Masters Thesis Electrical Engineering 2020

This thesis analyzes the end-user quality of service (QoS) in cellular mobile networks using device-side measurements. Quality of service in a wireless network is a significant factor in determining a user's satisfaction. Customers' perception of poor QoS is one of the core sources of customer churn for telecommunications companies. A core focus of this work is on assessing how user density impacts QoS within cellular networks. Kernel density estimation is used to produce user density estimates for high, medium, and low density areas. The QoS distributions are then compared across these areas. The k-sample Anderson-Darling test is used to determine the degree to which user densities vary over time. In general, it is shown that users in higher density areas tend to experience overall lower QoS levels than those in lower density areas, even though these higher density areas service more subscribers. The conducted analyses highlight the value of mobile device-side QoS measurements in augmenting traditional network-side QoS measurements. / Graduate

Lagrangian particle models are useful for modelling pollutants in the atmosphere. They simulate the spread of pollutants by modelling trajectories of individual particles. However, to be useful, these models require a density estimate. The standard method to use has been boxcounting, but kernel density estimator (KDE) is an alternative. How KDE is used varies as there is no standard implementation. Primarily, it is the choice of kernel and bandwidth estimator that determines the model. In this report I have implemented a KDE for FOI’s Lagrangian particle model LPELLO. The kernel I have used is a combination between a uniform and Gaussian kernel. Four different bandwidth estimators has been tested, where two are global and two are variable. The first variable bandwidth estimator is based on the age of released particles, and the second is based on the turbulence history of the particles. The methods have then been tested against boxcounting, which by using an exceedingly large number of particles can be seen as the true concentration. The tests indicate that KDE method generally performs better than boxcounting at low particle numbers. The variable bandwidth estimators also performed better than both global bandwidth estimators. To achive a firmer conclusion, more testing is needed. The results indicate that KDE in general, and variable bandwidth estimators in specific, are useful tools for concentration estimate.

Water shortages resulting from increased demand or reduced supply may be addressed, in part, by redirecting recycled water for irrigation, industrial reuse, groundwater recharge, and as effluent discharge returned to streams. Recycled water is an essential component of integrated water management and broader adoption of recycled water will increase water conservation in water-stressed coastal communities. This study examined spatial patterns of recycled water use in Florida in 2009 and 2015 to detect gaps in distribution, quantify temporal change, and identify potential areas for expansion. Databases of recycled water products and distribution centers for Florida in 2009 and 2015 were developed by combining the 2008 and 2012 Clean Water Needs Survey databases with Florida’s 2009 and 2015 Reuse Inventory databases, respectively. Florida increased recycled water production from 674.85 mgd in 2009 to 738.15 mgd in 2015, an increase of 63.30 mgd. The increase was primarily allocated to use in public access areas, groundwater recharge, and industrial reuse, all within the South Florida Water Management District (WMD). In particular, Miami was identified in 2009 as an area of opportunity for recycled water development, and by 2015 it had increased production and reduced the production gap. Overall, South Florida WMD had the largest increase in production of 44.38 mgd (69%), while Southwest Florida WMD decreased production of recycled water by 1.68 mgd, or 3%. Overall increase in use of recycled water may be related to higher demand due to increased population coupled with public programs and policy changes that promote recycled water use at both the municipal and individual level.

An extremal property of normal distributions is that they have the smallest Fisher Information for location among all distributions with the same variance. A new test of normality proposed by Terrell (1995) utilizes the above property by finding that density of maximum likelihood constrained on having the expected Fisher Information under normality based on the sample variance. The test statistic is then constructed as a ratio of the resulting likelihood against that of normality. Since the asymptotic distribution of this test statistic is not available, the critical values for n = 3 to 200 have been obtained by simulation and smoothed using polynomials. An extensive power study shows that the test has superior power against distributions that are symmetric and leptokurtic (long-tailed). Another advantage of the test over existing ones is the direct depiction of any deviation from normality in the form of a density estimate. This is evident when the test is applied to several real data sets. Testing of normality in residuals is also investigated. Various approaches in dealing with residuals being possibly heteroscedastic and correlated suffer from a loss of power. The approach with the fewest undesirable features is to use the Ordinary Least Squares (OLS) residuals in place of independent observations. From simulations, it is shown that one has to be careful about the levels of the normality tests and also in generalizing the results. / Ph. D.

Deux sujet sont traités dans cette thèse: l'agrégation d'estimateurs et la régression sous contrainte de formes.La régression sous contrainte de forme étudie le problème de régression (trouver la fonction qui représente un nuage de points),avec la contrainte que la fonction en question possède une forme spécifique.Par exemple, cette fonction peut être croissante ou convexe: ces deux contraintes de forme sont les plus étudiées. Nous étudions en particulier deux estimateurs: un estimateur basé sur des méthodes d'agrégation et l'estimateur des moindres carrés avec une contrainte de forme convexe. Des inégalités d'oracle sont obtenues, et nous construisons aussi des intervalles de confiance honnêtes et adaptatifs.L'agrégation d'estimateurs est le problème suivant. Lorsque plusieurs méthodes sont proposées pour le même problème statistique, comment construire une nouvelle méthode qui soit aussi performante que la meilleure parmi les méthodes proposées? Nous étudierons ce problème dans trois contextes: l'agrégation d'estimateurs de densité, l'agrégation d'estimateurs affines et l'aggrégation sur le chemin de régularisation du Lasso. / This PhD thesis studies two fields of Statistics: Aggregation of estimatorsand shape constrained regression.Shape constrained regression studies the regression problem (find a function that approximates well a set of points) with an underlying shape constraint, that is, the function must have a specific "shape". For instance, this function could be nondecreasing of convex: These two shape examples are the most studied. We study two estimators: an estimator based on aggregation methods and the Least Squares estimator with a convex shape constraint. Oracle inequalities are obtained for both estimators, and we construct confidence sets that are adaptive and honest.Aggregation of estimators studies the following problem. If several methods are proposed for the same task, how to construct a new method that mimics the best method among the proposed methods? We will study these problems in three settings: aggregation of density estimators, aggregation of affine estimators and aggregation on the regularization path of the Lasso.

京都大学 / 新制・課程博士 / 博士(経済学) / 甲第24375号 / 経博第662号 / 新制||経||302(附属図書館) / 京都大学大学院経済学研究科経済学専攻 / (主査)教授 西山 慶彦, 教授 江上 雅彦, 講師 柳 貴英 / 学位規則第4条第1項該当 / Doctor of Economics / Kyoto University / DFAM