A Violência entre Parceiros Íntimos (VPI) tem sido reconhecida como um importante problema de saúde pública e um fator de risco para agravos a saúde de mulheres e crianças. Os serviços de saúde desempenham importante papel na detecção precoce da VPI, especialmente em momentos da vida nos quais se preconiza o atendimento sistematizado, como na gestação e primeira infância. Esta dissertação tem como objetivo principal estimar a probabilidade de ocorrência de Violência Física entre Parceiros Íntimos (VFPI) durante a gestação e/ou pós-parto em população atendida em Unidades Básicas de Saúde (UBS), segundo diferentes características sócio-econômicas e demográficas da clientela. Trata-se de um estudo transversal, realizado com usuárias de 5 UBS da cidade do Rio de Janeiro no ano de 2007. Foram entrevistadas 811 mães de crianças de até cinco meses de idade, que não possuíam nenhuma contra-indicação formal para a amamentação e que relataram ter tido pelo menos uma relação amorosa um mês ou mais durante a gestação ou no período do pós-parto. Condições socioeconômicas e demográficas e relativas aos hábitos de vida do casal foram consideradas como potenciais preditores de violência. Utilizou-se a versão em português da CTS2 para identificar as situações de VPI. A variável de desfecho foi analisada em três níveis: ausência de VFPI, presença de VFPI no período da gestação ou do pós-parto e presença de VFPI em ambos os períodos. Utilizou-se um modelo logito-multinomial para as projeções de prevalências segundo os descritores selecionados. Os fatores que mais aumentaram a probabilidade de ocorrência de violência durante a gestação e/ou nos primeiros cinco meses de vida da criança foram: idade materna < 20 anos, escolaridade materna inferior ao 2 grau completo, ter 2 ou mais filhos menores de cinco anos, tabagismo materno, uso inadequado de álcool pela mãe e/ou companheiro, uso de drogas pela mãe e/ou companheiro e percepção materna sobre a saúde do bebê aquém da esperada. Entre mães com todas estas características, a estimativa de prevalência projetada de VFPI na gestação e/ou no pós-parto chegou a 96,4%, sendo 59,4% a estimativa de ocorrência em apenas um dos dois períodos e 37% em ambos. Por outro lado, a probabilidade de ocorrência de VFPI cai a 3,6% em famílias sem estas características. Os resultados indicam que a presença de certas características da criança e de sua família aumenta enormemente a probabilidade de ocorrência de VFPI, devendo ser levadas em consideração ao se estabelecer estratégias de intervenção que visem à detecção precoce e uma efetiva intervenção. / Intimate partner violence (IPV) has been recognized as an important public health problem and a risk factor for health problems of women and children. Health services play an important role in early detection of IPV. This issue is especially important in moments of life when systematized care is recommended, such as the pregnancy and early childhood. The main objective of this research was to estimate the probability of physical intimate partners violence (PIPV) during pregnancy or postpartum period. This is a cross-sectional study carried out in 5 primary health care units in Rio de Janeiro during the year 2007. Eight hundred and eleven interviews with mothers of children up to five months of age were carried out. Mothers presenting any formal contraindication for breastfeeding were considered ineligible, as well as women not reporting at least one month of amorous relationship during the pregnancy or postpartum period. Socioeconomic, demographic and life habits of the couple were considered as potential predictors. The outcome variable was assessed at three levels: no PIPV; PIPV during the pregnancy or postpartum; and PIPV in both periods. A multinomial logit model was used for projecting the respective prevalence according to a range of selected descriptors. The risk factors significantly associated with a greater risk of PIPV were: maternal age < 20; women with less than twelve years of schooling; women with 2 or more children under five years old; alcohol misuse by mother or partner; illicit drug abuse by the women or partner; and mothers perception of infants health lower than expected. Between mothers with all those characteristics, the projected prevalence of PIPV during pregnancy and/or postpartum was 96.4%, and as much as 59.4% accounted for the occurrence in one of the periods and 37% for the occurrence in both periods. In the absence of all of those factors, estimates reached only 3.6% during pregnancy and/or postpartum Summarizing, this study disclosed that some characteristics of children and their family increases the probability of PIPV during the pregnancy and postpartum period. This information is valuable for achieving early identification and effective resolution of the problem.

Les crises monétaires sont les premières crises financières de l'histoire économique. Elles se traduisent par l'élimination ou la substitution des monnaies nationales. L'objectif de cette thèse est de poser les fondements théoriques et empiriques des crises monétaires, mais également de proposer un cadre de prévention de ce type de crise qui sévit en zone euro depuis 2008. Les débats économiques et politiques actuels autour des questions de désintégration monétaire témoignent de la persistance et de l'ampleur de la crise, où la légitimité et la souveraineté de la monnaie unique est menacée à moyen long terme. Les divers plans de sauvetage et les politiques d'austérité dans les pays-membres en difficulté ne sont que les conséquences et les coûts directs d'une telle crise. Ces derniers doivent interpeller les autorités de supervision à une plus grande vigilance, ainsi qu'à une politique de prévention plus avisée. L'objectif de cette thèse est double : dans une première partie, nous analysons le concept, les fondements historiques et théoriques des crises monétaires, puis dressons une typologie de celles-ci. Dans une seconde partie, nous apportons une contribution empirique relative aux déterminants des crises monétaires en zone euro et proposons un outil de prévention des crises monétaires, grâce à la mise en place d'un système d'alerte avancée (Early Warning System), par l'approche économétrique de type logit multinomial. Pour ce faire, la détection et la mesure des mésalignements des taux de change réels à l'intérieur de la zone euro est cruciale, puisqu'il constitue l'indicateur premier des crises monétaires. L'estimation des taux de change d'équilibre permettent ainsi d'apprécier la sur ou sous-évaluation des monnaies, indispensable à la mise en place d'un système d'alerte avancée, à des fins de prévention des crises monétaires. / Monetary crises are the first financial crises in economic history, which result in the elimination or substitution of national currencies. The aim of this thesis is to study the theoretical and empirical foundations of monetary crises. Furthermore, a framework for the prevention of such crises, raging in the Euro zone since 2008, is provided. The current economic and political debates about this issue reflect the persistence and the extent of this crisis, in which the Euro's legitimacy and sovereignty is threatened in the medium term. The various rescue plans and austerity policies in troubled member states are direct consequences and costs of this crisis. Consequently, supervisory authorities need to be more vigilant in strengthening their prevention policy. The purpose of this thesis is twofold: in the first part, we analyze the concept, the historical and theoretical foundations of monetary crises, and then develop a typology of them. In the second part, we provide an empirical contribution on the determinants of monetary crises in the euro area and propose a tool for preventing currency crises by setting up an Early Warning System, through the econometric approach of the multinomial logit model. As the primary indicator of monetary crises, the detection and measurement of real exchange rate misalignments within the euro area is decisive. The equilibrium exchange rates estimation allows the assessment of currency over- or undervaluation, which is essential for the implementation of an early warning system

[ES] Los problemas de selección de talla y ajuste en la compra online de ropa son la causa de aproximadamente el 70% de las devoluciones. Esto se debe principalmente a la falta de estandarización del tallaje y al funcionamiento deficiente de los métodos de recomendación de talla. Actualmente, el comercio online de ropa tiene mucho potencial, sin embargo, las elevadas tasas de devolución, suponen costes relevantes en logística y gestión de stocks. En el estado del arte de esta tesis se refleja la complejidad del problema del ajuste de ropa, en el que intervienen múltiples factores tanto objetivos (tipo de tejido, patronaje, número de tallas, moda, morfotipo del usuario, medidas corporales, etc.), como subjetivos (percepción de ajuste y preferencias del usuario). Siendo el ajuste de las prendas uno de los problemas de mayor relevancia en el sector de la confección es de crucial interés avanzar en la generación de un conocimiento que permita relacionar con mayor precisión las dimensiones corporales, el ajuste de las prendas y el tallaje. El objetivo general de esta tesis consiste en establecer los fundamentos para desarrollar un sistema de recomendación del ajuste y talla de prendas de ropa a partir de medidas antropométricas del usuario y valoraciones de ajuste planteando un desarrollo metodológico que sirve de punto de partida para posteriormente escalar el proceso a cualquier tipo de prenda, estilo y sistema de tallaje de ropa. La aproximación propuesta consiste en la predicción del ajuste por zonas de la prenda a partir de medidas antropométricas del usuario y pruebas de ajuste previas utilizando el método estadístico de regresión logística multinomial. A partir de esta predicción de ajuste por zonas, y aplicando de nuevo modelos de regresión logística multinomial, se obtiene la probabilidad de ajuste de la serie de tallas de la prenda analizada. En primer lugar, se ha determinado la fiabilidad de las medidas antropométricas obtenidas a partir de escaneados 3D del cuerpo. Para desarrollar los modelos de predicción, se ha puesto a punto un método de caracterización del ajuste de ropa mediante valoración subjetiva de usurarios y expertos. Además, se han definido los conjuntos de medidas antropométricas relacionadas con el ajuste de la prenda en cada zona. El proceso de entrenamiento de los modelos de predicción de ajuste ha permitido determinar cuáles son las medidas antropométricas más relevantes para el ajuste de cada tipo de prenda, así como las zonas de ajuste que influyen en la selección de la talla. En la fase de validación, se ha demostrado que, con un porcentaje de acierto entre el 80-100%, los modelos de predicción de talla basados en probabilidades de ajuste obtenidas mediante regresión logística multinomial en zonas relevantes de la prenda, ofrecen mayor fiabilidad que los métodos actuales que solo consideran una variable corporal y sus intervalos. Finalmente, se ha propuesto un método para extrapolar los modelos individuales de predicción de talla a toda población objetivo, estimar la cuota de mercado potencial y optimizar la distribución de tallas de cada prenda. / [CA] Els problemes de selecció de talla i ajust en les compres de roba en la xarxa representen aproximadament el 70% de les devolucions. Això es degut principalment a la manca d'estandardització en les talles i al funcionament deficient dels mètodes de recomanació de talles. Actualment, el comerç de roba en la xarxa té molt potencial, no obstant això, les altes taxes de devolució comporten costos rellevants en logística i gestió d'estocs. L'estat de l'art en aquesta tesi reflecteix la complexitat del problema de l'ajust de la roba, que implica múltiples factors, tant objectius (tipus de teixit, patró, nombre de talles, tendències de moda, tipus de cos de l'usuari, mesures corporals, etc.) com subjectius (percepció de l'ajust per part de l'usuari i preferències). Ja que l'ajust de les peces de vestir és un dels problemes més importants en la indústria de la moda, és crucial avançar en la generació de coneixement que permeti establir una relació més precisa entre les dimensions del cos, l'ajust de la roba i les talles. L'objectiu general d'aquesta tesi és establir els fonaments per al desenvolupament d'un sistema de recomanació de l'ajust i la talla de peces de roba basat en les mesures antropomètriques de l'usuari. Això implica un desenvolupament metodològic que serveix com a punt de partida per a posteriorment escalar el procés a qualsevol tipus de peça de roba, estil i sistema de mides. L'aproximació proposada consistix en predir l'ajust per zones de la peça de roba basat en les mesures antropomètriques de l'usuari i proves prèvies d'ajust mitjançant el mètode estadístic de la regressió logística multinomial. A partir d'aquesta predicció d'ajust per zones, i aplicant novament models de regressió logística multinomial, s'obté la probabilitat d'ajust per a la gamma de talles de la peça de roba analitzada. S'ha determinat la fiabilitat de les mesures antropomètriques obtingudes a partir d'escaneigs 3D del cos. Per desenvolupar els models de predicció, s'ha posat a punt un mètode per caracteritzar l'ajust de la roba mitjançant avaluacions subjectives dels usuaris i experts. A més, s'han definit conjunts de mesures antropomètriques relacionades amb l'ajust de la peça a cada zona. El procés de formació dels models de predicció de l'ajust ha permès determinar les mesures antropomètriques més rellevants per a l'ajust de cada tipus de peça, així com les zones d'ajust que influeixen en la selecció de la talla. En la fase de validació, s'ha demostrat que, amb un percentatge d'encert entre el 80-100%, els models de predicció de talla basats en les probabilitats d'ajust obtingudes mitjançant la regressió logística multinomial en zones rellevants de la peça de roba ofereixen una major fiabilitat que els mètodes actuals que només consideren una variable corporal i els seus intervals. Finalment, s'ha proposat un mètode per extrapol·lar els models individuals de predicció de talla a tota la població objectiu, estimar la quota de mercat potencial i optimitzar la distribució de talles per a cada peça. / [EN] The problems of size selection and fit in online clothing purchases account for approximately 70% of returns. This is primarily due to the lack of standardization in sizing and the inefficient performance of current size recommendation methods. Currently, online clothing retail has a lot of potential; however, the high product return rates result in significant costs in logistics and stock management. The state of the art in this thesis reflects the complexity of the clothing fit problem, which involves multiple factors, both objective (fabric type, pattern, number of sizes, fashion trends, user body type, body measurements, etc.) and subjective (user's perception of fit and preferences). Since garment fit is one of the most significant issues in the fashion industry, it is crucial to advance in generating knowledge that allows for a more precise relationship between body dimensions, garment fit, and sizing. The general objective of this thesis is to establish the foundations for developing a recommendation system for clothing fit and size based on user anthropometric measurements and fit evaluations. This involves a methodological development that serves as a starting point for subsequently scaling the process to any type of garment, style, and sizing system. The proposed approach consists of predicting the fit by garment zones based on user anthropometric measurements and previous fit trials using the statistical method of multinomial logistic regression. From this prediction of fit by zones, and by once again applying multinomial logistic regression models, the probability of fit for the range of sizes of the analyzed garment is obtained. The reliability of anthropometric measurements obtained from 3D body scans has been determined. To develop the prediction models, a method for characterizing garment fit through subjective assessments by users and experts has been refined. In addition, sets of anthropometric measurements related to garment fit in each zone have been defined. The training process of the fit prediction models has enabled determining the most relevant anthropometric measurements for the fit of each type of garment, as well as the fit zones that influence size selection. In the validation phase, it has been demonstrated that, with an accuracy rate between 80-100%, size prediction models based on fit probabilities obtained through multinomial logistic regression in relevant garment zones offer greater reliability than current methods that only consider a single body variable and its intervals. Finally, a method has been proposed to extrapolate individual size prediction models to the entire target population, estimate the potential market share, and optimize the distribution of sizes for each garment. / Alemany Mut, MS. (2023). Desarrollo de modelos estadísticos de predicción del ajuste y talla de prendas de ropa a partir de la percepción y características antropométricas del usuario [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/202617

Das Verkehrsmittelwahlverhalten von Menschen ist komplex. So spielen soziodemografische, sozioökonomische sowie raum- und siedlungsstrukturelle Merkmale eine Rolle. In dieser Arbeit wird dieses Verhalten untersucht. Dabei wird eine homogene Grundgesamtheit geschaffen, welche alle Personen beinhaltet, die sowohl über eine Dauerkarte des öffentlichen Personenverkehrs als auch einen Personenkraftwagen verfügen. Anhand derer soll eine deskriptive Analyse und eine multinomiale logistische Regression Aufschluss geben, ob es Unterschiede zwischen den jeweiligen Nutzergruppen gibt. So lässt sich die Gruppe der ÖV-Nutzer durch folgende Charakteristiken beschreiben: der Großteil sind Frauen, sowie Personen, die eine hohe schulische und berufliche Bildung besitzen. Des Weiteren werden eher weniger Wege mit dem ÖV als mit dem PKW zurückgelegt. Erwerbstätige hingegen entscheiden sich eher für den PKW. / Human behavior towards the choice of transportation varies in very complex ways such as sociodemographics, socioeconomics as well as settlement structures. For this paper a homogenous population is created from season ticket holders for public transportation and car owners. Based on this population a descriptive analysis followed by a multinomial logistic regression is supposed to generate the differences between the user groups. The group of users of the public transportation system can be characterized as followed: the majority of users are women as well as highly educated people. Within this specific group distances are more likely to be covered by public transportation rather than by car. However the working population prefers to go by passenger car.

Title: Option Pricing Author: Radek Moravec Department: Department of Probability and Mathematical Statistics Supervisor: doc. RNDr. Jan Hurt, CSc., Department of Probability and Mathematical Statistics In the present thesis we deal with European call option pricing using lattice approaches. We introduce a discrete market model and show a way how to find an arbitrage price of financial instruments on complete markets. It's equal to the discounted value of future expected cash flow. We present the binomial option pricing model and generalize it into multinomial model. We test the resulting formula on real market data obtained from NYSE and NASDAQ. We suggest a parameter estimate method which is based on time series of historical observations of daily close price. We compare calculated option prices with their real market value and try to explain the reasons of the differences. 1

Based on the curriculum of 2011 a new grading scale ranging from A-F was introduced in the Swedish upper secondary school system. Previous research on similar data have focused on the earlier grading scale, and its crucial that the new circumstances are addressed to understand the impact on grading. Using 348 essays from the national writing test this study investigates the use of automated essay scoring as a way of grading in this new setting. Using various classication methods the models for younger students outperform the corresponding models for older students. This implies that it is harder to predict grades on essays written by older students. Based on the current data the result shows that with the new grading scale the use of automated essay scoring should be used with caution.

Incorporating survey weights into likelihood-based analysis is a controversial issue because the sampling weights are not simply equal to the reciprocal of selection probabilities but they are adjusted for various characteristics such as age, race, etc. Some adjustments are based on nonresponses as well. This adjustment is accomplished using a combination of probability calculations. When we build a logistic regression model to predict categorical outcomes with survey data, the sampling weights should be considered if the sampling design does not give each individual an equal chance of being selected in the sample. We rescale these weights to sum to an equivalent sample size because the variance is too small with the original weights. These new weights are called the adjusted weights. The old method is to apply quasi-likelihood maximization to make estimation with the adjusted weights. We develop a new method based on the correct likelihood for logistic regression to include the adjusted weights. In the new method, the adjusted weights are further used to adjust for both covariates and intercepts. We explore the differences and similarities between the quasi-likelihood and the correct likelihood methods. We use both binary logistic regression model and multinomial logistic regression model to estimate parameters and apply the methods to body mass index data from the Third National Health and Nutrition Examination Survey. The results show some similarities and differences between the old and new methods in parameter estimates, standard errors and statistical p-values.

The Slovenian Public Opinion survey (SPOS), which carried out in 1990, was used by the government of Slovenia as a benchmark to prepare for an upcoming plebiscite, which asked the respondents whether they support independence from Yugoslavia. However, the sample size was large and it is quite likely that the respondents and nonrespondents had divergent viewpoints. We first develop an ignorable nonresponse model which is an extension of a bivariate binomial model. In order to accommodate the nonrespondents, we then develop a nonignorable nonresponse model which is an extension of the ignorable model. Our methodology uses an EM algorithm to fit both the ignorable and nonignorable nonresponse models, and estimation is carried out using the bootstrap mechanism. We also perform sensitivity analysis to study different degrees of departures of the nonignorable nonresponse model from the ignorable nonresponse model. We found that the nonignorable nonresponse model is mildly sensitive to departures from the ignorable nonresponse model. In fact, our finding based on the nonignorable model is better than an earlier conclusion about another nonignorable nonresponse model fitted to these data.

Automation of document handling in the construction industries could save large amounts of time, effort and money and classifying a document is an important step in that automation. In the field of machine learning, lots of research have been done on perfecting the algorithms and techniques, but there are many areas where those techniques could be used that has not yet been studied. In this study I looked at how effectively the machine learning algorithm multinomial Naïve-Bayes would be able to classify 1427 documents split up into 19 different categories from a construction project. The experiment achieved an accuracy of 92.7% and the paper discusses some of the ways that accuracy can be improved. However, data extraction proved to be a bottleneck and only 66% of the original documents could be used for testing the classifier.

Every day, traditional statistical methodology are used world wide to study a variety of topics and provides insight regarding countless subjects. Each technique is based on a distinct set of assumptions to ensure valid results. Additionally, many statistical approaches rely on large sample behavior and may collapse or degenerate in the presence of small, spare, or correlated data. This dissertation details several advancements to detect these conditions, avoid their consequences, and analyze data in a different way to yield trustworthy results. One of the most commonly used modeling techniques for outcomes with only two possible categorical values (eg. live/die, pass/fail, better/worse, ect.) is logistic regression. While some potential complications with this approach are widely known, many investigators are unaware that their particular data does not meet the foundational assumptions, since they are not easy to verify. We have developed a routine for determining if a researcher should be concerned about potential bias in logistic regression results, so they can take steps to mitigate the bias or use a different procedure altogether to model the data. Correlated data may arise from common situations such as multi-site medical studies, research on family units, or investigations on student achievement within classrooms. In these circumstance the associations between cluster members must be included in any statistical analysis testing the hypothesis of a connection be-tween two variables in order for results to be valid. Previously investigators had to choose between using a method intended for small or sparse data while assuming independence between observations or a method that allowed for correlation between observations, while requiring large samples to be reliable. We present a new method that allows for small, clustered samples to be assessed for a relationship between a two-level predictor (eg. treatment/control) and a categorical outcome (eg. low/medium/high).