A Estatística é uma ferramenta imprescindível para a aplicação do método científico, estando presente em todos os campos de pesquisa. As metodologias estatísticas usuais estão bem estabelecidas entre os pesquisadores das mais diversas áreas, sendo que a análise de dados em muitos trabalhos costuma ser feita pelos próprios autores. Nos últimos anos, a área conhecida como Ciência de Dados vem exigindo de estatísticos e não-estatísticos habilidades que vão muito além de modelagem, começando na obtenção e estruturação das bases de dados e terminando na divulgação dos resultados. Dentro dela, uma abordagem chamada de aprendizado automático reuniu diversas técnicas e estratégias para modelagem preditiva, que, com alguns cuidados, podem ser aplicadas também para inferência. Essas novas visões da Estatística foram pouco absorvidas pela comunidade científica até então, principalmente pela ausência de estatísticos em grande parte dos estudos. Embora pesquisa de base em Probabilidade e Estatística seja importante para o desenvolvimento de novas metodologias, a criação de pontes entre essas disciplinas e suas áreas de aplicação é essencial para o avanço da ciência. O objetivo desta tese é aproximar a ciência de dados, discutindo metodologias novas e usuais, da área de pesquisa em poluição do ar, que, segundo a Organização Mundial da Saúde, é o maior risco ambiental à saúde humana. Para isso, apresentaremos diversas estratégias de análise e as aplicaremos em dados reais de poluição do ar. Os problemas utilizados como exemplo foram o estudo realizado por Salvo et al. (2017), cujo objetivo foi associar a proporção de carros rodando a gasolina com a concentração de ozônio na cidade de São Paulo, e uma extensão desse trabalho, na qual analisamos o efeito do uso de gasolina/etanol na mortalidade de idosos e crianças. Concluímos que suposições como linearidade a aditividade, feitas por alguns modelos usuais, podem ser muito restritivas para problemas essencialmente complexos, com diferentes modelos levando a diferentes conclusões, nem sempre sendo fácil identificar qual delas é a mais apropriada. / Statistics is a fundamental part of the scientific method and it is present in all the research fields. The usual statistical techniques are well established in the scientific community, and, regardless of the area, the authors themselves perform the data analysis in most papers. In the last years, the area known as Data Science has been challenging statisticians and non-statisticians to perform tasks beyond data modeling. It starts with importing, organizing and manipulating the databases, and ends with the proper communication of the results. Another area called Machine Learning created a framework to fit predictive models, where the goal is to obtain the most precise predictions to a variable under study. These new approaches were not completely adopted by the scientific community yet, mainly due to the absence of statisticians in most of the studies. Although basic research in Probabilities and Statistics is important, the link between these disciplines and their application areas is essential for the advancement of science. The goal of this thesis was to bring together the news views of Data Science and Machine Learning and air pollution research. We presented several strategies of data analysis and apply them to reanalyze the real world air pollution problem presented by Salvo et al. (2017) explore the association between ozone concentration and the proportion of bi-fuel vehicles running on gasoline in the city of São Paulo, Brazil. We also extended this analysis to study the effect of using gasoline/ethanol in mortality (child and elderly). We concluded that assumptions such as linearity and additivity, commonly required by usual models, can be very restrictive to intrinsically complex problems, leading to different conclusions for each fitted model, with little information about which one is more appropriate.

abstract: A major challenge in health-related policy and program evaluation research is attributing underlying causal relationships where complicated processes may exist in natural or quasi-experimental settings. Spatial interaction and heterogeneity between units at individual or group levels can violate both components of the Stable-Unit-Treatment-Value-Assumption (SUTVA) that are core to the counterfactual framework, making treatment effects difficult to assess. New approaches are needed in health studies to develop spatially dynamic causal modeling methods to both derive insights from data that are sensitive to spatial differences and dependencies, and also be able to rely on a more robust, dynamic technical infrastructure needed for decision-making. To address this gap with a focus on causal applications theoretically, methodologically and technologically, I (1) develop a theoretical spatial framework (within single-level panel econometric methodology) that extends existing theories and methods of causal inference, which tend to ignore spatial dynamics; (2) demonstrate how this spatial framework can be applied in empirical research; and (3) implement a new spatial infrastructure framework that integrates and manages the required data for health systems evaluation. The new spatially explicit counterfactual framework considers how spatial effects impact treatment choice, treatment variation, and treatment effects. To illustrate this new methodological framework, I first replicate a classic quasi-experimental study that evaluates the effect of drinking age policy on mortality in the United States from 1970 to 1984, and further extend it with a spatial perspective. In another example, I evaluate food access dynamics in Chicago from 2007 to 2014 by implementing advanced spatial analytics that better account for the complex patterns of food access, and quasi-experimental research design to distill the impact of the Great Recession on the foodscape. Inference interpretation is sensitive to both research design framing and underlying processes that drive geographically distributed relationships. Finally, I advance a new Spatial Data Science Infrastructure to integrate and manage data in dynamic, open environments for public health systems research and decision- making. I demonstrate an infrastructure prototype in a final case study, developed in collaboration with health department officials and community organizations. / Dissertation/Thesis / Doctoral Dissertation Geography 2017

Wastewater Treatment Plants (WWTPs) are sophisticated systems that have to sustain long-term qualified performance, regardless of temporally volatile volumes or compositions of the incoming wastewater. Membrane filtration in the Membrane Bioreactors (MBRs) reduces the WWTPs footprint and produces effluents of proper quality. The energy or electric power consumption of the WWTPs, mainly from aeration equipment and pumping, is directly linked to greenhouse gas emission and economic input. Biological treatment requires oxygen from aeration to perform aerobic decomposition of aquatic pollutants, while pumping consumes energy to overcome friction in the channels, piping systems, and membrane filtration. In this thesis, we researched full-scale WWTPs Influent Conditions (ICs) monitoring and forecasting models to facilitate the energy consumption budgeting and raise early alarms when facing latent abnormal events. Accurate and efficient forecasts of ICs could avoid unexpected system disruption, maintain steady product quality, support efficient downstream processes, improve reliability and save energy. We carried out a numerical study of bioreactor microbial ecology for MBRs microbial communities to identify indicator species and typical working conditions that would assist in reactor status confirmation and support energy consumption budgeting. To quantify membrane fouling and cleaning effects at various scales, we proposed quantitative methods based on Matern covariances to analyze biofouling layer thickness and roughness obtained from Optical Coherence Tomography (OCT) images taken from gravitydriven MBRs under various working conditions. Such methods would support practitioners to design suitable data-driven process operation or replacement cycles and lead to quantified WWTPs monitoring and energy saving. For future research, we would investigate data from other full-scale water or wastewater treatment process with higher sampling frequency and apply kernel machine learning techniques for process global monitoring. The forecasting models would be incorporated into optimization scenarios to support data-driven decision-making. Samples from more MBRs would be considered to gather information of microbial community structures and corresponding oxygen-energy consumption in various working conditions. We would investigate the relationship between pressure drop and spatial roughness measures. Anisotropic Matern covariance related metrics would be adopted to quantify the directional effects under various operation and cleaning working conditions.

<p>This dissertation details one of the first lines of inquiry into the rhetorical strategies used in scientific data analysis. The study primarily concerns the relationships between data work and knowledge making in the analysis of so-called “big data,” and how rhetoric and technical communication theories might inform those relationships. Hinging on five months embedded at a national science laboratory, this study uses ethnographic methods to detail the ways in which data analysis is neither purely data-driven and objective, nor purely situated in a local context or problem. Rather, data work requires both analytical processes and artful <i>techne</i> embedded in ongoing reflective praxis. As purely analytic, data work focuses on mathematical treatments, step by step procedures and rote formulas. As <i>techne</i>, data work requires interpretation. Rhetorical data analysis is not the opposite of data-driven work. Instead, rhetorical <i>techne</i> stands as the midpoint between the extremes of purely data-driven and purely context-driven analysis. Based on three cases that compare the practices of data novices, seasoned experts, and interdisciplinary teams, I argue that the ways in which scientists go about their data cleaning, collaboration, and analysis change based on their levels of expertise and the problem at hand. A number of principles that outline how data analysis is a form of rhetorical inscription are also defined, including the ways data dictionaries, model building and the construction of proxies intimately link scientific insights with language. The set of principles detailed in this dissertation are key areas that should be considered in both data science education and professional and technical writing curricula. Therefore, the project should be of particular interest to instructors and administrators in both Technical Writing and Data Science programs, as well as well as critical data studies scholars.</p>

Newsvendor problems in Operations Research predict the optimal inventory levels necessary to meet uncertain demands. This thesis examines an extended version of a single period multi-product newsvendor problem known as the ice cream vendor problem. In the ice cream vendor problem, there are two products – ice cream and hot chocolate – which may be substituted for one another if the outside temperature is no too hot or not too cold. In particular, the ice cream vendor problem is a data-driven extension of the conventional newsvendor problem which does not require the assumption of a specific demand distribution, thus allowing the demand for ice cream and hot chocolate respectively to be temperature dependent. Using Discrete Event Simulation, we first simulate a real-world scenario of an ice cream vendor problem via a demand whose expected value is a function of temperature. A sample average approximation technique is subsequently used to transform the stochastic newsvendor program into a feature-driven linear program based on the exogenous factors of probability of rainfall and temperature. The resulting problem is a multi-product newsvendor linear program with L1-regularization. The solution to this problem yields the expected cost to the ice cream vendor as well as the optimal order quantities for ice cream and hot chocolate, respectively.

This work has its focus on decision support in operational business situations and especially on the very short-term decisions in Online Surgery Scheduling, which has the goal of efficient and structured operations in the Operating Room area at minimal costs. This use case includes all intra-day decisions needed to ensure the execution of all planned and unplanned surgeries of the surgery schedule, with all of the concomitant uncertainties like unexpected events, delays, cancellations and emergency patients. This so far barely considered problem needs research for decision support, since few approaches are available that relieve the OR manager through tool support and reduce the informational, communicational and cognitive workloads needed to ensure efficient and seamless operations. With the strong growth of generated data and the digitization of business processes that make previously unobtrusive business elements become more visible, and their combination with large-scale data processing technologies and intelligent methods of the fields of AI or Analytics, new opportunities for data-driven real-time Decision Support Systems become evident. The objective of this research is the development of an approach that supports the operational decision processes in Operating Room Management and Online Surgery Scheduling, like facilitating the information collection and reducing the cognitive effort for decision-making by providing predictive information or alternative actions. In order to achieve this goal, a decision support approach is developed that utilizes streaming data of medical and surgical devices in a Situation Detection Subsystem, a Prediction Subsystem and a Rescheduling Subsystem. These components combine intelligent methods and scalable data processing technologies, consequently contributing a data-driven Decision Support System for Online Surgery Scheduling. The scientific contribution relates to the field of Business and Decision Analytics with its main challenges of increasing complexity and dynamics of today’s business decisions. This work provides a novel DSS approach, innovative models and concepts which consider exactly these problems with regards to the characteristics of OSS.:Table of Contents ................................ I List of Figures .................................. III List of Tables ................................... IV List of Abbrevations ............................. V 1 Introduction.................................... 1 1.1 Motivation ................................... 1 1.2 Research Objective and Questions ............. 2 1.3 Research Methodology ......................... 4 1.4 Outline ...................................... 7 2 Background ..................................... 9 2.1 Operational Decisions and Decision Support Systems .................................. 9 2.1.1 Decisions and Decision-making .............. 9 2.1.2 Operational Decision-making ................ 11 2.1.3 Decision Support Systems ................... 13 2.1.4 Business Value and Benefits ................ 18 2.2 Business Analytics ........................... 19 2.2.1 Characterization and Definition ............ 19 2.2.2 Delimitation of Areas ...................... 20 2.2.3 Types of Business Analytics ................ 21 2.2.4 Methods and Technologies ................... 23 3 Use-Case: Operational Decisions in Operating Room Management .................................. 29 3.1 Preliminary Considerations ................... 29 3.2 Online Surgery Scheduling .................... 31 3.2.1 Mapping of Decision Theory and Online Surgery Scheduling ............................... 32 3.2.2 Information Demands ........................ 33 3.2.3 State of the Art in Decision Support Systems 36 4 Motivation and Requirements .................... 38 4.1 Development of an Information System Architecture for Online Surgery Scheduling ....... 38 4.2 Summary ...................................... 49 5 Evaluation of Big Data Processing Frameworks.... 51 5.1 Evaluating new Approaches of Big Data Analytics Frameworks ............................. 51 5.2 Summary ...................................... 63 6 Stream Processing for Intra-surgical Phase Detection ........................................ 64 6.1 Method for Intra-surgical Phase Detection by Using Real-time Medical Device Data .............. 64 6.2 Summary ...................................... 71 7 Real-time Predictive Analytics in Operating Room Management ....................................... 72 7.1 A Big Data Architecture for Intra-surgical Remaining Time Predictions ....................... 72 7.2 Summary ...................................... 81 8 Data-driven Online Surgery Rescheduling ........ 83 8.1 Online Surgery Rescheduling - A Data-driven Approach for Real-time Decision Support .......... 83 8.2 Summary ...................................... 92 9 Prototypical Implementation: Decision Support in Online Surgery Scheduling ........................ 93 9.1 Implementation of a Situation Aware and Real-time Approach for Decision Support in Online Surgery Scheduling ............................... 93 9.2 Summary ...................................... 99 10 Conclusion .................................... 100 10.1 Summary and Contributions ................... 100 10.2 Limitations and Future Work ................. 103 Bibliography ..................................... VII Appendix ......................................... XXV Wissenschaftlicher Werdegang ..................... XXXI Selbständigkeitserklärung ........................ XXXII

The Newsvendor problem is a classical supply chain problem used to develop strategies for inventory optimization. The goal of the newsvendor problem is to predict the optimal order quantity of a product to meet an uncertain demand in the future, given that the demand distribution itself is known. The Ice Cream Vendor Problem extends the classical newsvendor problem to an uncertain demand with unknown distribution, albeit a distribution that is known to depend on exogenous features. The goal is thus to estimate the order quantity that minimizes the total cost when demand does not follow any known statistical distribution. The problem is formulated as a mathematical programming problem and solved using a Deep Neural network approach. The feature-dependent demand data used to train and test the deep neural network is produced by a discrete event simulation based on actual daily temperature data, among other features.

aus dem Inhalt: „Seit mehr als zwei Dekaden wird sich in der Marktforschung darum bemüht, neue Wege zu erschließen, um der digitalen Transformation und den damit verbundenen gesellschaftlichen Veränderungsprozessen mit geeigneten Forschungsmethoden zu begegnen. Insbesondere in der Online-Marktforschung wird gefragt, inwiefern digitale Ressourcen erschlossen werden können, um zielgruppenorientierte Datenanalysen für (neue) Customer Insights nutzbar zu machen.”

aus dem Inhalt: „In den vergangenen Jahren haben wir einen immensen Anstieg an verfügbaren Textdaten feststellen dürfen. Nicht nur Verbraucherkommentare in Social Media, Text-Transkripte von Sprachassistenten, Dialoge aus Customer-Feedback und Support Chat (Bots), sondern auch offene Antworten in Umfragen bis hin zu automatisierten Text2Speechgestützten Audiointerviews sorgen für steigenden Bedarf, Erkenntnisse aus unstrukturierten Textdaten zu gewinnen.”

aus dem Inhalt: „Text Mining ist eine interdisziplinäre Forschungsrichtung, jedoch befassen sich nur sehr wenige Arbeiten bislang mit dem Extrahieren, Analysieren und Erkennen von freien Antworten aus Online-Befragungen.”