Problems in statistical analysis, economics, and many other disciplines often involve a trade-off between rewards and additional information that could yield higher future rewards. This thesis investigates such a trade-off, using a class of problems known as bandit problems. In these problems, a reward-seeking agent makes decisions based upon his beliefs about a parameter that controls rewards. While some choices may generate higher short-term rewards, other choices may provide information that allows the agent to learn about the parameter, thereby potentially increasing future rewards. Learning occurs if the agent's subjective beliefs about the parameter converge over time to the parameter's true value. However, depending upon the environment, learning may or may not be optimal, as in the end, the agent cares about maximizing rewards and not necessarily learning the true value of the underlying parameter.

This thesis investigates the injunction to experiment in the social sciences and, more specifically, geography. This is both a geography of certain ways of thinking experiment, and an exploration of how particular strands of geographical thinking are being re-imagined and reworked as experimental under the influence of ideas and practices from within and beyond the discipline. Against the backdrop of recent debates about the status of experiment, it poses a number of key questions about what it means to be experimental, how experimental practices emerge and travel, and how these processes are inflected by the organization and atmospheres of particular sites of experimentation. These questions are addressed through a form of attentive participation at four key sites: The SenseLab and the Topological Media Lab in Montreal, the Institut für Raumexperimente in Berlin, and FoAM in Brussels. Based upon these encounters, and drawing upon the work of a range of exemplary experimentalists, the thesis develops the argument that there are new spaces of experiment which are worthy of such examination as part of a renewal of experimentation within geographical thinking. As such, the thesis outlines the logics of these forms of experiment and proposes the notion of ecologies of experiment. It also speculates on the possibilities for re-imagining what constitutes a geographical experiment, foregrounding the necessity of reactivating experiment as an ongoing ethos that needs careful cultivation and tending.

This thesis is called the crime reconstruction. Adapted from a historical context, development of views on this issue, including legislation crime reconstruction and development of reconstruction of the crime, the notion of the nature and significance of forensic reconstruction, the concept of crime reconstruction, reconstruction of the essence of the offense, the importance of reconstruction of the crime; types of forensic reconstruction, as is the reconstruction of the scene, reconstruction of physical objects, reconstruction of the outer form of a man, thought reconstruction; function rekonstrukce offense; entities rekonstrukce offense; psychological assumptions of the reconstruction; tactics rekonstrukce crime, ie crime rekonstrukce preparation, basic tactical principles and procedures reconstruction of the crime, progress and reconstruction stages of the offense; documentation and evaluate the results and the reconstruction of the crime; relation to the reconstruction of a crime other investigative capacity, similarity with the reconstruction of the crime investigative experiment.

Legal regulation of animal testing The diploma thesis is analyzing current legislation in the field of animal experiments. It is formed by five chapters and each chapter is concerned with different aspects of animal protection, when focusing on experimental animals in particular. The first chapter is divided into three subchapters and it is successively describing development of animal experiments, progress of legislation dealing with animal protection and experimental animal protection on the international and European level, when analyzed is also Czech legislation in the last subchapter. Second chapter is focused on the basic legal sources and its legal foundation. It has four subchapters, when every subchapter explains relevant terms, which are being formulated in the experimental animal protection law. This chapter of the diploma thesis also involves sections dedicated to experiments and alternative methods of the experiments. Following chapter of the thesis (chapter nr. 3) describes legislation of the experimental animal protection and it is constituted by three subchapters, which are regarding ensuring expertness of the experiments, farming of and caring for the experimental animals and also approving, progress and finishing the process of the experiment. Every subchapter is then divided into...

The two most popular classification tree algorithms in machine learning and statistics - C4.5 and CART - are compared in a benchmark experiment together with two other more recent constant-fit tree learners from the statistics literature (QUEST, conditional inference trees). The study assesses both misclassification error and model complexity on bootstrap replications of 18 different benchmark datasets. It is carried out in the R system for statistical computing, made possible by means of the RWeka package which interfaces R to the open-source machine learning toolbox Weka. Both algorithms are found to be competitive in terms of misclassification error - with the performance difference clearly varying across data sets. However, C4.5 tends to grow larger and thus more complex trees. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Doctor of Philosophy / Department of Agricultural Economics / Hikaru Hanawa Peterson / Consumers’ food demand has been found to be affected not only by prices and income, but also by their increasing concern about factors like health benefits, animal welfare, and environmental impacts. Thus, many food producers have differentiated and advertised their products using relevant attributes. The increasing demand and supply of differentiated food products have raised questions regarding consumer preferences and producer strategies. This dissertation consists of three essays and empirically examines the egg market to shed light on related issues. The first question that this study aims to answer is whether consumers are willing to pay a premium for livestock and dairy products associated with improved animal welfare. Consumers’ attitude towards such products not only affect manufacturers’ production decisions, but also influence policy makers and current legislations. Using a national online survey with choice experiments, the first essay found that consumers in the study sample valued eggs produced under animal-friendly environment, suggesting incentives for producers to adopt animal welfare friendly practices. In an actual shopping trip, consumers usually need to choose from products with multiple attributes and labels. Studying how consumers with heterogeneous preferences process these information simultaneously and make decisions is important for producers to target interested consumer segments and implement more effective labeling strategies. In the second essay, a different national online survey was administered. The analysis using a latent class model categorized the sample respondents into four classes, and their preferences toward attributes and various label combinations differed across classes. Scanner data, which record actually purchased choices, are an important source of information to study consumer preferences. Diverging from the traditional demand approaches that are limited in studying differentiated product markets using scanner data, this study used a random coefficient logit model to overcome potential limitations and examine the demand relationship as well as price competition in the differentiated egg market. The third essay found that conventional and private labeled eggs yielded higher margins due to less elastic demand and cautioned producers of specialty eggs, which are usually sold at high prices despite their much more elastic demand.

<div> We consider the existence of a force that could produce a non-null result in the Eötvös experiment while producing a null result in the Eöt-Wash experiment. We introduce a general force, in the form of its Taylor series expansion, and determine the response of each experiment to that force. We can then determine which terms of the expansion are important to each experiment. A trial force, in the form of a mixed vector-scalar interaction is introduced and we analyze the resulting Eötvös parameters for various values of the strengths and ranges of the interactions. We find that under certain conditions the Eötvös parameter for the Eöt-Wash experiment can be made zero while the Eötvös parameter for the Eötvös experiment is nonzero.</div><div> </div><div><br></div><div> Next, we examine the possibility of a wind force appearing in the MICROSCOPE experiment. This wind would be due to the satellite's motion through a particle background which couples to the differential accelerometer through a baryon-number dependent interaction. We determine the signal that would be measured by MICROSCOPE satellite and compare the power spectrum density of this signal to the published power spectrum density of the experiment. </div><div> </div><div><br></div><div> Additionally, we present a new theoretical framework for the time-variation of nuclear decay rates. This new framework is motivated by the results of numerous experiments which show a periodicity of one year. The fractional decay rate of these experiments are constant, regardless of isotope. We find that a novel neutrino interaction, in the form of an index of refraction, successfully generates the constant fractional decay rates. Using the optical theorem and the relativistic Breit-Wigner distribution makes the index of refraction consistent with neutrino speed measurements. We conclude by describing other systems where the index of refraction could create observable oscillations.</div><div><br></div><div> </div><div> Finally, we consider the suppression of beta decay rates through the Pauli exclusion principle due to the presence of background cosmic and solar neutrinos. We derive the suppression factor for both thermalized and non-thermalized neutrinos.</div>

Background: Simulation of physics is something several modern video games use. These simulations are often used to create more believable and realistic environments. Physics-based animations in the form of muscle simulations is one such technique. Objectives: The aim of this thesis is to investigate three different stages of full-body muscle deformation and observe which of these are preferred. One using no degree of deformation being the control condition and the other two using different degrees of deformation being the treatments. This study is conducted by creating animationswith three different degrees of muscle simulation. These animations are then rendered in Maya as well as put into a small fighting scenario implemented in Unreal Engine 4. A user experiment will be conducted where a number of participants will be asked to choose between different scenarios using two-alternative forced choice. After the user study is completed, the data will be analyzed and used to form a conclusion. Methods: Implementations needed to create the stimulus was first done in Maya where the meshes, muscles and animations were created. Renders were done in Maya of all animations and a scene was also implemented in Unreal Engine 4 simulating a small fighting game using the assets created in Maya. To evaluate player preference,a user experiment was conducted with 13 participants where each participant was asked to watch 27 scenarios containing two side-by-side comparisons with different degrees of muscle deformation. The user experiment stimulus was created using PshycoPy which also collected the data of user preference. The scenarios where presented in an arbitrary order. The study was held in a room where the participant was undisturbed. Results: The results showed that no muscle deformation was preferred in all cases where a statistical difference could be found. Conclusions: While the results show that the control condition is mostly preferred, most cases did not yield a conclusive result. Thus further research in the area is necessary.

The study investigates whether it is possible to drive sales-oriented marketing for B2C companies in social media through data-driven marketing, and thus increasing the relevance of the ads. The study has had an experimental approach where hypotheses are tested from a company towards a target group of just over 480 000 people in the Småland region. Two separate marketing campaigns were run on Facebook and Instagram where the fundamental difference between the two was the strategic arrangement. The experimental design included a campaign, containing three separate ads, where strategy and target group segmentation were based on the "The purchasing funnel" model. The result of the campaign, as part of the study's experiments, was then compared against the campaign run without strategy or audience segmentation. This showed a final result just over 20 times better than the previous campaign and a ROAS (return of ad spent) of 25.

Thesis advisor: Michael Russell / For more than a half century surveys and questionnaires with Likert-scaled items have been used extensively by researchers in schools to draw inferences about students; however, to date there has not been a single study that has examined whether alternative item response types on a survey might lead to different results than those obtained with Likert scales in a K-12 setting. This lack of direct comparisons leaves the best method of framing response options in educational survey research unclear. In this study, 4th through 12th grade public school students were administered two versions of the same survey online: one with Likert-scaled response options and the other with visual analogue-scaled response options. A randomized, fixed-effect, between-subjects experimental design was implemented to investigate whether the survey with visual analogue-scaled items yielded results comparable to the survey with Likert-scaled items based on the following four methods and indices: 1) factor structure; 2) internal consistency and test-retest reliability; 3) survey summated scores; and 4) main, interaction, and simple effects. Results of the first three indices suggested that both the Likert scale and visual analogue scale produced similar factor structures, were equally reliable, and yielded summated scores that were not significantly different across all three school levels (elementary, middle, and high school). Results of the factorial ANOVA suggested that only the main effect of school level was statistically significant but that there was no significant interaction between item response type and school level. Results of the post-survey questionnaires suggested that students at all school levels preferred answering questions on the survey with the VAS compared to the LS nearly three to one. / Thesis (PhD) — Boston College, 2008. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement, and Evaluation.