19 research outputs found
Fraud Analytics: A Decade of Research -- Organizing Challenges and Solutions in the Field
No full textThe literature on fraud analytics and fraud detection has seen a substantial increase in output in the past decade. This has led to a wide range of research topics and overall little organization of the many aspects of fraud analytical research. The focus of academics ranges from identifying fraudulent credit card payments to spotting illegitimate insurance claims. In addition, there is a wide range of methods and research objectives. This paper aims to provide an overview of fraud analytics in research and aims to more narrowly organize the discipline and its many subfields. We analyze a sample of almost 300 records on fraud analytics published between 2011 and 2020. In a systematic way, we identify the most prominent domains of application, challenges faced, performance metrics, and methods used. In addition, we build a framework for fraud analytical methods and propose a keywording strategy for future research. One of the key challenges in fraud analytics is access to public datasets. To further aid the community, we provide eight requirements for suitable data sets in research motivated by our research. We structure our sample of the literature in an online database. The database is available online for fellow researchers to investigate and potentially build upon
Fraud analytics : a decade of research: organizing challenges and solutions in the field
No full textAbstract: The literature on fraud analytics and fraud detection has seen a substantial increase in output in the past decade. This has led to a wide range of research topics and overall little organization of the many aspects of fraud analytical research. The focus of academics ranges from identifying fraudulent credit card payments to spotting illegitimate insurance claims. In addition, there is a wide range of methods and research objectives. This paper aims to provide an overview of fraud analytics in research and aims to organize the discipline and its many subfields. We analyze a sample of almost 300 records on fraud analytics published between 2011 and 2020. In a systematic way, we identify the most prominent domains of application, challenges faced, performance metrics, and methods used. In addition, we build a framework for fraud analytical methods and propose a keywording strategy for future research. One of the key challenges in fraud analytics is access to public datasets. To further aid the community, we provide eight requirements for suitable data sets in research motivated by our research. We structure our sample of the literature in an online database. The database is available online for fellow researchers to investigate and potentially build upon
Can causal machine learning reveal individual bid responses of bank customers? A study on mortgage loan applications in Belgium
No full textAbstract: Personal loan pricing requires accurate estimates of individual customer behavior, such as the willingness to take out a loan at a given price, the "bid response". This is challenging due to the nonlinearity of responses hindering the discretionary definition of models, as well as the confoundedness of observational training data. This paper investigates the application of data-driven and machine learning (ML) methods to estimate individual bid responses. We argue that framing bid response modeling as a problem of causal inference is crucial for accurate modeling and understanding of challenging factors. We test established ML algorithms and state-of-the-art causal ML methods on a dataset on mortgage loan applications in Belgium and investigate the effects of different levels of confounding in the data. Our results demonstrate that methods that address confounding can improve bid response estimation, especially when established non-causal methods are negatively affected
Sources of Gain: Decomposing Performance in Conditional Average Dose Response Estimation
No full textEstimating conditional average dose responses (CADR) is an important but
challenging problem. Estimators must correctly model the potentially complex
relationships between covariates, interventions, doses, and outcomes. In recent
years, the machine learning community has shown great interest in developing
tailored CADR estimators that target specific challenges. Their performance is
typically evaluated against other methods on (semi-) synthetic benchmark
datasets. Our paper analyses this practice and shows that using popular
benchmark datasets without further analysis is insufficient to judge model
performance. Established benchmarks entail multiple challenges, whose impacts
must be disentangled. Therefore, we propose a novel decomposition scheme that
allows the evaluation of the impact of five distinct components contributing to
CADR estimator performance. We apply this scheme to eight popular CADR
estimators on four widely-used benchmark datasets, running nearly 1,500
individual experiments. Our results reveal that most established benchmarks are
challenging for reasons different from their creators' claims. Notably,
confounding, the key challenge tackled by most estimators, is not an issue in
any of the considered datasets. We discuss the major implications of our
findings and present directions for future research.Comment: 25 pages, 9 figure
Predicting Employee Turnover: Scoping and Benchmarking the State-of-the-Art
No full textEmployee turnover presents a significant challenge to organizations. High turnover rates impose substantial costs on organizations, e.g., direct costs resulting from rehiring efforts and training new employees, and indirect costs resulting from the loss of expertise and declining organizational productivity. Hence, predicting employee turnover is an important task for human resource departments and organizations as a whole, as it can help to proactively approach employees at risk of churning to improve retention and workforce stability. With ever more data at hand and increasing competition in the labor market, analytical tools are essential to improve workforce management and aid human resource managers in their decision-making. Yet, the existing literature on predictive analytics for employee turnover is scattered and fails to present a coherent and holistic view. To find common ground in the established literature, the paper provides a scoping and benchmarking of the state-of-the-art. The scoping concludes that established research results are difficult to compare due to inconsistent methodologies and experimental setups. To address these issues, an extensive benchmarking experiment is conducted involving 14 classification methods and 9 datasets. The results provide a unique focal point for research on employee turnover prediction and aim to benefit academic research and industry practitioners. The code and public datasets are available on Github to facilitate further extension of the research
Using representation balancing to learn conditional-average dose responses from clustered data
No full textEstimating a unit\u27s responses to interventions with an associated dose, the conditional average dose response (CADR), is relevant in a variety of domains, from healthcare to business, economics, and beyond. Such a response typically needs to be estimated from observational data, which introduces several challenges. That is why the machine learning (ML) community has proposed several tailored CADR estimators. Yet, the proposal of most of these methods requires strong assumptions on the distribution of data and the assignment of interventions, which go beyond the standard assumptions in causal inference. Whereas previous works have so far focused on smooth shifts in covariate distributions across doses, in this work, we will study estimating CADR from clustered data and where different doses are assigned to different segments of a population. On a novel benchmarking dataset, we show the impacts of clustered data on model performance and propose an estimator, CBRNet, that learns cluster-agnostic and hence dose-agnostic covariate representations through representation balancing for unbiased CADR inference. We run extensive experiments to illustrate the workings of our method and compare it with the state of the art in ML for CADR estimation.21 pages, 7 figures, v2: updated methodology and experiment
The Desegregation of a Historically Black High School in Jacksonville, Florida
No full textThis historical study examines the desegregation of a historically African- American high school during the period between 1965-1975. The Mims v. The Duval County School Board (1971) decision brought about radical changes in the operation of the Duval County Public Schools. The mass transfer of teachers and reassignment of students as a result of the federal judge\u27s order in this case resulted in a school system that was dramatically different from the one that previously existed. The author seeks to determine why the desegregation of William Raines High School was short-lived and questions the continued effort of the school system to desegregate this school.
The author conducted a multi-faceted investigation to answer the research questions. Following a case study approach, both archival and oral data were collected and examined. Focused interviews were conducted with former William Raines High School students, faculty and parents. In addition, written documents and local newspaper accounts were studied. The oral interviews support and expand the findings of the archival documents.
The findings of the study indicate that the history and traditions at William Raines High School are founded on a strong sense of pride and identity. However, changes in the school over time have resulted in a school that has lost its focus on academic excellence. In order for lasting desegregation to take place, substantive changes will be required. The pride that was the school\u27s legacy must be restored. Excellence in all aspects of school life should be the overarching goal. PALMM
The Ne Bis in Idem Principle in EU Law
No full textThe legal principle of ne bis in idem restricts the possibility of a defendant being prosecuted repeatedly on the basis of the same offence, act, or facts. Although few would dispute its relevance to the regulation of transnational justice, there is as yet no universally accepted ne bis in idem rule or provision available at the international level, although it is to some extent recognized and respected in Europe, via Article 54 of the Convention on the Implementation of the Schengen Agreement (CISA; integrated into EU law by the Treaty of Amsterdam) and Article 4 of the 7th Protocol of the European Convention on Human Rights. The relevant case law of the ECJ and the ECtHR has implications for the systems of criminal and administrative law in European states, as well as for the interpretation and application of the principle in some areas of EU law, such as competition law. This book analyses these important implications, fulfilling a genuine need to assess the need for – and the prospects of – a ‘European’ ne bis in idem principle. The author identifies and describes obstacles that stand in the way of a single, autonomous, and uniformly applicable general ne bis in idem principle of EU law – differently worded provisions within the different ‘European’ frameworks; a measure of confusion and conflict within the case law of the Community courts; positive conflicts of jurisdiction and the allocation of cases between the Member States; and the vague exception possibilities laid down in Article 55 CISA. Among the issues examined are the following: • the problem of defining the substance of the guarantee; • scope of application of the guarantee; • mechanisms for the coordination of the allocation of cases between Member State authorities and the role of Eurojust; • developments in enhanced cooperation in criminal matters; • fining in competition law; • extraterritoriality and convergence issues; and The author approaches the subject along two lines: first by way of conceptual analysis, and then through a jurisprudential analysis of the relevant case law of the ECJ and the ECtHR, offering well-informed, insightful recommendations and suggestions along the way. As the first contribution to an in-depth understanding of the fundamental problems and issues associated with the interpretation and application of a ‘European’ ne bis in idem principle, this study takes a giant step towards developing and refining the principle within the legal order of the EU. Jurists and legal scholars concerned with the foundations of European law will welcome the book, as will competition law practitioners and officials concerned with transnational police and judicial enforcement
Avaliação em médio prazo da pressão intra-saco após correção endovascular de aneurisma de aorta abdominal com o uso de sensor sem fio
No full textDissertação (mestrado) - Universidade Federal de Santa Catarina, Centro de Ciências da Saúde. Programa de Pós-Graduação em Ciências Médicas.Justificativa: O objetivo do tratamento endovascular do aneurisma de aorta abdominal é a exclusão do saco aneurismático da circulação sistêmica. Para acompanhar a eficácia do tratamento é necessária a realização de angiotomografias seriadas. A monitorização continuada da pressão poderia diagnosticar precocemente casos em que esteja ocorrendo falha desse tratamento, auxiliando ou substituindo a tomografia. Objetivo: Avaliar a eficácia em médio prazo da medida de pressão através de sensor sem fio implantado no saco aneurismático após tratamento endovascular de aneurisma de aorta abdominal (AAA). Desenho do Estudo: Quarenta pacientes submetidos a tratamento endovascular de aneurisma de aorta abdominal receberam implante de sensor sem fio para monitorização em médio prazo da pressão intra-saco. Os dados foram analisados no primeiro, sexto, décimo segundo meses e anualmente. A cada análise, uma angiotomografia informava o diâmetro do aneurisma, presença e tipo de vazamento. A pressão arterial sistêmica assim como todos os dados pressóricos obtidos através do sensor foram coletados. Resultados: No seguimento, dos 40 sensores implantados apenas em dois o sinal não pode ser detectado. A pressão de pulso obtida apresentou boa sensibilidade em afastar a presença de vazamento quando inferior a 25mmHg no primeiro ano. Após esse período, o Índice de Endotensão mostra-se como a variável que mais se aproxima dos dados tomográficos. Na presença de vazamantos tipo I ou III, o sensor é capaz de identificar sucesso ou falha após a correção do vazamento. Até o segundo ano de seguimento, resultados apresentados pelo sensor são concordantes com dados tomográficos, a partir daí, perdem acurácia. Conclusão: Na amostra estudada não foi possível obter sensibilidade suficiente para substituir a realização da tomografia no acompanhamento dos AAA tratados por técnica endovascular. Um maior número de pacientes é necessário para estabelecer o verdadeiro papel da monitorização pressórica sem fio no seguimento do tratamento endovascular do aneurisma de aorta abdominal.Racional: The objective of endovascular treatment of an abdo-minal aortic aneurysm is to exclude the aneurysm sac from systemic circulation. In order to assess treatment effectiveness it is necessary to perform serial computed tomography (CT). Continued pressure moni-toring could provide an early diagnosis of cases in which this treatment is failing, helping or replacing CT. Objective: Evaluation of medium-term effectiveness of pressure measurement using a wireless sensor implanted in the aneurysm sac after endovascular treatment of abdominal aortic aneurysm. Method: Forty patients undergoing endovascular treatment for abdominal aortic aneurysm had a wireless sensor implanted for medium-term monitoring of intrasac pressure. The data were analyzed in the first, sixth and twelfth month and annually. At each analysis, a CT scan in-formed the aneurysm diameter, presence and risk of endoleaks. System-ic arterial pressure was collected, as well as all pressure data obtained through the sensor. Results: Of the 40 sensors implanted, the signal could not be de-tected only in two. The pulse pressure obtained presented good sensitivi-ty to rule out the presence of leakage when it was less than 25mmHg during the first year. After this period, the Endotension Index is the variable that comes closest to the CT scan data. With type I or III endo-leaks, the sensor can identify success or failure after endoleak repair. Until the second year of follow up, the results presented by the sensor agree with CT data, and then become less accurate. Conclusions: In the sample studied sufficient sensitivity could not be achieved to replace the CT when following AAA treated by the en-dovascular technique. A larger number of patients is needed to deter-mine the true role of wireless pressure monitoring when following en-dovascular treatment of abdominal aortic aneurysm
