56 research outputs found
Elastic SCAD as a novel penalization method for SVM classification tasks in high-dimensional data
No full textAbstract Background Classification and variable selection play an important role in knowledge discovery in high-dimensional data. Although Support Vector Machine (SVM) algorithms are among the most powerful classification and prediction methods with a wide range of scientific applications, the SVM does not include automatic feature selection and therefore a number of feature selection procedures have been developed. Regularisation approaches extend SVM to a feature selection method in a flexible way using penalty functions like LASSO, SCAD and Elastic Net. We propose a novel penalty function for SVM classification tasks, Elastic SCAD, a combination of SCAD and ridge penalties which overcomes the limitations of each penalty alone. Since SVM models are extremely sensitive to the choice of tuning parameters, we adopted an interval search algorithm, which in comparison to a fixed grid search finds rapidly and more precisely a global optimal solution. Results Feature selection methods with combined penalties (Elastic Net and Elastic SCAD SVMs) are more robust to a change of the model complexity than methods using single penalties. Our simulation study showed that Elastic SCAD SVM outperformed LASSO (L1) and SCAD SVMs. Moreover, Elastic SCAD SVM provided sparser classifiers in terms of median number of features selected than Elastic Net SVM and often better predicted than Elastic Net in terms of misclassification error. Finally, we applied the penalization methods described above on four publicly available breast cancer data sets. Elastic SCAD SVM was the only method providing robust classifiers in sparse and non-sparse situations. Conclusions The proposed Elastic SCAD SVM algorithm provides the advantages of the SCAD penalty and at the same time avoids sparsity limitations for non-sparse data. We were first to demonstrate that the integration of the interval search algorithm and penalized SVM classification techniques provides fast solutions on the optimization of tuning parameters. The penalized SVM classification algorithms as well as fixed grid and interval search for finding appropriate tuning parameters were implemented in our freely available R package 'penalizedSVM'. We conclude that the Elastic SCAD SVM is a flexible and robust tool for classification and feature selection tasks for high-dimensional data such as microarray data sets.</p
<it>FACT </it>– a framework for the functional interpretation of high-throughput experiments
No full textAbstract Background Interpreting the results of high-throughput experiments, such as those obtained from DNA-microarrays, is an often time-consuming task due to the high number of data-points that need to be analyzed in parallel. It is usually a matter of extensive testing and unknown beforehand, which of the possible approaches for the functional analysis will be the most informative Results To address this problem, we have developed the Flexible Annotation and Correlation Tool (FACT). FACT allows for detection of important patterns in large data sets by simplifying the integration of heterogeneous data sources and the subsequent application of different algorithms for statistical evaluation or visualization of the annotated data. The system is constantly extended to include additional annotation data and comparison methods. Conclusion FACT serves as a highly flexible framework for the explorative analysis of large genomic and proteomic result sets. The program can be used online; open source code and supplementary information are available at http://www.factweb.de.</p
CebraNET Cellular Membranes in Volume SEM
No full textDownload RDF Package
You need to sign in or sign up before continuing.
EMBL GitLab Server
The EMBL Git server offers all EMBL staff members a convenient way to store, control and publish the source code of their projects.
It uses gitlab, which offers commandline and web access to the stored data.
Have a look at the public projects
EMBL
Standard
EMBL Username
Password
Remember me
Username or email
Password
Remember me
Forgot your password?
Sign in
Usage:
read our list of Frequently Asked Questions
git is meant predominantly for handling source code and text files, not for storing large files such as images, etc. Please make sure your repository size stays below 2GB!
Login is provided using EMBL LDAP authentication, external users need to be registered first by an EMBL user. For this, send an email to [email protected] with the external user's name, affiliation, and email address. External users can not create their own repositories
Some hints about the visibility levels of your projects:
Public: visible for everyone - google will find it
Internal: only EMBL users will be able to see it
Private: for project member eyes only
The following rules apply:
Repositories on the EMBL Git server are only granted to EMBL members
External users can be added as cooperators on a project, but the projects themselves have to be lead by someone with an active EMBL contract
Should the project leader leave EMBL, then the project has to be transferred to someone else or the complete repository will be removed
Contact
Please contact [email protected] if you have questions regarding this service
Administration and Maintenance:
Renato Alves
Jelle Scholtalbers
Acknowledgements:
The system was originally set up by Holger Dinkel and Grischa Toedt
Former project members:
Marc Gouw, former admin
Toby Hodges, former admin
Frank Thommen (Structures IT Management & Support), who together with Grischa and Michael implemented the original git repository service at EMBL
Wasiu Akanni
Jakob Wirbel
People involved in the setup: Michael Wahlers and Carlos Fernandez San Millan
The EMBL Git server is part of the Bio-IT Project
The server, disk space and backup are kindly provided by EMBL IT Services (internal link)
By signing in you accept the Terms of Use and acknowledge the Privacy Policy and Cookie Policy.
Sign in with
Login with EMBL/EBI credentials (SSO)
Remember me
Explore
Help
About GitLab
Community forumCellular membrane prediction model for volume SEM datasets (Uploaded via https://bioimage.io
Identifizierung kausaler Zusammenhänge : Aufsätze zur empirischen Wirtschaftsforschung
No full textThis dissertation discusses the potentials and pitfalls of empirical economic research. Several pieces of applied research illustrate the discipline's diverse use of statistical methods as well as their applicability to different topics.
Empirical economic research uses empirical evidence to test hypotheses and statistical inference to uncover general rules. However, very often several rules or causal mechanisms exist that can equally well explain the investigated outcome. This can be problematic whenever statistical inference does not yield convincing results, i.e. the degree of the study's internal validity is low. Although many different statistical tools and techniques have been developed to increase the degree of internal validity, in practice it remains difficult to claim causality. One reason for this is that the quality of statistical inference depends on the appropriateness of the chosen statistical method. Another reason is that causality requires that competing alternative explanations for an estimated statistical relationship are addressed and at best can be dismissed. Thus, an empirical study's overall quality depends critically on an author's judgement and knowledge of the environment in which the outcome is nested. Given the importance of personal perception it is not surprising that the validity of results in many studies in empirical economics is heatedly discussed in- and outside the community.
At the beginning of this dissertation the current status of this academic debate is reproduced, leading to the conclusion that there is no panacea for causal inference. Instead, it is proposed that several equally sensible strategies to strengthen causality exist and that their selection depends on the specific research question and setting. While this still allows the author to base decisions on personal perception it also stresses that justifications are required. Hence, each new study demands a tailored research agenda in which the choice of statistical methods and the existence of alternative explanations are transparently discussed. Subsequently to the discussion, three independent papers illustrate that there is indeed no blueprint procedure to conduct empirical economic research.
The first paper addresses the question whether individuals react to natural disasters by adjusting their saving behavior. The study applies statistical tools commonly used in applied microeconomic research. The research design uses quasi-experimental variation in a panel survey to infer a causal relationship between flooding and saving behavior. The study finds that from the flooding affected individuals save less in subsequent years. While the study's internal validity is rather high, the generalizability of the relationship remains to be seen. Several alternative explanations for the observed behavior are discussed and evaluated. The concluding explanation is that unusually high amounts of post-disaster financial aid induces moral-hazard-behavior. Thus, the paper makes a case for policy makers to carefully design post-disaster aid payments so as to minimize the possibility of detrimental reductions in individual precautionary efforts.
The second paper investigates the link between foreign education and domestic productivity. The paper uses aggregate data, thus encountering statistical challenges commonly occurring in applied macroeconomic research. The research design focuses on the dynamic structure of the data. The paper finds that the more students a country sends to the U.S., the higher subsequent domestic productivity growth rates will be. Additional analyses show that this effect is driven by developing countries. It is argued that the relationship is causal because foreign students transfer productivity enhancing skills from the U.S. to their home country. However, the data does not reveal whether foreign students indeed return and therefore causal inference is weaker than it could be otherwise. Measures to overcome this shortage are presented and applied. Nonetheless, the extent of the data allows for a certain generalizability of the results. In conclusion, the study suggests that foreign education poses a viable additional strategy for economic development.
Finally, the third paper addresses a research question from the field of empirical industrial organization. Specifically, the paper tests whether prices for an abatement technology are influenced by the type of environmental regulation of polluting sources. In order to test this relationship, the paper's research design combines a structural economic model with quasi-experimental empirical evidence. The paper finds that the price of abatement technology is significantly higher for those polluting sources that are participating in a permit trading scheme. Causal inference relies on the quasi-experimental nature of the data and the theoretical derivations from the structural model. However, it remains empirically challenging to exclude alternative explanations as doing so considerably strains the scope of our data. In the end, the study's results should caution policy makers to consider that regulatory instruments can have unintended side-effects hampering the diffusion and adoption of abatement technology by increasing its price.
The final section discusses the role of empirical research in the overall process of scientific progress. The importance of diverse and comprehensive empirical economic research is emphasized. Finally, it is concluded that empirical research with all its outgrowth is essential to establish something like an objective truth'' in economic science.Die vorliegende Dissertation beschäftigt sich mit den Möglichkeiten und Grenzen der empirischen Wirtschaftsforschung. Anhand mehrerer Forschungsarbeiten verdeutlicht sie die Anwendbarkeit statistischer Verfahren auf verschiedene Fragestellungen aus den Wirtschaftswissenschaften.
In der empirischen Wirtschaftsforschung werden Beobachtungen statistisch ausgewertet, um Hypothesen zu testen und allgemeine Regeln aufzudecken. Die Herleitung eines kausalen Zusammenhangs zwischen zwei Ereignissen gilt dabei als ein wichtiges Ziel. In der Praxis erweisen sich kausale Schlussfolgerungen allerdings als überaus schwierig. Ein Grund hierfür ist, dass der Gegenstand empirischer Wirtschaftsforschung - unsere Gesellschaft - ein komplexes und dynamisches System ist. Eine allgemeingütige Blaupause, mithilfe welcher sich kausale Zusammenhänge belegen lassen, lässt sich daher kaum entwickeln. Vielmehr hängt die Herleitung eines kausalen Zusammenhangs von den konkreten Faktoren des Einzelfalls ab. Solche Faktoren sind die Forschungsfrage, die Qualität der Daten und das Umfeld, in welchem diese erhoben wurden. Sie bestimmen anschließend das Forschungsdesign und die Auswahl eines geeigneten statistischen Verfahrens. Eine empirische Studie ist somit in vielerlei Hinsichten einzigartig. – Daher ist es auch nicht verwunderlich, dass Kausalitätsbehauptungen in Bezug auf empirische Ergebnisse in den Wirtschaftswissenschaften häufig kontrovers diskutiert werden.
Zu Beginn dieser Dissertation wird eine aktuelle Diskussion zur Herleitung von Kausalität in der empirischen Wirtschaftsforschung wiedergegeben. Aus dieser Diskussion geht hervor, dass es momentan kein Allheilmittel für kausale Inferenz gibt, sondern stattdessen mehrere gleichermaßen sinnvolle Strategien für die Herleitung von Kausalität existieren. An die Wiedergabe und Auswertung der Diskussion schließt sich die Darstellung dreier unabhängiger empirischer Studien an. Jede dieser Studien befasst sich mit einem anderen Themengebiet der Wirtschaftswissenschaften, wobei Forschungsdesign, Auswahl der empirischen Methoden und die Art der kausalen Herleitung variieren. Die drei Studien illustrieren somit mehrere Punkte, die sich aus der Diskussion in der Einleitung ergeben haben.
Der erste empirische Beitrag in dieser Dissertation geht der Frage nach, ob Opfer von Naturkatastrophen im Anschluss an ihre Erlebnisse ihr Sparverhalten verändern. In der Studie werden statistische Methoden verwendet, die üblicherweise in der angewandten mikroökonomischen Forschung verwendet werden. Das Forschungsdesign nutzt die durch eine Flut generiete quasi-experimentelle Variation in einer Panelbefragung aus, um einen Kausalzusammenhang zwischen Betroffenheit und Sparverhalten abzuleiten. Die Studie zeigt, dass von den Überschwemmungen Betroffene in den Folgejahren weniger sparen. Während der Grad der kausalen Schlussfolgerung hoch ist, bleibt es abzuwarten, ob sich der Zusammenhang auf andere Situationen übertragen lässt. Es werden mehrere Gründe für das beobachtete Verhalten diskutiert und gegeneinander abgewogen. Die Studie kommt zu dem Ergebnis, dass ungewöhnlich hohe Hilfszahlungen zu einem sogenannten Moral Hazard-Verhalten‘‘, also einem verantwortungslosen Verhalten aufgrund von Fehlanreizen, führen können. In dem untersuchten Fall haben die Hilfszahlungen zu einer Verringerung im Vorsorgeverhalten bei Betroffenen geführt. Die Studie plädiert daher dafür, dass politische Entscheidungsträger etwaige Hilfszahlungen nach einer Katastrophe sorgfältig planen, um einen nachteiligen Einfluss auf individuelle Vorsorgemaßnahmen zu vermeiden.
In der zweiten Arbeit wird die Verbindung zwischen einem Studium im Ausland und heimischer Produktivität untersucht. Das Papier verwendet dafür aggregierte Daten und befasst sich aus ökonometrischer Sicht mit bestimmten statistischen Herausforderungen, die häufig in der angewandten makroökonomischen Forschung auftreten. Das Forschungsdesign fokussiert sich auf die dynamische Struktur der Daten, um einen kausale Herleitung zu ermöglichen. Gezeigt wird, dass die Anzahl von Studenten, die ein Land in die USA schickt, sich positiv auf die Produktivitätszuwächse dieses Landes in den Folgejahren auswirkt. Des Weiteren kann gezeigt werden, dass dieser positive Zusammenhang nur für Entwicklungsländer gilt. Dies erscheint plausibel, da insbesondere die Bevölkerung aus Entwicklungsländern durch den Transfer produktivitätssteigernder Fähigkeiten aus den USA in ihr Heimatland profitieren sollte. Die Ergebnisse deuten daher an, dass es tatsächlich einen positiven Kausalzusammenhang zwischen Auslandsstudium und heimischer Produktivität gibt. Da die Daten jedoch nicht darüber informieren, ob ausländische Studenten wirklich zurückkehren, ist die kausale Inferenz schwächer, als sie es sonst sein könnte. Hingegen erlaubt der Umfang der Daten eine gewisse Generalisierbarkeit der Ergebnisse. Zusammenfassend lässt sich festhalten, dass die Förderung eines Auslandsstudiums eine sinnvolle zusätzliche Strategie für eine erfolgreiche internationale Entwicklungszusammenarbeit darstellen kann.
Das dritte und letzte Papier befasst sich mit einer Forschungsfrage, die dem Bereich der empirischen Industrieökonomik zugeordnet werden kann. Darin wird untersucht, ob die Regulierung von Schwefeldioxidemissionen von Kohlekraftwerken die Preissetzungsstrategie von Herstellern einer geeigneten Vermeidungstechnologie beeinflusst. Um diese Beziehung zu testen, nutzt die Studie ein Forschungsdesign, das auf einem Strukturmodell und Daten mit quasi-experimenteller Variation basiert. Die Ergebnisse der empirischen Untersuchung zeigen, dass die Preise für die Vermeidungstechnologie höher sind, wenn ein Kohlekraftwerk an einem Emissionshandelssystem teilnehmen muss. Diese Entwicklung ist kontraproduktiv, da es den Anreizen eines Emissionshandelssystems, die Verbreitung von Vermeidungstechnologien zu fördern, entgegen wirkt. Die Herleitung eines kausalen Zusammenhangs beruht auf dem quasi-experimentellen Charakter der Daten sowie einem theoretischen Modell, welches den empirischen Befund ebenfalls vorhersagt. Die wichtigsten alternativen Erklärungen für das Ergebnis können ausgeschlossen werden. Einschränkend wirkt hierbei jedoch der Umfang der Daten. Dieser lässt eine rigorose Untersuchung alternativer Erklärungen nur begrenzt zu und schwächt somit den kausalen Zusammenhang etwas ab. Am Ende unterstreichen die Ergebnisse der Studie allerdings, dass politische Entscheidungsträger bei der Ausgestaltung regulatorischer Instrumente umfassend auf unbeabsichtigte Nebenwirkungen achten sollten.
Im letzten Abschnitt der Dissertation wird die Rolle empirischer Forschung im Gesamtprozess des wissenschaftlichen Fortschritts diskutiert. Dabei wird die Bedeutung einer umfangreichen und vielfältigen empirischen Wirtschaftsforschung hervorgehoben. Abschließend wird festgestellt, dass die empirische Forschung mit all ihren Ergebnissen und Methoden notwendig ist, um eine objektive Wahrheit'' in der Wirtschaftswissenschaft zu generieren
Chester Morris and Betty Compson: in 'The case of Sergeant Grischa'
No full textNo description availabl
Valuation of altmetrics in research funding
No full textThis master s thesis is about the potential valuation of altmetrics or alternative metrics in research funding, which is apparent in current high-level policy debates in higher education. Altmetrics measure the outputs of scholarly research online. Valuation is defined not in the monetary sense of the word, but as giving worth to something as a social construct. Based on the Sociology of Valuation and Evaluation, the author intends to map the potential usage and valuation of altmetrics in research funding. A mixed method research design was chosen for this study. Firstly, a review of policy papers from supranational organisations, national governments, and organisations in higher education was carried out. Secondly, qualitative interviews (n=6) with research policy makers and members of a research funding organisation in Finland were conducted. Thirdly, the quantitative phase consisted of four online surveys (n=290) with researchers at a university and reviewers in Finland and on an international level. Finally, these data sets were analysed together (N=296). The findings suggest that altmetrics is mostly unknown and of low importance among the study participants, and only a small amount of altmetrics users could be identified. It is a prominent research policy topic these days, and considered as on the rise in debates on higher education. And, despite the unawareness and little valuation of altmetrics, some respondents use altmetrics in some way or the other, and are highly-aware of the concept of altmetrics. Altmetrics might be more important in future in the reporting phase compared to the research funding application phase. Considering the current high-level policy debates, it is recommended to stakeholders in the higher education system to become familiar with altmetrics, as they might play a larger role in future. Policy makers need to communicate more clearly on the challenges of research impact assessments, and altmetrics
Software Engineering PhD and Licentiate Theses in Sweden: Publication statistics
No full textThis simple dataset contains publication statistics of Swedish PhD and Licentiate thesis in Software Engineering from 1999 to 2018. The contents of this dataset were discussed in a blog post on https://grischaliebel.de.
The data is offered in two formats, xlsx and csv, but with the same content. Names and affiliation are anonymised in the data set to prevent identification of subjects. In the following, we describe the content of the different columns in the table.
Level: 'lic' for Licentiate theses or 'phd' for PhD theses
Year: The year of publication of the thesis
Included: The total number of papers included in the compilation-style thesis.
Listed: Number of papers listed in addition to the included papers (basically "I have also published these, but they are not relevant to the thesis). Note that we cannot distinguish between cases, where no papers are listed because none are published, or because the author decided not to list them.
IncludedPublished: The amount of included papers that are published or accepted for publication.
IncludedSubmitted: The amount of included papers that in submission/under review.
IncludedPublishedISI: The amount of included, published papers that are in ISI-ranked journals.
IncludedPublishedNonISIJ: The amount of included, published papers that are in non ISI-ranked journals.
IncludedPublishedConf: The amount of included, published papers that are in CORE-ranked conferences (any grade).
IncludedPublishedWS: The amount of included, published papers that are in workshops. Non CORE-ranked conferences are counted as workshops as well.
IncludedPublishedOther: The amount of included, published papers that do not fit in any other category (e.g., book chapters, technical reports).
IncludedSubmitted*: Amount of included, submitted papers broken down by category (Journal, conference, workshop, and other).
ListedPublished*: Amount of listed, published papers broken down by category (ISI/Non-ISI Journal, conference, workshop, and other).
ListedSubmitted*: Amount of listed, submitted papers broken down by category (Journal, conference, workshop, and other)
A Conceptual Model For Web Accessibility Requirements In Agile Development
No full textPublisher Copyright:
© 2024 Copyright is held by the owner/author(s). Publication rights licensed to ACM.Accessibility is the practice of making content and functionality accessible to all users, regardless of their abilities. Although accessibility is a highly relevant quality attribute, it is often treated as an afterthought in software development, unfortunately excluding people with disabilities from using many web-based systems. Specifically in agile development, sprints focus on new features and quality attributes, such as accessibility, are often not considered sufficiently. In these cases, using conceptual models to understand and analyze requirements that developers have formulated as a set of related user stories is a research opportunity. To increase agile professionals' focus on accessibility, we built a conceptual model for web accessibility, identifying artifacts and concepts used in agile development to specify accessibility. We discuss how this model can be used as a guide to better integrate accessibility considerations into agile software development. Researchers can use the result to define resources that are not currently covered or improve underutilized practices. We plan to use the conceptual model in the next steps to adapt existing agile artifacts and create support tools for web accessibility in agile development.publishersversionpublishe
- …
