Claudia Koschtial

Representing Multidimensional Cancer Registry Data

Epidemiology requires the analysis and visualization of massive data sets. The field of cancer statistics in particular is facing the challenging task of visualizing a large data set that contains a wide range of available dimensions. The existing work of epidemiologists has been time-consuming because of visualization techniques that could not be scaled to support an unguided exploration process. This limitation has led to the inefficient use of data representations that are mainly used for detailed analysis. Our goal was to find a scalable visualization technique that focused on covering a wide range of categorical information. For this purpose, a task by data type taxonomy is used to analyze the existing data visualization techniques. The chosen representation was based on the implemented flow visualization and provided an overview for exploring the data by epidemiologists. In this way, a more scalable visualization delivered the ability to support the creation of hypotheses by finding relationships of interest.

A Novel Approach for Decision Support in Uncertain Environments: The Case of Identifying Similar News Tickers in Natural Gas Trading

As the natural gas market is characterized by price volatility and uncertainty, the participants need in using founded information for decision making in order to manage risks and profits is immense. Analyzing the information content of news tickers can provide additional information about the environment. This paper concerns the features of a novel method that encourages price prognosis in gas trading. By knowing how the market has developed regarding a certain former situation, this knowledge can be used in predicting the future market by associating a similar former state to a present state. Thereby, uncertainty about future might be reduced. Framing the research in design science, we use task-technology-fit theory and technology-acceptance-model to identify requirements and to appraise the artifact. The novel method integrates structured and unstructured data in decision support.

Detecting similar news tickers in the area of natural gas trading: improving decision support in uncertain situations

The volatility of the natural gas market founded a need for the ability to analyze upcoming events in real time in order to manage profits and risks for participants. News ticker provide information being of utmost importance for the analysis. The research presented among this paper describes features of a software prototype supporting the analytical price prognosis tasks for gas traders. By knowing market development at the time of a certain past situation, the outcome of that situation can be used to predict the future market development of a current analyzed situation with similar content. For that, similar situations have to be detected in order to reduce uncertainty about future. Fitting into design science, we use task-technology-fit theory and technology-acceptance-model to identify information needs and to evaluate the artifact. This novel approach serves as a further step to gain a decision support with integrated structured and unstructured data.

Science with and without e

Though the final extent of the change is difficult to estimate, the perception of e-science as a transformation of the everyday professional lives of scientists (scholars, researchers) is common. Besides, there is only little knowledge about the everyday professional live of scientists. Therefore, it is difficult to measure the change, but that would facilitate an understanding. Furthermore, it is important to recognize performed tasks or relevant processes and how they are carried out in order to understand potentials and limitations of e-science. We analyzed everyday professional lives of scientists by a qualitative study with 19 researchers among various disciplines and identified attributes like role, collaboration, and internationalization leading to a usage of tools among the context of e-science. The presented research is developed in the theoretical foundation of Heinrichs Human-Task-Technology framework and enables herewith a guided further exploration of the topic.

Towards an Evaluation Framework to Structure Business Intelligence Project Patterns as Enhancement of Business Intelligence Maturity Models

Maturity models are introduced to support the evaluation of the status quo of IT inside a company and to identify next applicable steps, but it has been revealed that Business Intelligence (BI) maturity models are still in a learning phase so far. Existing models are not mainly based within practical environment, yet, so that the positioning of the own company’s status quo and comparison to competitors does not help in decision making in context of next steps in BI investments. We are analyzing several existing BI project descriptions, which were part of the TDWI BI Best Practice Award, to be able to contribute to the discussion of BI project definitions. Project patterns are identified, which are organized in a blueprint. This result serves as a first step and road sign towards a scalable BI management support, to increase the discussion about maturity models and to support BI strategies and investment decisions for project definitions.

On the Way from a Knowledge Discovery in Databases to a Predictive Analytics

Business Intelligence has “decision support” as a characterizing element. Decisions are done as a selection process based on alternatives. The choice depends on prospective developments whereby those developments are predicted with uncertainty. Due to this reason, forecasts are getting more into focus of the strategic and tactical level. But forecasts, usually based on Knowledge Discovery in Databases (KDD), are limited, yet. They often produce non-adequate results, which can lead to wrong decisions. Such a forecast quality demands further research in identifying improvements to increase reliability of forecast results and its usage in practice. This chapter modifies the Knowledge Discovery in Databases to improve the forecast quality. The associated process is supplemented by further steps to enhance the analyzed data set with additional future oriented data by using the KDD markup language. First results of an evaluation implementation at a German saving and loans bank shows motivating results.

Providing a Method for Supporting the Decision Making About a Meaningful XBRL Implementation According to the Specific Situation of an Organization

XBRL is in an increasing number of countries defined as the obligate format for the transmission of financial data to authorities by legal requirements. The legal requirement itself refers solely to the transmission of the information in XBRL format. How a formatting of a company’s report becomes realized is not regulated. This decision has to be done by the decision makers of the affected companies. There is no generally recommended realization or global valid implementation process. Each realization (bolt-on, built-in, and embedded) comprises a specific realization effort and offers different potential benefits for the implementing organization. Therefore, it is necessary to decide about the integration depth within an organization. The paper presents a methodology to support the decision process for the adoption of XBRL into the organization. By doing design science research, the constructed artifact enables a structured decision based on facts, individual properties, and needs of the organization.

Break-Up Analysis: A Method to Regain Trust in Business Transactions

The financial crisis resulted in a loss of trust; not only within the investment or banking sector, but in general between creditors and debtors, because many organizations faced insolvency. Such a financial situation can even result in a company’s bankruptcy. Therefore it is necessary to get a realistic understanding of the solvency or the possible insolvency of a company. The support of a decision on a debtor’s creditability is not yet sufficiently provided by the most prominent method (Altman’s Z’’-score). The paper presents a procedure called Break-Up Analysis (BUA). It helps to decide on the solvency of a company. The comparison of the BUA to Altman’s Z’’-score shows an improvement of the identification of solvent and insolvent companies by 22 %. The BUA enables herewith to regain trust in business transaction by not identifying only the insolvent companies but the solvent ones as well.

Predictive Analytics in der Strategischen Anlagenwirtschaft

Predictive Analytics (ein deutschsprachiger Begriff hat sich bisher nicht ausgepragt) dient dem Erkenntnisgewinn zur Entscheidungsunterstutzung in Unternehmen. Anwendungsgebiete finden wir immer dort, wo komplexe Verfahren zur zukunftsorientierten Datenauswertung notwendig sind. Beispielsweise wird es angesichts einer gestiegenen Kapitalintensitat in vielen Branchen, insbesondere der Energiewirtschaft, zunehmend erforderlich, den Anlagenlebenszyklus im Ganzen zu betrachten, um diesen effizient zu steuern. Dabei erfordern analytische Verfahren einen ganzheitlichen Rahmen, um auf Wechselwirkungen und Optimierungspotenziale hinzuweisen. In diesem Sinne ist Predictive Analytics ein Werkzeug zur strategischen Unternehmenssteuerung [44]. Sich mit der Predictive Analytics auseinandersetzende Publikationen zeigen eine Zunahme im Zeitraum von 2000 bis 2010 (siehe Abb. 22.1, links). Der rechte Abbildungsteil zeigt die Beitrage, die das Thema Predictive Analytics im expliziten Fokus aufweisen. Dabei ist ein zeitlicher Verlauf erkennbar, der an den Gartner Hype Cycle erinnert. In der Finanzmarktkrise wurden Prognoseprozesse nicht mehr ausgefuhrt, weil man deren Ergebnissen nicht mehr vertraute [17]. Eine solche Desillusionierung kann den Ruckgang an praxisorientierten Arbeiten begrundet haben.