Big Data and Analytics: Pathways to Maturity

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 5 of 6
  • Item
    The Risk Management Process for Data Science: Gaps in Current Practices
    ( 2022-01-04) Lahiri, Sucheta ; Saltz, Jeffrey
    Data science projects have unique risks, such as potential bias in predictive models, that can negatively impact the organization deploying the models as well as the people using the deployed models. With the increasing use of data science across a range of domains, the need to understand and manage data science project risk is increasing. Hence, this research leverages qualitative research to help understand the current practices with respect to the risk management processes organizations currently use to identify and mitigate data science project risk. Specifically, this research reports on 16 semi-structured interviews, which were conducted across a diverse set of public and private organizations. The interviews identified a gap in current risk management processes, in that most organizations do not fully understand, nor manage, data science project risk. Furthermore, this research notes the need to a risk management framework that specifically addresses data science project risks.
  • Item
    Deep Learning Strategies for Industrial Surface Defect Detection Systems
    ( 2022-01-04) Martin, Dominik ; Heinzel, Simon ; Kunze Von Bischhoffshausen, Johannes ; Kühl, Niklas
    Deep learning methods have proven to outperform traditional computer vision methods in various areas of image processing. However, the application of deep learning in industrial surface defect detection systems is challenging due to the insufficient amount of training data, the expensive data generation process, the small size, and the rare occurrence of surface defects. From literature and a polymer products manufacturing use case, we identify design requirements which reflect the aforementioned challenges. Addressing these, we conceptualize design principles and features informed by deep learning research. Finally, we instantiate and evaluate the gained design knowledge in the form of actionable guidelines and strategies based on an industrial surface defect detection use case. This article, therefore, contributes to academia as well as practice by (1) systematically identifying challenges for the industrial application of deep learning-based surface defect detection, (2) strategies to overcome these, and (3) an experimental case study assessing the strategies' applicability and usefulness.
  • Item
    Applying Forensic Analysis Factors to Construct a Systems Dynamics Model for Failed Software Projects
    ( 2022-01-04) Kaisler, Stephen ; Money, William ; Cohen, Stephen
    Forensic analysis of failed software projects can aid in managerial understanding of the issues and challenges of delivering a successful project. The factors and their interrelationships causing software project failure are not well understood or researched with a strong forensic-analytic approach. Previous papers have not adequately explored how dynamic interaction of multiple factors can lead to critical events that ultimately portend eventual failure. This paper proposes the development of a System Dynamics (SD) model that will represent the key factors, their dynamic interactions, and the influence of exogenous events in causing software project failure. Forensic data will be used as inputs to the SD model to assist managers in understanding the factor interactions, the importance of individual factor metrics, as well as the sequence of interactions in causing possible software project failure. Outcomes from the model will include a likelihood of software project failure, possible factor sequences leading to failure, and suggestions of remediation activities that might mitigate eventual failure.
  • Item
    Application of the Technology Acceptance Model to an Intelligent Cost Estimation System: An Empirical Study in the Automotive Industry
    ( 2022-01-04) Bodendorf, Frank ; Franke, Joerg
    Cost estimation methods are crucial to support inter- and intraorganizational cost management. Despite intense research on machine learning and deep learning for the prediction of costs, the acceptance of such models in practice remains unclear. The aim of this study is to evaluate the acceptance of an implemented deep learning-based cost estimation system. In an empirical study at a large Bavarian automotive manufacturer we use surveys to collect opinions and concerns from experts who regularly use the system. The evaluation is framed by the basic theories of the Technology Acceptance Model. The results from 50 questionnaires and qualitative participant observations show further development potentials of intelligent cost estimation systems in terms of perceived usefulness and user-friendliness. Building on our empirical findings we provide implications for both research and practice.
  • Item
    A Novel Population Analysis Approach for Analyzing Financial Markets under Crises – 2008 Economic crash and Covid-19 pandemic
    ( 2022-01-04) Hatami, Zahra ; Chetti, Prasad ; Ali, Hesham ; Volkman, David
    Stock markets play an important role in shaping an economic portfolio in many countries and are often used as critical ways to measure economic health and financial status in numerous studies. Financial markets are often volatile and can be influenced by a wide range of direct and indirect variables. The current Covid-19 Pandemic has severely impaired the economic markets in many parts of the world and has negatively affected millions of investors. While some financial markets or stocks are expected to recover or partially recover from this crisis, others may not. With recent crises, such as the 2008 economic crash or the economic impact of the 9/11 event, researchers are looking for innovative ways to analyze the behavior of financial markets under crisis. Can we apply traditional analytical approaches to study the behavior of financial markets under crises, or a different new approach is required to conduct such a study? This paper proposes a new network model and employs a population analysis approach to address such an important research question. We present the basic steps that illustrate how to construct correlation networks of financial stocks and how to utilize graph-theoretic properties of the constructed networks to analyze the behavior of stocks over a given period of time. The proposed population analysis approach allows us to compare the behavior of various groups of companies and their relevant economic sectors in the stock market. We apply the correlation network analysis on different financial data and study the financial implications of two major events, the 2008 economic crash, and Covid-19. In particular, we use the networks to compare the behavior of different economic sectors and uncover the similarities and differences between sectors and their reactions or behavior during these two events. We were able to see certain patterns and extract useful information from the correlation networks. For example, we observed that companies in finance sectors behave in a similar way under the effect of both events and identify some similarities between the behavior of the energy sector during the current pandemic and the utility sector during the 2008 crash.