Deep learning models have revolutionized research fields like computer vision and natural language processing by outperforming traditional models in multiple tasks. However, the field of time series analysis, especially time series forecasting, has not seen a similar revolution, despite forecasting being one of the most prominent tasks of predictive data analytics. One crucial problem for time series forecasting is the lack of large, domain-independent benchmark datasets and a competitive research environment, e.g., annual large-scale challenges, that would spur the development of new models, as was the case for CV and NLP. Furthermore, the focus of time series forecasting research is primarily domain-driven, resulting in many highly individual and domain-specific datasets. Consequently, the progress in the entire field is slowed down due to a lack of comparability across models trained on a single benchmark dataset and on a variety of different forecasting challenges. In this paper, we first explore this problem in more detail and derive the need for a comprehensive, domain-unspecific overview of the state-of-the-art of commonly used datasets for prediction tasks. In doing so, we provide an overview of these datasets and improve comparability in time series forecasting by introducing a method to find similar datasets which can be utilized to test a newly developed model. Ultimately, our survey paves the way towards developing a single widely used and accepted benchmark dataset for time series data, built on the various frequently used datasets surveyed in this paper.

Exploratory data analysis (EDA) is an iterative process where data scientists interact with data to extract information about their quality and shape as well as derive knowledge and new insights into the related domain of the dataset. However, data scientists are rarely experienced domain experts who have tangible knowledge about a domain. Integrating domain knowledge into the analytic process is a complex challenge that usually requires constant communication between data scientists and domain experts. For this reason, it is desirable to reuse the domain insights from exploratory analyses in similar use cases. With this objective in mind, we present a conceptual system design on how to extract domain expertise while performing EDA and utilize it to guide other data scientists in similar use cases. Our system design introduces two concepts, interaction storage and analysis context storage, to record user interaction and interesting data points during an exploratory analysis. For new use cases, it identifies historical interactions from similar use cases and facilitates the recorded data to construct candidate interaction sequences and predict their potential insight—i.e., the insight generated from performing the sequence. Based on these predictions, the system recommends the sequences with the highest predicted insight to data scientist. We implement a prototype to test the general feasibility of our system design and enable further research in this area. Within the prototype, we present an exemplary use case that demonstrates the usefulness of recommended interactions. Finally, we give a critical reflection of our first prototype and discuss research opportunities resulting from our system design.

With the latest advances in digitalization and Industry 4.0, the manufacturing industry is collecting more and more production data. However, with the increasing interconnection of machines, not only the volume but also the variety of data is being expanded. The data life cycles of collection, processing, combining, analyzing and feeding new findings back into sources are becoming increasingly challenging for data scientists to complete. Reference architectures such as the RAMI 4.0 provide conceptual guidelines to address these problems. In this paper, we focus on the implementation of an agent-based architecture that is in line with RAMI 4.0. This architecture implements the guidelines provided by RAMI 4.0 by applying modern approaches from the areas of data lake based data acquisition, semantic description, look up and processing as well as information utilization.

In exploratory data analysis, domain knowledge and experience play a central role in order to extract information from the data and to derive proof and knowledge. However, experienced domain experts are rarely the same people who carry out the data analyses. Therefore, utilizing domain expertise for guidance in analytic processes is a complex challenge. In recent years, machine learning has seen great advances. Increasing processing power and growth in data as well as affordable storage have led to more advanced algorithms. Therefore, with the emergence of applicable machine learning algorithms, there is now a method for preserving and making use even of complex knowledge. In this paper, we present a concept that allows to extract and utilize domain knowledge for exploratory data analysis. We introduce concepts of interaction store and analysis context store to record user interaction and context during an exploratory analysis. We use the recorded data to construct semantic interaction sequences and predict their potential insight. The prediction can then be used to guide other data scientist in their sensemaking while performing exploratory data analysis in similar domains and use cases. Furthermore, we discuss possible research opportunities and implications resulting from the presented concept.