Learning to Read Data

So that students are able to use tools and libraries in handling data, they need to learn a scripting language. The decision as to which language is suitable depends not only on how easy it is to learn it but also on how broadly it can be applied. The aim is to give students the tools with which they can subsequently work in any research area no matter what the subject matter is. Python offers great advantages here because of its widespread use in many disciplines and because of its open-source nature. Together with Jupyter Notebooks and the JupyterHub, Python-Universum also offers two browser-based tools that greatly facilitate teaching and learning, and allow a prompt start.

In this part of the lecture the focus is not only on introducing the student to collecting and managing data but also on facilitating (automated) reading, writing and editing data in standard formats. For this purpose, students require an introduction to the nature of data that is short and practical, and that puts what is learned into a greater social and scientific context. Next, the lecture shows how to deal with questions on organising, manipulating and converting data with the help of established and generally used tools from the Python “toolkit” (Python Standard Library, Pandas). Finally, the lecture discusses the issue of incomplete, corrupted entries in datasets and what effects incomplete datasets can have on analytical results.

This part of the lecture deals in more detail with the question of how a library and its methods can be developed and used. Generally accessible libraries (Pandas, Matplotlib) that facilitate dealing with data are used to calculate and visualise simple statistical quantities. With the help of these tools students can learn how to interpret the results.

Here again, the focus is not so much on acquiring in-depth theoretical knowledge of statistics as on the ability to apply methods. With the help of libraries (SciPy, Scikit-learn) linear regressions and simple methods from the subject area of machine learning can be quickly applied to accessed datasets. In doing so, it is possible to draw attention to the problems that may occur using the methods or interpreting the results. The aim here is to give guidance in questions relating to

• what methods are suitable to deal with the respective questions and data,
• how the results gained can be used to find answers to the original questions and where the weaknesses of different methods lie.

As part of a broad review, the lecture on “Data Literacy Basics” puts the discussed contents into the overall context of working with data and underlines the significance of aspects of data ethics, data protection and publishing. In this way, students are encouraged to extend and enhance their successfully acquired skills according to their individual needs in further course modules. For this purpose, they can choose from a collection of suitable Open Educational Resources, which forms the third pillar of this project.

Weekly, complementary tutorials are offered with the lecture on “Data Literacy Basics”. In the first phase of these tutorials, students are to learn how to work independently using their data-handling skills with the help of practical and subject-specific examples. The emphasis here is on help in solving subject-specific problems and on acquiring further independent skills in applying the script language and the tools to deal with data-related problems. The following are three practical examples taken from the Faculties of Economics, Social Sciences and Arts respectively:

• In tutorials for students of the Faculty of Economics the complex relationships between socio-economic variables such as working hours, wages and unemployment can be analysed with the help of comprehensive panel databases and by using regression techniques.
• In a tutorial for the Faculty of Social Sciences the online, freely available election manifestos of the parliamentary parties can be uploaded and compared with the help of corpus-based research methods.
• The acquired data-handling skills can be used in the Faculty of Arts, for example, to elucidate topometric tests carried out in connection with the 3D campus laboratory in the Department of Classical Archaeology. The computer-based comparison and analysis here can give the students more detailed insight.

In addition to the tutorials, the examination projects at the end of the lecture on “Data Literacy Basics” are also supervised by the DataLab. Students work at a data-related problem that is relevant to their subject and thus demonstrate that they can apply the acquired skills to a practical situation. Projects are allotted to small groups of 3-5 undergraduate students, whereby each group is supervised by a tutor. Intensive supervision thus guarantees all-round support for the students.

Since 2012 the Centre for Statistics has been offering advice in statistics for those completing a bachelor’s thesis in any area of study at the University of Göttingen. This is complemented by a range of subject-specific advice options offered to undergraduate and postgraduate students by seven different faculties. In setting up a DataLab a central advisory facility can give undergraduate and postgraduate students at the Göttingen Campus an opportunity to follow up any questions about data handling. In this way, existing facilities are being improved and extended. A policy of central and focused information adds to more awareness among students of the DataLab as a source of support.

Part of the work of the advice service is to systematically collect relevant questions and datasets for the purpose of providing, where possible, a feedback for the university teaching programme. In particular, questions, datasets and insight gained are to be prepared in such a way that they can be used as practical examples in the introductory lecture and the corresponding tutorials, the aim being to cover as many areas of data-handling skills as possible. In addition, our Data Consulting also helps tutors to deal with questions in examination projects relating to the planned course.

Based on already existing networks within the Göttingen Campus, the SüdNiedersachsenInnovationsCampus [Southern Lower Saxony Innovation Campus] and the Measurement Valley, the Datalab is intended to function not only as a focal point for businesses and further social actors but also as a platform for an exchange of questions between students, researchers, businesses and non-university institutions addressing specific issues relating to data analysis. This feeds into the planned project dealing with data-specific issues from non-university partners that can be a topic for examination projects.

The emphasis here is primarily on less demanding courses that are also viable for students from the Faculties of Arts and Social Sciences. Within the framework of the project only a limited number of innovatory ideas relating directly to the planned lecture are to be included (for example, recordings of the lecture); instead, with the target group in mind, material already available from other providers is to be compiled and catalogued. The project staff are responsible for this on the one hand, but on the other hand students and teachers may make relevant suggestions and state specific needs. Modules varying in length of time and depth of content can meet the different needs.

With relevant extramural and internal advisory boards collaborating, there is a project-related regulatory procedure in place that determines the overall framework and practical issues of recognition of credits within the curricula. For the purpose of cataloguing, module descriptions are compiled and approved on the basis of this. Similar modules, that are also offered at the same time, are systematically compared as to their suitability and recommended accordingly.

The assembled information on the overall procedure and the catalogued OER is coordinated with partners in the eCULT-Netzwerk and published online. Based on this arrangement, students and course coordinators at other universities can also access data that is linked to the respective ER material. In this way, relevant information about corresponding courses can be made available beyond the University of Göttingen.