The Data Factory

Project introduction and background information

Within data science education, students experience four layers of learning that together form the Data Science Pipeline, as follows:

1.    Data Sourcing – covering big data, variety in data (i.e., structured/unstructured) and data collection methodologies;

2.    Data Cleaning and Manipulation – involving governance, storage, removal of errors, fixing artefacts and feature extraction;

3.    Data Analysis – encompassing the exploration processes, descriptive statistics, machine/deep learning and generation of results;

Data Visualisation and Storytelling – presentation of the findings in a meaningful and communicative manner.

These layers take the students on a journey from raw data collection to establishing new understanding and insight in their own specific field of study. The pipeline can be considered as comparable to a Data Factory production line.

However, some specific components of the Data Science Pipeline are abstract in nature and because of this we see that students often struggle to understand form and function, how processes impact on each other and that the layers build on one another; thus, not understanding one layer would make it challenging to understand the subsequent layers.

Therefore, there are two major problems we aim to solve: 1) students do not always understand how form (data concepts) and function (the data processes) go together, and 2) that the processes are dynamic. 

We theorise that these two problems can be addressed by means of harnessing the advanced communication potential of XR technologies (an umbrella term referring to Augmented, Virtual and Mixed Reality technologies); working towards a shift in the way complex data-focused subjects are taught on a wider scale.

XR is proven to have enormous potential for education.

Objective and expected outcomes

The objectives included: 1) investigating the design of an XR software (to be created using game engine technology), whereby the inclusion of the third dimension would enable students to better visualise data science concepts; 2) conducting a creation workshop/hackathon for WUR-based students to come together and design concepts for an XR-game that can be used for wide data educational application; to 3) design and develop a student-led software application.

Results and learnings

The project kicked off with a student hackathon (called the ‘Data Factory hackathon’); an event that was promoted as open to any student at WUR. The event – which took place in November 2023 – and took place over two days on the WUR campus. During the event, student participants were divided into teams, provided with a workshop on XR technologies, given guest talks and then tasked with designing their own creative XR solutions for teaching data science. The resulting showreel of what the students made is available here: https://shorturl.at/wwOCU.

From the event, tangible findings included the project team learning that the students preferred Virtual Reality as a medium for learning over Augmented Reality – this allowed us to narrow the software focus. We also took the ‘best innovation’ elements from the students and combined them into one student-led software application. The software was then developed by students in the SCT Lab (https://www.linkedin.com/company/sct-lab). The demo video is available here: https://shorturl.at/oqNwd. Most recently, the findings from the hackathon were published in the IEEE International Conference on Virtual Reality (https://ieeexplore.ieee.org/abstract/document/10867973). Within an educational setting, the software was used as a study aid during the INF34306-Data Science Concepts course to help students prepare for the exam and was used in the MCB33806- Strategic Marketing for Sustainability Transitions course for students to be introduced to the notion of VR and data-driven innovations.

Recommendations

Always involve the students, their creativity is really surprising. Even though we expected it to be so, the level of creativity and innovation surprised our project team. The results we found from asking the students about XR in education revealed that the usefulness of XR to foster learning is evaluated positively with a mean score of M=5.82 on a 7-point Likert-scale; with wase-of-use is evaluated slightly lower (M=5.50); Even though teams were not expected to develop a prototype, 4 of the 5 did so. We had a nice quote from a student:  ‘At first I did not see the added value of teaching new things in XR, but now I have changed my mind. I now think that it is a fun and interactive way to learn new concepts’. Students particularly valued the learning strategies of ‘effort’ and ‘reflection’, both of which are integral to XR applications. Survey response declined during the hackathon- likely due to the event's complex and competitive nature, making it challenging for participants to focus on completing surveys – so our advice to others doing a hackathon would be, try not to over-burden students with additional questions; acting as a distractor in the creative process.

Practical outcomes

Practical outcomes include:

• new software on GitHub – the Data Factory - which works on PC, Laptop, VR headset and Mobile phones.
• new publication – specifically related to how to create hackathons for XR applications for data science education, and second in progress
• new study aid for students, currently used in two courses.
• 1 new grant funding application to Horizon EU