Written Catherine Upex and Rachel Wood
Introducton
Earlier this summer, the University of Bristol and the JGI welcomed a group of visiting researchers from the “Oceanic and Southeast Asian Navigators” (OCSEAN) project. OCSEAN is a worldwide interdisciplinary consortium researching the demographic history of ancient seafarers across Oceania and Southeast Asia. The visiting humanities researchers from Indonesia and the Philippines arrived in Bristol with the aim of learning more about quantitative methods, how to apply them to their research, and to take these skills home to help their research community do the same.
When asked, most said they had little to no knowledge or experience in coding. The task therefore was to design a training approach to help them feel confident independently using Python for research – all in the space of a few weeks.
Our Approach
The training style followed a traditional workshop format, but importantly with two instructors. This allowed one to talk through the course content, and the other to provide one-to-one help to individuals. Initially, the sessions consisted of lecture-style teaching, but as confidence grew, they transitioned to a more independent format, where small groups collaborated to solve data science problems directly related to their research interests.
As most participants has no prior coding experience, it was important not assume any knowledge of technical terms. Over eight two-hour sessions spanning three weeks, the training slowly built-up coding knowledge, covering the following topics:
- Introduction to Python (e.g. variables, data types, operators, lists, dictionaries)
- Intermediate concepts (e.g. using/writing functions, loops, conditional statements)
- How to use Chatbots for coding (e.g. how to write good prompts, refine responses, when/when not to use, error handling, and sanity checking)
- Data analysis (e.g. loading/cleaning data, plotting using seaborn and matplotlib, summarising data)
The training also coincided with Bristol Data Week 2025, so the OCSEAN researchers had the opportunity to cement their knowledge by revisiting concepts in similar training sessions from the event.
Comparing training styles
The approach differed to a recent pilot training scheme run by JGI Research Data Science Advocates. The aim of the pilot was to run training on data analysis in Python in a low-stress environment, via a self-led approach. Participants were supplied with materials to work through independently, with optional contact time with facilitators.
Both training styles were designed for researchers with no prior coding experience. It was interesting to see how the hands-on and hands-off approaches compared in order to understand how to most effectively encourage engagement with data science.
Feedback from OCSEAN researchers
By the end of our training period, all the OCSEAN researchers said that they found the training very beneficial for their research. Many acknowledged that they found learning Python challenging. However, the format of the sessions, especially the opportunity to draw upon help from not only facilitators but also ChatGPT, and importantly each other, allowed them to get to grips with new concepts. Intensive successive trainings with a clear syllabus were seen as more beneficial than one-off unconnected sessions.
The importance of structured training was echoed by feedback from the self-led pilot training. Here, participants highlighted that despite a self-led approaching being easier to fit into a working week, they would have benefitted from group discussions and the opportunity to compare their results with others. Additionally, while most of the self-led participants agreed that the pilot scheme facilitated their learning outcomes and expressed a desire to apply what they learnt to their work, some commented that they lacked a basic understanding of Python to independently apply these skills.
Importantly, OCSEAN researchers commented on how it wasn’t just the training structure that facilitated learning. Aspects such as the use of a small meeting room and the inclusion of regular breaks, further encouraged collaboration between participants and drove better understanding. Additionally, the use of datasets adapted to participants’ research fields made coding seem much more accessible and engaging. This highlighted how important it is to facilitate a supportive and personalised teaching environment in order to fully grasp new complex concepts.
Reflections and moving forward
This training was facilitated by two PhD students developing their own teaching skills, and the experience taught the team a lot about what makes effective data science training. To feel confident in independently using data science, intensive face-to-face training is needed to make sure basic coding skills are cemented. This can be difficult for many to fit in, but a weekly commitment, combined with a hand-on collaborative atmosphere can effectively drive key concepts home.
Additionally, to drive engagement particularly from disciplines with little data science background, it is important to cater training to specific research questions in that field i.e. using relevant data sets. This way, participants can see how data science can help them in their own research and be more inspired to try for themselves.
So, what’s next? The aim of this training was to provide OCSEAN researchers with data science skills to apply to their own research. It’s been brilliant to see that some have already taken this leap. Using their coding skills and connections made in Bristol, many are developing new projects, applying for PhD positions and forming future collaborations. In the Autumn, the team plan to travel to Bali to aid OCSEAN researchers in sharing coding skills with their research communities, as well as developing more research collaborations.
This blog was written by Catherine Upex and Rachel Wood
Learn more about the OCSEAN project here or contact Daniel Lawson (Dan.Lawson@bristol.ac.uk) or Monika Karmin (monika.karmin@ut.ee) for more information.