Today, 07.10.2026 at 15:00, Manuel Valle Torre successfully defended his PhD thesis “Sequences in Data to Represent Learning Processes: What have you tried until now?” at Delft University of Technology. Hendrik Drachsler was honoured to take part as an independent member of the doctoral committee.

Manuel-defense-slides
Manuel’s thesis starts with a question every programming tutor knows: “What have you tried until now?” Digital learning environments usually assess only the final product, so this question is hard for them to answer. In three phases, Manuel shows how sequences of learner actions can be used to understand learning processes and support them while they happen. He developed ELAT for analysing edX log data, reviewed the field of learning sequence analysis, and studied problem-solving in data science notebooks. He then built JELAI, an open-source platform that combines learning analytics with an AI tutor directly inside Jupyter Notebooks.

With JELAI, Manuel studied how students ask generative AI for help. His findings are timely. Without guidance, many students use AI to get ready-made solutions, and these habits do not correct themselves. Teaching students how to ask better questions changed their behaviour. Small motivational nudges from the AI tutor helped students keep going. One of his most striking results is that novices who delegated their work to the AI reached the same scores as more experienced learners. When AI makes the final product look the same for everyone, the learning process becomes the most important thing to understand.

These findings fit closely with our own research on Highly Informative Learning Analytics (HILA) and Data-enriched Learning Activities (DeLA). We look forward to exploring how JELAI could support a DeLA for programming.

Dr. Valle Torre

Congratulations, Dr. Valle Torre, and many thanks to the promotors Prof. Marcus Specht and Dr. Catharine Oertel. In the words of our congratulations to Manuel: measure what you value, not just what is easy to measure.