Identifying problem solving strategies of programming novices in a serious game

First steps in programming often take place in a self-regulated learning process, online and without supervision or assistance of a teacher. When problems occur, novices depend on automated feedback from the programming learning environment or sample solutions, which do not necessarily fit the approach of the learners. Our goal is to identify and classify the problem-solving strategies of programming novices. In the long term, these insights might help with adaptive feedback fitting to the current trial of the individual player and to learn from strategies of successful learners. Using self-created visualization tools we found complex indicators for small-scale strategy patterns. These patterns, along with interaction data from the user interface, allow clustering the learners by their behaviour. Those insights should form a future basis for automatic recognition of strategies.