The automated analysis of programming code developed during Computational Thinking
development activities has demonstrated a potential to identify employed strategies for
problem solving. However, the evolution of code produced by students and its
relationships with Computational Thinking skill acquisition has not been explored in full
detail. In this study we used selforganizing maps to analyze games produced by
students in a Game Building Workshop offered in 2013 and 2017. As a result of the
training process, clusters in the produced map were consistent with employed strategies
that were previously observed in classroom activities