Study area of process mining exhibits great potential in explaining the design performance and behavior pattern in architecture, engineering and construction industries based on event data. However, research into creativity-related behavior pattern in sustainable building design through process mining remains blank. With the global climate change and energy crisis, creativity-driven problem-solving design process is becoming crucial. Based on >41-million lines of event data from 115 participants from a green building design competition, this study explored four dimensions on creative design behavioral pattern: 1) frequently used commands; 2) frequently used objects; 3) duration of used commands and 4) experience with the tool. Secondly, this study establishes and trains artificial neural network models according to the four dimensions to evaluate level of creativity of design works. Training results prove the effectiveness of the discoveries and realize a preliminary screening on creativity for large amount of data in sustainable building design.