Contents I. Introduction
ICT literacy has been regarded as one of the essential skills in modem life. It has been raising broad attention from various fields including the workplace and education [1]. In order to understand and make effective use of computational mechanisms and information technology, it is crucial to learn programming—many studies on various topics have been conducted, such as teaching and learning methods [2]. In particular, some studies that aim to enhance educational outcomes focus on the coding process of programming novices to help instructors analyze their coding behaviors and progress. The typical approach for this end is a content analysis of questions raised by students in class and quantitative analysis of syntax and compile errors. However, such overt questions and error logs do not help instructors find and deal with students who are actually in trouble but hesitate to ask questions or have some technical troubles that do not produce explicit errors. If instructors can find out such students who are stuck while they are coding, they could provide appropriate instructions or formative feedback. Besides, in recent programming courses, it is common for students to create programs to draw some shapes or web pages freely. The conventional data such as compile log and the time to finish assignments does not help evaluate their progress on this type of assignments. Instructors need to analyze data at a higher granularity level to identify students who are in trouble.
