Learning analytics to assess students’ behavior with scratch through clickstream

The construction of knowledge through computational practice requires to teachers a substantial amount of time and effort to evaluate programming skills, to understand and to glimpse the evolution of the students and finally to state a quantitative judgment in learning assessment. This suposes a huge problem of time and no adecuate intime feedback to students while practicing programming activities. The field of learning analytics has been a common practice in research since last years due their great possibilities in terms of learning improvement. Such possibilities can be a strong positive contribution in the field of computational practice such as programming. In this work we attempt to use learning analytics to ensure intime and quality feedback through the analysis of students behavior in programming practice. Hence, in order to help teachers in their assessments we propose a solution to categorize and understand students’ behavior in programming activities using business technics such as web clickstream. Clickstream is a technique that consists in the collection and analysis of data generated by users. We applied it in learning programming environments to study students behavior to enhance students learning and programming skills. The results of the work supports this business technique as useful and adequate in programming practice. The main finding showns a first taxonomy of programming behaviors that can easily be used in a classroom. This will help teachers to understand how students behave in their practice and consequently enhance assessment and students’ following-up to avoid examination failures.