In this paper, we present a software to support educators collecting and processing data in each step of the experimental learning cycle for cognitive apprenticeship. We do so by developing a system that allows educators to (1) define a set of competences, (2) link keywords to each competences, (3) analyse reflexive papers written by students as evidences of their acquired competences after practical projects and (4) analyse multiple documents to assign students to different clusters to better coach them. The theoretical ground for our artefact can be found in the body of knowledge concerning pedagogy to define a process that would allow combining different teaching approaches and statistical learning to define the software approach to use to gather data and predict future outcomes. We support our claims by illustrating results coming from 15 students of a Team Academy course and 100 students of a university course.