[en] To provide quality feedback, the risk of two types of errors must be minimized. The type 1 error, which is to provide unnecessary information to the recipient and the type 2 error, which is not to provide information that could be useful to the recipient. When Learning Analytics is used in academic contexts, it is important to be sure that the delivered information is strongly related to critical aspects of the learning experience on which the student would have to apply some level of reflexivity. This is why we introduce a new model meant to tackle this issue from a strongly asserted pedagogical viewpoint. This model is concerned by teaching and by learning at the same time. A specific feature of it is therefore to help establishing the relevance of learning analytics for both sides of the instruction process. In addition, our modeling integrates a facet of the pedagogy too often overlooked in the reflections on the Learning Analytics, it is about the human being. This complex dimension is part of our reasoning, meeting the wish of George Siemens, when he wanted the Learning Analytics to become "more human" (2015).
