We start with a problem which is the separation between the conceptual and algorithmic in the teaching of integral Calculus. Thus, to address the problem according to its nature, we rely on the theoretical approach called socioepistemology helped us as well as the theory of the conceptual fields. Based on the above we present a field of social practices organized in the order of three areas: prediction, accumulation and constancy of what is variable. It has been necessary to build the field, to the extent possible, from different levels: the historical genesis, the contemporary genesis and the artificial genesis. In conclusion, we argue about the feasibility of treating social practices as the unit of analysis for the redesign of the school mathematical discourse, in particular the integral Calculus. So, the prediction (immersed in a field of social practices) is by its nature interweaving knowledge without a rigid boundary between concepts without a linear order, and that transcends the domain of mathematics.