Abstract : In this paper, we propose an interactive data-processing environment for algorithmic training and autoevaluations. Learner will acquire knowledge on algorithmic, but also experiment it. The use of new technologies of information is not the gaol in it self, nevertheless it's could be a rich support on which we can base a new methodological approach in teaching. Our approach is based on the use of the examples and their experimentation. We focus on the activity of the subject to achieve the teaching goals discounted. Several scenarios are possible according to the knowledge and rate of learner. Several types of exercises are designed to sensitize learner, and enable him to develop easily abstract models of data, temporal tasks and their sequences in an algorithm. The environment proposes evaluations for learner and gives him recommendations which change according to the type of errors. The management of the profile of learning is an integral part of this environment.