Summary
    This study concerns the phases of creating, and implementing the neural networks designated to make a forecast sales of goods. This forecast will be done using the computer program "Qnet" Vesta Services Inc.
    The main problem in this study is pointing stages concerning the choice of network topology, testing it and implementation. The end task is comparing gained results with real values and literature data, as well as drawing appropriate from this compression. These conclusions will be helpful in answering if in a concrete situation a neural network is able to execute an established task. But in the case when a neural network is unable to execute properly the task conclusions enable to establish causes that caused it.