A framework is a set of concepts used to solve a problem in a specific domain and this paper, one is proposed through the use of the data collected from Brazilian res judicata processes related to cyber crime and machine learning. Using machine learning methodology, specifically knowledge discovery in database, respecting each of its phases and through the execution of the XGBoost algorithm and SHAP, the results produced tend to help experts to discover patterns of the application of pecuniary fines by the court in the face of a set of applied laws and make predictions.