Abstract: This work contributes with a new approach for tuning hyperparameters of machine learning models, based on sequences of optimization studies based on an initial range of hyperparameters.