Técnicas de aprendizado de máquina para previsão de perdas severas em rochas carbonáticas de reservatórios do pré-sal

This work aims to present binary classification models to assist in determining the occurrence of circulation loss phenomenon in the construction of subsea wells in the Santos Basin’s pre-salt. Prior knowledge about the possibility of the phenomenon occurrence, makes it possible to allocate rigs with the appropriate technology for the well's construction. In this context, classification systems based on machine learning can support decision making. In this work, classifiers are proposed based on classic machine learning algorithms (Pedregosa, F. et al. 2011) and the results are presented using the Area Under the ROC Curve (AUC) as the metric.