Stone is a company that offers payment solutions for card machines. The company s solutions go through not only card machines, but also through a financial control portal that allows you to safely manage the company s health, in addition to providing credit to your customers. In this market there is great competition and because of this, a well-evaluated metric is the turnover rate, or Churn rate, which shows the rate of consumers that a company lost in a given period and the total revenue involved in this process. Another and distinct point of attention is the team s ability to detect and reject bad credit risks, as the success of an individual credit operation and the sustainability of fintech depend on it. Consequently, in order to separate good and bad payers and identify the turnover rate, there is a need for credit and churn classification, through data science. The dissertation addresses the combination of the decision tree technique, a machine learning algorithm, with optimization for credit and churn risk classification for Stone Pagamentos S.A. The literature lacks studies on the combination of both techniques, which promising tools have been shown, since implicit knowledge can be extracted from the learning algorithms and complemented with the explicit information obtained from the optimization. The decision tree facilitates the interpretation of the results in comparison with other machine learning algorithms that are black box. This technique performs better than techniques like boosting and random forest when applied with mixed integer optimization. This approach can benefit from the theoretical guarantees and state-of-the-art algorithms available in the literature.