Aplicação de técnicas de aprendizado de máquina no setor de ressarcimento de uma empresa securitária

Abstract

Faced with variables that affect a company's process, sometimes, solving problems using the Cartesian method becomes very costly. In this sense, modeling and simulation, especially machine learning algorithms, can contribute to the improvement of manufacturing processes. Thus, the objective of this work was to test machine learning algorithms for legal collection proceedings of an insurance company, trying to predict the outcome of the action. The method used in this work was Modeling and Simulation, where it was possible to use the algorithms of Neural Networks, Logistic Regression and Decision Trees. The results achieved reached almost 80% prediction, indicating an adequate choice of predictors. With this research, it was possible to demonstrate the possibility of prediction through algorithms and thus, have one more tool for decision making during the achievement of the insurer's processes.