Claim Status Prediction for OSIPTEL Using Neural

In recent years, the demand for complaints from users towards mobile operators has increased notably according to OSIPTEL indicators. Only for the year 2017, the operating companies registered 2 million 728 thousand 430 claims, of which 233 thousand 342 claims passed to second instance (Tribunal Administrativo de Solucion de Reclamo - TRASU of OSIPTEL), putting OSIPTEL in big trouble. This research seeks to solve this problem by implementing a system to predict the meaning of the claim, making use of neural networks. It was decided to implement a Multilayer Perceptron, as a learning algorithm, Backpropagation of the error was chosen. As for the architecture of the Perceptron, several were tested, where the changing factor was the neurons in the hidden layer. The results show that the system has 86.969% accuracy.