Séries temporais: um estudo de previsão

Abstract

This work presents a prediction study of time series. A time series and a set of data points (information) observed over time. The study uses the ARIMA and Neural Networks and methodologies to make data predictions, seeking to identify among the studied methods the most effective based on the results obtained. Each methodology has its own characteristics in the search for the best result. ARIMA models are a combination of three components which can be interpreted as "filters", which are important parameters for the model in predicting the data. Neural Networks is based on the idea of learning, as well as in the brain, the information received is processed, used to train the network and a model is used that can he used to medict new dat.