| Summary: | This paper deals with the application of neural networks in predicting the fuel sales index. Neural networks, through their learning ability, can understand the variability of parameters and from this, infer about their future behavior. Most of the sales forecasts made by ANP (National Agency of Petroleum, Natural Gas and Biofuels) are based on fuel consumption, where in this work this index was disregarded and other indicators that were considered relevant in the prediction process were used. The best network consists of a multilayered perceptron, trained with the backpropagation algorithm, consisting of five neurons in the input and intermediate layers and with only one output node. This presents a relative mean square error of 27% for the expected sales figures. The results generated were satisfactory for the chosen variables.
|
|---|
