Comparación de comportamiento de redes neuronales recurrentes con otros métodos en pronósticos con series de tiempo – caso series hidrológicas.
Associacao Iberica de Sistemas e Tecnologias de Informacao (AISTI)
MetadataShow full item record
One of the fundamental activities of any organization to be able to carry out its planning processes and budgets is the preparation of forecasts. To achieve this, those responsible for the planifcation must apply techniques or models. These models can be qualitative, quantitative or a mixture of both. As for quantitative techniques, the models can be causal or for time series analysis; this depending on how the data is handled, according to how the problem is approached. For both cases statistical techniques are used, but also Artifcial intelligence techniques such as neural networNs or the hybridization of diɣerent techniques. This article focuses on models and techniques for time series forecasts, showing a comparison of the behavior of recurrent neural networks, Elman type, against other statistical techniques and diɣerent neural networN architectures such as Multilayer perceptron - MLP, single layer FeedForward, Radial Basis Function netework - RBF and a modular network composed of MLP networks.
- Artículos 
The following license files are associated with this item: