Hierarchical Mixtures of Autoregressive Models for Time-Series Modeling
- First Online:
- Cite this paper as:
- Vidal C., Suárez A. (2003) Hierarchical Mixtures of Autoregressive Models for Time-Series Modeling. In: Kaynak O., Alpaydin E., Oja E., Xu L. (eds) Artificial Neural Networks and Neural Information Processing — ICANN/ICONIP 2003. Lecture Notes in Computer Science, vol 2714. Springer, Berlin, Heidelberg
A hierarchical mixture of autoregressive (AR) models is proposed for the analysis of nonlinear time-series. The model is a decision tree with soft sigmoidal splits at the inner nodes and linear autoregressive models at the leaves. The global prediction of the mixture is a weighted average of the partial predictions from each of the AR models. The weights in this average are computed by the application of the hierarchy of soft splits at the inner nodes of the tree on the input, which consists in the vector of the delayed values of the time series. The weights can be interpreted as a priori probabilities that an example is generated by the AR model at that leaf. As an illustration of the flexibility and robustness of the models generated by these mixtures, an application to the analysis of a financial time-series is presented.
Unable to display preview. Download preview PDF.