Autoregressive Tree Models For Time-Series Analysis

Chris Meek, Max Chickering, David Heckerman

Proceedings of the Second International SIAM Conference on Data Mining |

Published by SIAM

View Publication

The analysis and modeling of time-series data is an important area of research for many communities. In this paper, our goal is to identify models for continuous valued time-series data that are useful for data mining in that they (1) can be learned eficiently from data, (2) support accurate predictions, and (3) are easy to interpret. To these ends, we describe an interpretable class of models that we call AutoRegressive Tree models, or ART models, that are a generalization of standard autoregressive (AR) models. We describe learning methods for ART models and compare these methods to those for alternative models. Our experiments, performed on 2,494 time-series data sets from the International Institute of Forecasters, demonstrate that ART models provide superior predictive accuracy. We concentrate on the problem of modeling the evolution of values of a continuous variable over time; that is, we model a univariate time series. The generalization to multivariate time-series analysis is straightforward and is discussed in Section 6.