State space modeling of time series
State space modeling of time series
Paper: Modeling by shortest data description
Automatica (Journal of IFAC)
| Hi-index | 0.98 |
The problem of deriving a low dimensional model from a given data record arises frequently in fields such as engineering, economics, or biology. In the context of a stochastic realization algorithm, this study considers three variations of model reduction techniques based on singular value decomposition. Applying Monte Carlo methods, the accuracy of the alternatives in approximating impulse responses is investigated for various sample sizes.