Computational Modeling of Multivariable Non-Stationary Time Series in the State Space by the AOKI_VAR Algorithm

In this work we propose an iterative e algorithm for identification in the state space of multivariable non-stationary discrete time series considering and implementing a computational process which we will call AOKI_VAR. The proposed algorithm is based on an algorithm proposed by Masanao Aoki for computational modeling of time series. A modeling example is presented as well as discussions on validation, prediction and modeling of time series.