Seasonal time series forecasting with a state-dependent model

This paper predicts the seasonal time series using a state-dependent autoregressive model. To improve the forecasting performance of the model, this paper considers automatic selection of the number of network nodes and the proper input variables, and simultaneously optimizing the parameters of the model. The nodes and inputs of the model is represented in one chromosome and evolved by genetic algorithm. The performance of the presented approach is evaluated by predicting a seasonal time series. Comparison results show the effectiveness of the proposed method.