An Introduction to Hybrid Method for Interval-Valued Time Series

Interval-valued data appear quite naturally in many situations in which the values of each observation are defined by two points, a lower and an upper limits. This paper presents hybrid method to forecast the interval-valued time series. Hybrid model is a nonlinear method that is the additive combination of autoregressive integrated moving average and artificial neural networks models. The performance of the estimation procedure and prediction methods is evaluated via a simulation study. Then, our methodology is illustrated on a practical data set.