Forecasting Data Published at Irregular Time Intervals Using an Extension of Holt's Method

In practice many data series contain observations at irregular times whereas most forecasting methods are restricted to the case of equal time intervals between data points. This paper provides extensions of Single Exponential Smoothing and Holt's Method to the case of irregularly spaced data and shows them to be highly efficient computationally. The new methods are applied to six published series, and their performance is analyzed via four error measures with respect to changes in the smoothing parameters.