Time series AR model parameter estimation with missing observation data

This paper focuses on identification problems of auto-regression (AR) models with missing output observation data. The standard least squares algorithm cannot be applied to the AR models due to the missing output data. To estimate the parameters of the AR models, we employ the polynomial transformation technique to transform the AR models into the auto-regression moving average (ARMA) models, which can be identified from available scarce observation data. Then, we analyze the convergence properties of the algorithm in details and give an example to test and illustrate the algorithm involved.