How to model and generate forecasts for time series with missing observations?

Question

I am trying to model some historical variables which are discontinuous. I am working with monthly observations so I have 12 observations per year.

However, there are cases in which, for example, I have data for January-2000, February-2000, and then the data jumps to May-2000, and so on.

I am using R to perform the time series and forecasting analysis.

Is there a way to cope with this kind of irregular time series in R?

Alternatively, you may be able to consider it a regularly-spaced time series with missing values. In that case, any state-space model which can be estimated through the Kalman filter algorithm will work automatically with missing values, and the algorithm will also provide non-anticipating estimates for the missing values, and forecasts. — Chris Haug, Mar 23 '17 at 17:53

score 1 · Answer 1 · answered Mar 23 '17 at 16:43

1

answered Mar 23 '17 at 16:43

DaBenski

1 Answers1