I would like to know if anyone knows how to apply the arima results to calculate missing values in the observation period. I am looking for something similar to forecast.arima(), that forecasts values based on arima results.

Does anyone has experiences in using data from other sources that are complete but with less accuracy, to replace the missing values? Is it feasible to decompose both datasets using stl and recalculate the new values using the observation data?

I am using daily climatic observations for a large number of years, although some of the missing data is between observations, I have also large periods of time without any observations.


Box and Jenkins all the way back in the first edition of their book (1970) proposed backcasting as a way to "fill in" missing values in the middle of a time series. The link I gave is for the most recent 2008 edition.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.