Greetings,

There are many ways to impute your data due to missing values. Take a look at the table on page 3 of this article (https://mran.microsoft.com/snapshot/2017-12-11/web/packages/imputeTS/vignettes/imputeTS-Time-Series-Missing-Value-Imputation-in-R.pdf) to get a feel for some of the common approaches. Personally, I use moving average but that is in the field of EEG motor imagery classification so I could not state with certainty if that would work best given your specific problem. If your application is machine learning, always start with removal, gauge how it affects performance, then move to more involved methods as needed.

Best,

Tyler J Grear

Interpolate. Least-squares fit a low-order polynomial to the available data that surround the missing datum. Then evaluate the fitted polynomial at the missing datum. You could use a weighted least-squares fit if you think some of that data are more relevant than others.

As per attached.