python statsmodels.tsa.stattools.pacf with masked array?

Question

Is there a general trick to using masked arrays (or arrays containing nan's) with the statsmodels routines? For example pacf and acf?

Answer 1

In general, the functions don't know anything about masked arrays. The Model classes accept a missing keyword for missing value handling. However, I don't think there's anything in acf/pacf. What would you expect these functions to do with missing data? If you file an issue on github with an example and ideally a reference of comparison results from another stats package, I can look into it.

https://github.com/statsmodels/statsmodels/issues