Inconsistent Handling of NAs using roll_median()

[csatzky]

Hi there,

Thank you for developing this extremely useful package.

When using roll_median() in the presence of NA values the output is inconsistent with all other functions, e.g. roll_mean(), roll_sum(), etc.

According to the documentation, all of the parameters are the same by default across roll_...() functions.

# a vector with NAs
vec <- c(1:5, rep(NA,3), 6:10, NA)

# expected output of computed & NA values
roll_mean(vec, 5)
roll_var(vec, 5)

# output using roll_median()
roll_median(vec, 5)

Also for this function, there is no difference whether na.rm = c(TRUE, FALSE).

Thank you.

Cheers,
Christian

[kevinushey]

Thanks for the bug report! I believe this should now be resolved in the development version of RcppRoll. You can give it a spin with:

remotes::install_github("kevinushey/RcppRoll")

[csatzky]

It seems to work well now! Thank you very much for the fix!

Cheers,
C