assaforon / cir Goto Github PK

Dear @assaforon ,

First of all, thank you for your excellent R package. We have been enjoying using it with with much success for reweighting-based estimators.

It took me some time to debug the following issue:

my_df <- structure(list(x = c(49.0046744288143, 74.6004805025514, 14.0424434314875, 
                              15.0093007827047, 8.73911971302957), y = c(1225.32722130527, 
                                                                         2362.0066919363, 121.935555430898, 4.03728696991483, 32.9374863651358
                              ), wt = c(1, 1, 1, 1, 1)), row.names = 2:6, class = "data.frame")

my_df %>% cirPAVA # fails
my_df[order(my_df$x), ] %>% cirPAVA() # runs

... because (as I understand it) I didn't realize that the inputs needed to be pre-ordered by x.

I therefore write only with the request that this be included (perhaps more prominently) in the documentation, or that an error message can alert me to the relevant issue.

Thank you for considering, and thank you again for the excellent package.

Sincerely,

Amy