I have a csv file with a few columns, each with numerical values. I have an
int value for NA which is different for each column. I also have to rescale
some columns (division by 10, 1000, etc). I have found that converts is
processed first than na_values and that everything is processed as string
and only at the end is there some type inferencing. Is this correct? Any
way of changing the arguments processing order? I can just set the NA
values for post-converted ones, but nonetheless...