Use a column name as a function argument to then modify that column

[lusty shadow]

That's because you're not actually getting attribute.

You could manipulate expressions with macros.

But you need to interpolate them.

But in this case, what you need is the getattr...

Maybe try...
replace!(getfield(df, col_name), missing=>99)?

Though getfield isn't used quite often.

In that case... can you turn the string into the symbol?

Or...

another way... there might be a way to index the dataframe using name.

df[col_names]?

df[!, col] -> the vector contained in column col returned without copying; the same as df.col if col is a valid identifier.

The docs say this way.

Oh, you can index with an arbitrary type!

And since "!" is a function!

Julia doesn't fail to impress me.