Hello members!

For a term paper, I wanted to check whether a lin-log model or a cubic function is better for a prediction. We were taught Stata at university, but unfortunately this is the first time I am trying to use it practically. Anyway, one variable is GDP per capita. This can be logarithmized via "gen logGDPTotal = log(GDPTotal)". But there is apparently also for scatter the option qfit, which produces the same. So generation is redundant?
Unfortunately, when I use the following command, no useful result comes out: twoway (scatter GDPTotal FIW19) (lfit GDPTotal FIW19) (qfit GDPTotal FIW19)
I would like to generate a similiar graphic.