As I said earlier, it really is a great job and I really enjoy these kinds of datasets visualization :)
It's just in these process of combining as you said that mistakes happen ;)
I'm still in first year of my data engineering bachelor so I can't say anything for sure, but usually these kind of ratios, when you sum up elements, you either recalculate the ratio with the whole data or you approximate it with a geometrical mean, and most importantly the result is always between the highest (here 1.52) and lowest (0.24) percentage. I need to go and won't insist any more, but please check it with an engineer of the team ^^
And, as said, it really is a very pleasant and clean, readable data, thank you again for that ^^ (edited)