Base stat total alone is a bad metric, because stat distribution is equally as important.
If the stats are distributed heavily both on attack and special attack, it's usually bad because you generally want specialist attackers and these stats could be better somewhere else like speed.
Absolutely! In general I would expect a better model to incorporate a lot of weighed terms in the objective to choose less "extreme" solutions, but here I was mostly interested in illustrating the method.
Comments
Base stat total alone is a bad metric, because stat distribution is equally as important.
If the stats are distributed heavily both on attack and special attack, it's usually bad because you generally want specialist attackers and these stats could be better somewhere else like speed.
Absolutely! In general I would expect a better model to incorporate a lot of weighed terms in the objective to choose less "extreme" solutions, but here I was mostly interested in illustrating the method.
It was very impressive at that, congratulations.
You can go for smogon tiers as a proxy for pokemon strength.