Measuring the wrong things gives the wrong results though.
Miners by block reward. This should be count of all individual miners, including those in pools. Reason — the more miners there are (regardless if they get a reward or not), the more decentalized the subcomponent is.
Developers that commit. This should be developers that contribute. Reason — often there are only one or two developers with merge rights but hundreds of developers contributing code via pull requests.
Clients count. Someone else already pointed out why this is wrong.
Distribution / users by address count. This is misleading as one user can, and often does, have many addresses and wallets.