There is no single number that says how good someone is at a team game, and the four in common use disagree with each other often enough to be worth separating.
Kill/death ratio
Kills divided by deaths. Simple, old, and it rewards staying alive at least as much as it rewards killing. A player with 20 kills and 20 deaths and a player with 5 kills and 5 deaths have the same KD and have had completely different games.
The incentive it creates is caution, and caution is sometimes exactly wrong. Trading your life to take an objective lowers your KD and wins the match. A player optimising KD will not make that trade, which is why it is the ratio most often criticised in objective modes. The KD calculator handles the arithmetic, including the edge case of zero deaths, where the ratio is undefined and usually reported as the kill count.
KDA
Kills plus assists, divided by deaths. The addition is the whole point: it counts contribution to a kill rather than the last hit, which in most team games is a far better description of what someone did.
It has its own distortion. Assists are cheap in games that award them generously, and a support player can post an enormous KDA while a carry with the same impact posts a modest one. It measures participation, which correlates with contribution without being the same thing.
Win rate
Wins divided by games. The only one of the four that measures the thing you are actually trying to do, and the only one that cannot be farmed by playing selfishly.
Its weakness is noise. Over twenty games, win rate tells you almost nothing, because a coin flipped twenty times regularly comes up 13 to 7. It takes hundreds of games before a win rate separates a 52 per cent player from a 55 per cent player, and most people draw conclusions long before that. The winning percentage calculator handles the count, and the honest reading of a small sample is that it is a small sample.
There is also a mathematical trap in chasing a target. Climbing from 48 per cent to 50 over 200 games does not require winning two more games. If you have played 200 games at 48 per cent, you have 96 wins, and reaching 50 per cent needs eight straight wins to get to 104 out of 208. The further you are from the target and the more games you have played, the more the denominator works against you.
Score per minute
Whatever the game counts, divided by time played. It is the only rate among the four, and that is its advantage: it does not reward sitting in a corner, because time keeps running whether you contribute or not.
It is also the easiest to game, since it measures whatever the scoring system happens to reward rather than what wins.
Which to use
Use win rate to judge whether you are improving, because it is the only one aligned with the goal, and judge it over hundreds of games rather than tens. Use KDA to understand your role in a match. Use KD only in modes where staying alive genuinely is the objective.
And notice which one you check after a loss, because that is the one actually shaping how you play. For a rating system that tries to solve this properly, see the piece on Elo.