This data is prepared for CS334 Machine Learning course final project from Emory University. It can also be used and we are very welocome for doing any kind of data analysis and machine learning research by anyone.
Data is scraped from vlr.gg: one of the three major websites (vlr.gg, rib.gg, thespike.gg) that give out past Valorant pro matches data. For actual in-game data you can check out rib.gg which may have advanced data for selling.
Columns are described in abbreviation and here are the explanation:

name: str
team: str  # might be different for the same team
agent: str
rating: float
acs: int  # average combat score
k: int  # kills
d: int  # deaths
a: int  # assists
tkmd: int  # total kills minus deaths
kast: float  # kill, assist, survive, trade %
adr: int  # average damage per round
hs: float  # headshot %
fk: int  # first kills
fd: int  # first deaths
fkmd: int  # first kills minus first deaths
t # attack side
ct # defend side

Data may have missing values either because they were missing in the original source or the gameplay is so biased (score like 13:0) that one of the team or some of the players did not get to do anything.