If I set it to be 0.5, does it do column sampling with exact half of the # columns or do sampling with a random # within range (0, half of # columns)?
Thanks for help!
If I set it to be 0.5, does it do column sampling with exact half of the # columns or do sampling with a random # within range (0, half of # columns)?
Thanks for help!
It takes a randomly selected half of the columns at each split/node.
Thanks for help!