4 ms·
The main point of this article is really to explain how gradient boosting works and why. The math is really there to show what the algorithm looks like in its g
by parrt 8y ago
The main point of this article is really to explain how gradient boosting works and why. The math is really there to show what the algorithm looks like in its general form. The Discussion of parameters was really just a bit of motivation. Think of this as a good explanation of why it is performing gradient descent in function space. That tends to be very hard to explain.
- nonbel 8y agoI just disagree with the claim that understanding the mathematical machinery is necessary for tuning hyperparameters. I doubt it is even helpful.