What is 'Gradient Boosting'?

What is 'Gradient Boosting'?

Approach: Describe building trees to predict the gradient of the loss function.

A machine learning technique for regression and classification that produces a prediction model in the form of an ensemble of weak prediction models, typically decision trees, built sequentially to minimize loss.

Follow-up: What is the learning rate in gradient boosting?

Key concepts: residuals, weak learners, optimization.