liveBook · Manning

XGBoost

As explained in Build AI Drug Discovery Pipelines

XGBoost is a gradient-boosting implementation designed for efficient and high-performing supervised learning.

XGBoost is an efficient gradient-boosting implementation for supervised learning. It builds an ensemble sequentially: each weak learner focuses on errors left by earlier learners, producing stronger predictions on difficult cases. section 6.4.2

Unlike random forests, which train independent models in parallel, boosting trains models in sequence. Gradient boosting uses a loss function to communicate how well the current ensemble fits each training example, guiding the next tree. section 6.4.2

Small-molecule binding prediction

A drug-discovery workflow uses XGBoost to predict which compounds may bind an RNA target. Its ensemble can be inspected through feature-importance views based on gain, coverage, and frequency. section 6.4.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains weak learners, sequential error correction, and loss-guided updates, but stops before implementation details.
Examples
Uses RNA-target compound binding, a drug-discovery scientist analogy, and feature importance measured by gain, coverage, and frequency.
What it assumes
Assumes familiarity with ensemble methods and introduces boosting by contrasting it with random forests.
What you can do after
Evaluate/understand how sequential weak learners reduce boosting bias while preserving ensemble generalization.
How it connects
Connects XGBoost with ensemble methods, random forests, boosting, gradient boosting, weak learners, and loss functions.

Also explained in

Compare how 7 books explain XGBoost →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.