Learn AI · Level 2 · How AI works · step 6 of 18 · word 2 of 2
What is gradient boosting (XGBoost)?
Gradient boosting builds a strong model from many small decision trees, each one trained to fix the mistakes of the ones before it. XGBoost is the best-known version.
In plain English
Imagine a team of weak students taking turns on the same test. Each one studies only the questions the last got wrong. Together they score far higher than any one of them. Gradient boosting does this with small decision trees, adding them one at a time.
An example
XGBoost, a free software library, dominated data science competitions. Its creators counted that 17 of the 29 winning solutions published on Kaggle’s blog in 2015 used it.
Watch out
For predictions from tables of data, such as fraud, prices or which customers will leave, gradient boosting often beats neural networks and is much cheaper to run. It is a big part of predictive AI in business. It is not used for language or images, where deep learning wins.
Learn AI path · Level 2 · How AI works · step 6 of 18
Done? Next stepTensorFlow Playground ↗See all of level 2