Reference · Glossary
Random forest
Last updated
A **random forest** averages many decision trees trained on bootstrap samples and random feature subsets. Usually more accurate and stable than one tree.
#When to use
Structured/tabular prediction when you want strong baselines with less tuning than deep nets.
#When not to
Ultra-low-latency tiny models, or problems that need smooth probabilistic calibration out of the box without extra work.
#Practical tips
- Start with out-of-bag error estimates
- Cap tree depth to control size
- Still hold out a true test set for final numbers
#Learn next
- Lesson: `random-forests`
- Related: decision tree