Random Forest

  • RF
  • Bagged Trees

Best for: Strong baseline, robust tabular modeling Aliases: RF, Bagged Trees

How it works

$$\hat{f}(x)=\frac{1}{B}\sum_{b=1}^{B}T_b(x)$$

Grows $B$ decorrelated decision trees, each trained on a bootstrap sample of the data and allowed to consider only a random subset of features at every split (bagging plus feature subsampling). Prediction averages the trees for regression, $\hat{f}(x)=\frac{1}{B}\sum_b T_b(x)$, or takes a majority vote for classification. Averaging many de-correlated trees lowers variance without raising bias, which is why it is a strong, low-tuning baseline.

When to use

Strong out-of-the-box baseline for tabular data when you want robustness with minimal tuning and preprocessing.

Watch out

Large memory footprint and slower prediction than boosting; cannot extrapolate beyond the training range.

Common fields

Healthcare · credit risk · customer analytics · ecology · operations