Random Forest is an ensemble learning algorithm that combines many decision trees to produce a more stable and accurate predictor. Given training data {(xᵢ,yᵢ)}₁ⁿ, each tree is trained on a bootstrap sample of the observations, while considering a random subset of features at every split. If T₁(x),…,T_B(x) are the resulting trees, the forest prediction for regression is f̂(x) = 1/B ∑ᵦ Tᵦ(x), while classification typically uses majority voting. The key statistical idea is variance reduction: individual decision trees can have high variance, but averaging many sufficiently diverse trees substantially improves generalization. Random forests are widely used in statistical machine learning because they can model complex nonlinear relationships and high-order interactions without requiring a parametric specification. They are applied to classification, regression, feature selection, anomaly detection and risk prediction. Their built-in estimates of variable importance can help identify influential predictors, while out-of-bag observations provide a convenient form of internal validation. Random forests are particularly valuable in high-dimensional problems where linear models may be too restrictive and where interpretability of a single decision tree is traded for the predictive robustness of an ensemble.