A nonparametric model makes fewer assumptions about the data and adapts its complexity to the data, rather than using a fixed functional form and parameter count. section 3.3.1
Decision trees are a practical example: they represent predictions as feature-based questions, branching until reaching a leaf that supplies the final answer. section 5.2
Classifying cytochrome P450 inhibition
A tree-based model can classify compounds by repeatedly splitting on molecular features, narrowing the possibilities until a leaf predicts inhibition. section 5.2
