Tous les livres

Professionnel

Applis À propos Coach Connexion Commencer la lecture

Quantitative Finance · Glossaire

Qu'est-ce que « Decision tree » ?

Definition 5.1 Machine Learning for Markets · Chapitre 5 — Trees and Boosting

A decision tree partitions the feature space by a sequence of binary splits xj≤cx_j\le c, chosen greedily to reduce the loss most (for regression, the sum of squared errors), and predicts in each final cell, or leaf, the mean target of its training rows. Its capacity is set by its depth and by the minimum number of rows per leaf.

A tree of depth two on ranked characteristics, as a schematic: each split sends a stock-month left or right, each leaf predicts the mean demeaned return of its training rows. The leaf values and counts are illustrative, not fitted.
Figure 5.1. A tree of depth two on ranked characteristics, as a schematic: each split sends a stock-month left or right, each leaf predicts the mean demeaned return of its training rows. The leaf values and counts are illustrative, not fitted.
Lire dans le chapitre →