Partition Trees: Conditional Density Estimation over General Outcome Spaces

Felipe Angelim, Alessandro Leite · 2026

Read Partition Trees: Conditional Density Estimation over General Outcome Spaces by Felipe Angelim, Alessandro Leite (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-04042.txt.

Description

Estimating a full chance distribution, not just a likely value, is Felipe Angelim and 1 co-author's goal. Their 2026 paper unveils Partition Tree, which chops the input space into self-tuning regions and fits a constant density on each. It handles continuous and categorical outcomes under one roof.

Trees are grown to cut conditional negative log-likelihood, with no assumed distribution shape. A bagged ensemble version, Partition Forest, blends many such trees. Experiments show better probabilistic forecasts than classic regression trees and results on par with leading probabilistic methods. Honest error bars matter wherever forecasts feed choices, from health care to finance.

Catalog record

CREATORFelipe Angelim, Alessandro Leite
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE110 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-04042

Browse the full library · About the Tower of Babel