Simplifying Random Forests' Probabilistic Forecasts.
Since their introduction by Breiman, Random Forests (RFs) have proven to be useful for both classification and regression tasks. The RF prediction of a previously unseen observation can be represented as a weighted sum of all training sample observations. This nearest-neighbor-type representation is...
| Publicado en: | American Statistician Vol. 80; no. 1; pp. 135 - 146 |
|---|---|
| Autores principales: | , |
| Formato: | Artículo |
| Publicado: |
Taylor & Francis Ltd
Feb2026
|
| Materias: | |
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=191630189&site=ehost-live header: @attributes: shortDbName: ssf uiTerm: 191630189 longDbName: Social Sciences Full Text (H.W. Wilson) uiTag: AN controlInfo: bkinfo: jinfo: jid: 00031305 STT jtl: American Statistician issn: 00031305 maglogo: Y pubinfo: dt: Feb2026 vid: 80 iid: 1 pid: 377 pub: Taylor & Francis Ltd artinfo: ui: 191630189 10.1080/00031305.2025.2552284 ppf: 135 ppct: 11 formats: tig: atl: Simplifying Random Forests' Probabilistic Forecasts. aug: au: Koster, Nils Krüger, Fabian affil: Karlsruhe Institute of Technology, Karlsruhe, Germany Broad Institute of MIT and Harvard, Cambridge, MA su: Forecasting Random forest algorithms Machine learning Distribution (Probability theory) Statistical models K-nearest neighbor classification Parsimonious models sug: subj: Forecasting Random forest algorithms Machine learning Distribution (Probability theory) Statistical models K-nearest neighbor classification Parsimonious models keyword: Forecast distribution Random forests Simplicity Forecast distribution Random forests Simplicity ab: Since their introduction by Breiman, Random Forests (RFs) have proven to be useful for both classification and regression tasks. The RF prediction of a previously unseen observation can be represented as a weighted sum of all training sample observations. This nearest-neighbor-type representation is useful, among other things, for constructing forecast distributions (as in Meinshausen's Quantile Regression Forests). In this article, we consider simplifying RF-based forecast distributions by sparsifying them. That is, we focus on a small subset of k nearest neighbors while setting the remaining weights to zero. This simplification, which we refer to as "Topk", greatly improves the interpretability of RF predictions. It can be applied to any forecasting task without re-training existing RF models. In empirical experiments, we document that the simplified predictions can be similar to or exceed the original ones in terms of forecasting performance. We explore the statistical sources of this finding via a stylized analytical model of RFs. The model suggests that simplification is particularly promising if the unknown true forecast distribution contains many small weights that are estimated imprecisely. pubtype: Academic Journal doctype: Article src: R language: English refInfo: copyright: @attributes: flag: N holdings: @attributes: islocal: N |
|---|