Medoid splits for efficient random forests in metric spaces

Institut for Matematiske Fag

Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt

Standard

Medoid splits for efficient random forests in metric spaces. / Bulté, Matthieu; Sørensen, Helle.

I: Computational Statistics and Data Analysis, Bind 198, 107995, 2024.

Publikation: Bidrag til tidsskrift › Tidsskriftartikel › Forskning › fagfællebedømt

Harvard

Bulté, M & Sørensen, H 2024, 'Medoid splits for efficient random forests in metric spaces', Computational Statistics and Data Analysis, bind 198, 107995. https://doi.org/10.1016/j.csda.2024.107995

APA

Bulté, M., & Sørensen, H. (2024). Medoid splits for efficient random forests in metric spaces. Computational Statistics and Data Analysis, 198, [107995]. https://doi.org/10.1016/j.csda.2024.107995

Vancouver

Bulté M, Sørensen H. Medoid splits for efficient random forests in metric spaces. Computational Statistics and Data Analysis. 2024;198. 107995. https://doi.org/10.1016/j.csda.2024.107995

Author

Bulté, Matthieu ; Sørensen, Helle. / Medoid splits for efficient random forests in metric spaces. I: Computational Statistics and Data Analysis. 2024 ; Bind 198.

Bibtex

@article{fcbb0b82c65c4c049d5f16815a48187c,

title = "Medoid splits for efficient random forests in metric spaces",

abstract = "An adaptation of the random forest algorithm for Fr{\'e}chet regression is revisited, addressing the challenge of regression with random objects in metric spaces. To overcome the limitations of previous approaches, a new splitting rule is introduced, substituting the computationally expensive Fr{\'e}chet means with a medoid-based approach. The asymptotic equivalence of this method to Fr{\'e}chet mean-based procedures is demonstrated, along with the consistency of the associated regression estimator. This approach provides a sound theoretical framework and a more efficient computational solution to Fr{\'e}chet regression, broadening its application to non-standard data types and complex use cases.",

keywords = "Least squares regression, Medoid, Metric spaces, Random forest, Random objects",

author = "Matthieu Bult{\'e} and Helle S{\o}rensen",

note = "Publisher Copyright: {\textcopyright} 2024 The Authors",

year = "2024",

doi = "10.1016/j.csda.2024.107995",

language = "English",

volume = "198",

journal = "Computational Statistics and Data Analysis",

issn = "0167-9473",

publisher = "Elsevier",

}

RIS

TY - JOUR

T1 - Medoid splits for efficient random forests in metric spaces

AU - Bulté, Matthieu

AU - Sørensen, Helle

PY - 2024

Y1 - 2024

N2 - An adaptation of the random forest algorithm for Fréchet regression is revisited, addressing the challenge of regression with random objects in metric spaces. To overcome the limitations of previous approaches, a new splitting rule is introduced, substituting the computationally expensive Fréchet means with a medoid-based approach. The asymptotic equivalence of this method to Fréchet mean-based procedures is demonstrated, along with the consistency of the associated regression estimator. This approach provides a sound theoretical framework and a more efficient computational solution to Fréchet regression, broadening its application to non-standard data types and complex use cases.

AB - An adaptation of the random forest algorithm for Fréchet regression is revisited, addressing the challenge of regression with random objects in metric spaces. To overcome the limitations of previous approaches, a new splitting rule is introduced, substituting the computationally expensive Fréchet means with a medoid-based approach. The asymptotic equivalence of this method to Fréchet mean-based procedures is demonstrated, along with the consistency of the associated regression estimator. This approach provides a sound theoretical framework and a more efficient computational solution to Fréchet regression, broadening its application to non-standard data types and complex use cases.

KW - Least squares regression

KW - Medoid

KW - Metric spaces

KW - Random forest

KW - Random objects

U2 - 10.1016/j.csda.2024.107995

DO - 10.1016/j.csda.2024.107995

M3 - Journal article

AN - SCOPUS:85196724062

VL - 198

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

M1 - 107995

ER -

ID: 396942515