In this paper the Fuzzy K-expectiles clustering model is proposed. The model takes into account the asymmetry inherent in the data distribution, extending its applicability to a broader spectrum of data than the Fuzzy K-means. To achieve this, the Fuzzy K-expectiles clustering model introduces the cluster centroid expectile, and assigns data points based on expectile distances. An adaptive asymmetry parameter is specified for each variable and for each cluster The performance of the adaptive Fuzzy K-expectiles model is compared with other clustering models suggested in the literature. To show the performances of the proposed model three simulation studies and three applications to real datasets are presented.

Fuzzy K-expectiles clustering / D'Urso, Pierpaolo; De Giovanni, Livia; Federico, Lorenzo; Vitale, Vincenzina. - In: STATISTICS AND COMPUTING. - ISSN 1573-1375. - 35:48(2025), pp. 35-48. [10.1007/s11222-025-10566-1]

Fuzzy K-expectiles clustering

Pierpaolo D’Urso;Livia De Giovanni
;
Lorenzo Federico;Vincenzina Vitale
2025

Abstract

In this paper the Fuzzy K-expectiles clustering model is proposed. The model takes into account the asymmetry inherent in the data distribution, extending its applicability to a broader spectrum of data than the Fuzzy K-means. To achieve this, the Fuzzy K-expectiles clustering model introduces the cluster centroid expectile, and assigns data points based on expectile distances. An adaptive asymmetry parameter is specified for each variable and for each cluster The performance of the adaptive Fuzzy K-expectiles model is compared with other clustering models suggested in the literature. To show the performances of the proposed model three simulation studies and three applications to real datasets are presented.
2025
Asymmetric quadratic loss, Fuzzy clustering, Expectiles
Fuzzy K-expectiles clustering / D'Urso, Pierpaolo; De Giovanni, Livia; Federico, Lorenzo; Vitale, Vincenzina. - In: STATISTICS AND COMPUTING. - ISSN 1573-1375. - 35:48(2025), pp. 35-48. [10.1007/s11222-025-10566-1]
File in questo prodotto:
File Dimensione Formato  
s11222-025-10566-1.pdf

Open Access

Tipologia: Versione dell'editore
Licenza: Creative commons
Dimensione 928.83 kB
Formato Adobe PDF
928.83 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11385/247558
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
  • OpenAlex ND
social impact