Biases in the Simulation and Analysis of Fractal Processes - Université de Montpellier Accéder directement au contenu
Article Dans Une Revue Computational and Mathematical Methods in Medicine Année : 2019

Biases in the Simulation and Analysis of Fractal Processes

Clément Roume
  • Fonction : Auteur
Samar Ezzina
  • Fonction : Auteur
  • PersonId : 1041419
Didier Delignieres
  • Fonction : Auteur
  • PersonId : 1034549


Fractal processes have recently received a growing interest, especially in the domain of rehabilitation. More precisely, the evolution of fractality with aging and disease, suggesting a loss of complexity, has inspired a number of studies that tried, for example, to entrain patients with fractal rhythms. is kind of study requires relevant methods for generating fractal signals and for assessing the fractality of the series produced by participants. In the present work, we engaged a cross validation of three methods of generation and three methods of analysis. We generated exact fractal series with the Davies-Harte (DH) algorithm, the spectral synthesis method (SSM), and the ARFIMA simulation method. e series were analyzed by detrended fluctuation analysis (DFA), power spectral density (PSD) method, and ARFIMA modeling. Results show that some methods of generation present systematic biases: DH presented a strong bias toward white noise in fBm series close to the 1/f boundary and SSM produced series with a larger variability around the expected exponent, as compared with other methods. In contrast, ARFIMA simulations provided quite accurate series, without major bias. Concerning the methods of analysis, DFA tended to systematically underestimate fBm series. In contrast, PSD yielded overestimates for fBm series. With DFA, the variability of estimates tended to increase for fGn series as they approached the 1/f boundary and reached unacceptable levels for fBm series. The highest levels of variability were produced by PSD. Finally, ARFIMA methods generated the best series and provided the most accurate and less variable estimates.
Fichier principal
Vignette du fichier
Delignieres_05.pdf (622.26 Ko) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-02417699 , version 1 (18-12-2019)



Clément Roume, Samar Ezzina, Hubert Blain, Didier Delignieres. Biases in the Simulation and Analysis of Fractal Processes. Computational and Mathematical Methods in Medicine, 2019, 2019, pp.4025305. ⟨10.1155/2019/4025305⟩. ⟨hal-02417699⟩
37 Consultations
89 Téléchargements



Gmail Facebook X LinkedIn More