Detection of long-range dependence and estimation of fractal exponents through ARFIMA modelling.

  title={Detection of long-range dependence and estimation of fractal exponents through ARFIMA modelling.},
  author={Kjerstin Torre and Didier Deligni{\`e}res and Lo{\"i}c Lemoine},
  journal={The British journal of mathematical and statistical psychology},
  volume={60 Pt 1},
We evaluate the performance of autoregressive, fractionally integrated, moving average (ARFIMA) modelling for detecting long-range dependence and estimating fractal exponents. More specifically, we test the procedure proposed by Wagenmakers, Farrell, and Ratcliff, and compare the results obtained with the Akaike information criterion (AIC) and the Bayes information criterion (BIC). The present studies show that ARFIMA modelling is able to adequately detect long-range dependence in simulated… 

Figures from this paper

Estimating long-range dependence in time series: An evaluation of estimators implemented in R
The findings—confined to ARFIMA (0, d, 0) models and fractional Gaussian noise—identify the best estimators for persistent and antipersistent series.
Tenets and Methods of Fractal Analysis (1/f Noise)
This chapter deals with methodical challenges confronting researchers of the fractal phenomenon known as pink or 1/f noise, and compares and evaluates different approaches to estimating fractal parameters from observed data and outlines advantages, disadvantages, and constraints of some popular estimators.
Biases in the Simulation and Analysis of Fractal Processes
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.
Multifractality Versus (Mono-) Fractality as Evidence of Nonlinear Interactions Across Timescales: Disentangling the Belief in Nonlinearity From the Diagnosis of Nonlinearity in Empirical Data
ABSTRACT This article addresses the still popular but incorrect idea that monofractal (sometimes called “fractal” for short) structure might be a definitive signature of nonlinearity and, as a
Relative Roughness: An Index for Testing the Suitability of the Monofractal Model
An easily computable index, the relative roughness (RR), defined as the ratio between local and global variances, that allows to test for the applicability of fractal analyses, and shows that RR is confined within a limited range for long-range correlated series.
Fractional Brownian Motions and their multifractal analysis applied to Parana river flow
A number of different analysis techniques have been used to analyze long-term time series data from different rivers, starting with the determination of the Hurst coefficient. We summarize the
Testing for the presence of 1/f noise in continuation tapping data.
  • Loïc Lemoine, K. Torre, D. Didier
  • Environmental Science
    Canadian journal of experimental psychology = Revue canadienne de psychologie experimentale
  • 2006
The ARMA/ARFIMA modeling procedures proposed by Wagenmakers, Farrell, and Ratcliff (2005) are applied to test for the presence of long-range dependencies in continuation tapping data and offer strong support for the hypothesis that fluctuations in tapping series are fractal in nature.
Complexity, coordination, and health: avoiding pitfalls and erroneous interpretations in fractal analyses.
It is shown that an increase in serial correlations cannot be directly interpreted as an increaseIn long-range correlations, and the strength of correlations, as measured by classical methods, cannot be considered as predictive of the genuine presence of long- range correlations.


On estimating the intensity of long-range dependence in finite and infinite variance time series
A detailed comparison of eight estimators for long-range dependence, using simulated FARIMA(p; d; q) time series with diierent nite and innnite variance innovations, finds that the estimators tend to perform less well when p and q are not zero, that is, when there is additional short- range dependence structure.
Integrated approach to the assessment of long range correlation in time series data
  • Rangarajan, Ding
  • Computer Science
    Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics
  • 2000
An integrated approach is advocated which requires the use of both spectral method and random walk analysis, and the theoretical foundation of this approach is provided and the main ideas using examples are illustrated using examples.
Physiological time series: distinguishing fractal noises from motions
A new method using signal summation conversion (SSC) greatly improves the classification and the reliability of Ĥ, the estimates of H, for the times series, and suggests that the flow signal is the summation of a set of local velocities from neighboring vessels that are negatively correlated, as if induced by local resistance fluctuations.
Inference and Forecasting for Fractional Autoregressive Integrated Moving Average Models, with an application to US and UK inflation
We discuss computational aspects of likelihood-based specification, estimation,inference, and forecasting of possibly nonstationary series with long memory. We use the \\ARFIMA$(p,d,q)$ model with
Finite sample criteria for autoregressive order selection
The special finite sample information criterion and combined information criterion are necessary because of the increase of the variance of the residual energy for higher model orders that has not been accounted for in other criteria.
On the accuracy of spectrum analysis of red noise processes using maximum entropy and periodogram methods: Simulation studies and application to geophysical data
Power spectra, estimated by the maximum entropy method and by a fast Fourier transform based periodogram method, are compared using simulated time series. The times series are computer generated by
Fractal modeling of human isochronous serial interval production
Abstract.The Hurst exponent (H) was estimated for series of 256 time intervals produced by human participants, collected in 5 sessions performed on different days. Each series was obtained during the