Abstract:In practice, several time series exhibit long-range dependence or persistence in their observations, leading to the development of a number of estimation and prediction methodologies to account for the slowly decaying autocorrelations. The autoregressive fractionally integrated moving average (ARFIMA) process is one of the best-known classes of long-memory models. In the package afmtools for R, we have implemented some of these statistical tools for analyzing ARFIMA models. In particular, this package contains functions for parameter estimation, exact autocovariance calculation, predictive ability testing, and impulse response function, amongst others. Finally, the implemented methods are illustrated with applications to real-life time series.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the modeling and prediction of long - range dependence or persistence in time - series analysis. Specifically, the article focuses on how to process and analyze observational data with slowly decaying autocorrelation through the Autoregressive Fractionally Integrated Moving Average model (ARFIMA model). ### Main problem description 1. **Time - series with long - term dependence**: - In practice, many time - series exhibit long - term dependence, that is, there are persistent correlations among their observed values. This characteristic makes it difficult for traditional short - memory models (such as ARMA models) to accurately describe the behavior of these sequences. - The ARFIMA model is a model capable of capturing long - term dependence. Its main feature is the introduction of the fractional integration term \((1 - B)^{-d}\), where \(d\) is a real - valued parameter representing the degree of fractional differencing. 2. **Deficiencies of existing tools**: - Existing statistical software packages have some limitations when dealing with the ARFIMA model, such as the lack of methods for accurately calculating the autocovariance function (ACVF), estimating parameter variances, testing prediction performance, and calculating the impulse response function. - The paper aims to develop and improve an R package named `afmtools` to provide more comprehensive and accurate tools for analyzing the ARFIMA model. ### Specific objectives - **Achieve accurate autocovariance function calculation**: Use the Sowell algorithm to calculate the accurate autocovariance function of the ARFIMA model. - **Parameter estimation**: Conduct parameter estimation through the Whittle method and compare it with other methods (such as the Hessian method). - **Prediction ability test**: Evaluate the prediction ability of different models, especially by comparing them through the root mean square error (RMSE) and the Giacomini - White (GW) test. - **Impulse response function**: Calculate and analyze the impulse response function of the ARFIMA model to assess the impact of shocks on the time - series. - **Practical application**: Demonstrate the functions and performance of the `afmtools` package through practical cases of meteorological data (such as tree - ring width). ### Summary of mathematical formulas - **Definition of the ARFIMA model**: \[ \Phi(B)y_t=\Theta(B)(1 - B)^{-d}\varepsilon_t \] where \(\Phi(B)\) and \(\Theta(B)\) are the autoregressive and moving average operators respectively, \(B\) is the backshift operator, and \((1 - B)^{-d}\) is the fractional differencing operator. - **Autocovariance function**: \[ \gamma(h)=\frac{1}{2\pi}\int_{-\pi}^{\pi}f(\lambda)e^{-i\lambda h}d\lambda \] where \(f(\lambda)\) is the spectral density function. - **Whittle likelihood function**: \[ L(\Omega)\approx-\frac{1}{4\pi}\left[\int_{-\pi}^{\pi}\log[2\pi f(\lambda)]d\lambda+\int_{-\pi}^{\pi}\frac{I(\lambda)}{f(\lambda)}d\lambda\right] \] where \(I(\lambda)\) is the periodogram and \(\Omega\) is the set of model parameters. Through these methods and tools, the paper aims to provide more effective solutions for studying time - series with long - term dependence.

Statistical Analysis of Autoregressive Fractionally Integrated Moving Average Models

On Long Memory Origins and Forecast Horizons

Efficient Bayesian inference for long memory processes

A long-memory integer-valued time series model, INARFIMA, for financial application

Statistical inference for ARTFIMA time series with stable innovations

A Vector Autoregressive Moving Average Model for Interval-Valued Time Series Data

Navigating the Financial Landscape: The Power and Limitations of the ARIMA Model

On continuous-time autoregressive fractionally integrated moving average processes

Spatial autoregressive fractionally integrated moving average model

An irregularly spaced ARMA(1,1) model and an application to contamination data

Online learning of ARIMA for time series prediction.(2016)

Seasonal fractional long-memory processes. A semiparametric estimation approach

Estimation of parameters of autoregressive models with fractional differences in the presence of additive noise

Time series forecasting for the adobe software company’s stock prices using ARIMA (BOX-JENKIN’) model

A characterization of ARMA and Fractional ARIMA models with infinitely divisible innovations

Beta Autoregressive Fractionally Integrated Moving Average Models

A novel Bayesian approach to estimate long memory parameter

Autoregressive Times Series Methods for Time Domain Astronomy

A goodness-of-fit test for functional time series with applications to Ornstein-Uhlenbeck processes

Hilbert space-valued fractionally integrated autoregressive moving average processes with long memory operators

Statistical Inference for ARMA Time Series with Moving Average Trend