rolcor_estim_1win {NonParRolCor}

Estimates the rolling window correlation coefficients for only one window-length and their statistical significance

Description

The rolcor_estim_1win function estimates the rolling window correlation coefficients for only one window-length (time-scales) for two time series sampled on identical time points, and their statistical significance via a non-parametric computing-intensive method. To carry out the computational implementation we extend the works of Telford (2013), Polanco-Martínez (2019) and Polanco-Martínez (2020). The test/method to determine the statistical significance is described in Polanco-Martínez and López-Martínez (2021). The rolcor_estim_1win function is highly flexible since this contains several parameters to control the estimation of the correlation. A list of parameters are described in the following lines.

Usage

rolcor_estim_1win(inputdata, CorMethod="pearson", widthwin=3, Align="center", 
                  rmltrd=TRUE, Scale=TRUE, MCSim=1000, Np=2, prob=0.95) 

Arguments

inputdata

A matrix of 3 columns: time (regular/evenly spaced), the first (e.g., the independent) variable, and the second (e.g., the dependent) variable. Please verify if inputdata is a matrix.

CorMethod

The method used to estimate the correlations, by default is “pearson,” but other options (“spearman” and “kendall”) are available (please look at: R>?cor.test).

widthwin

The window size or length that indicates the window's size to compute the rolling window correlations.

Align

To align the rolling object, NonParRolCor uses three options: “left”, “center”, and “right” (please look at: R>?running). However, there are some restrictions that have been described lines above. We recommend to use the “center” option to ensure that variations in the correlations are aligned with the variations in the relationships of the variables under study, rather than being shifted to left or to right (Polanco-Martínez 2019, 2020), but this imply that the window-lengths must be odd.

rmltrd

Remove (by default is “TRUE”; “FALSE” otherwise) the linear trend in the variables under analysis. It is advisable to remove the trend before estimating the rolling window correlation coefficients, especially, for large window-lengths.

Scale

Scale (by default is “TRUE”; “FALSE” otherwise) is used to “normalize” or “standardize” the variables under analysis. It is highly advisable to ”normalize/standardize” the time series under study to have them in the same scales.

MCSim

Number of Monte-Carlo simulations to permute the second variable. It is advisable to use at least 1000 simulations.

Np

Number of CPU cores, by default is 2. Please verify the number of cores of your computer. WARNING: it is not advisable to use the maximum number of cores of your computer.

prob

Numeric vector of probabilities with values in the interval [0,1], by default prob=0.95 (p=0.05), please look at R?quantile, Telford (2013) or Polanco-Martínez and López-Martínez (2021) for more information.

Details

The rolcor_estim_1win function estimates the rolling window correlation coefficients for only one window-length and their statistical significance between two time series sampled on identical time points. The function rolcor_estim_1win uses the functions cor.test (package:stats) and running (package:gtools) to estimate correlation coefficients and to compute the rolling window correlations, and also the functions foreach (package:foreach) and makeCluster (package:parallel) to parallelize the estimation of the statistical significance.

Value

Outputs: A list containing six elements: Correlation_coefficients and CRITVAL contain the rolling window correlation coefficients and their respective critical values to determine the statistical significance of these coefficients, CorMethod is the method used to estimate the correlation coefficients (e.g., Pearson, Spearman or Kendall), widthwin contain the window-length (time-scales), and left_win and righ_win are used to accommodate the times of the rolling window correlation coefficients.

Author(s)

Josué M. Polanco-Martínez (a.k.a. jomopo).
Excellence Unit GECOS, IME, Universidad de Salamanca, Salamanca, SPAIN.
BC3 - Basque Centre for Climate Change, Leioa, SPAIN.
Web1: https://scholar.google.es/citations?user=8djLIhcAAAAJ&hl=en/.
Web2: https://www.researchgate.net/profile/Josue-Polanco-Martinez/.
Email: [e-mail].
José L. López-Martínez.
Faculty of Mathematics, Universidad Autónoma de Yucatán (UADY), Tizimín, MEXICO.
Web1: https://scholar.google.es/citations?user=552PKVEAAAAJ&hl=es/.
Web2: https://www.researchgate.net/profile/Jose-Lopez-Martinez-3/.
Email: [e-mail].

References

Polanco-Martínez, J. M. and López-Martínez, J.M. (2021). A non-parametric method to test the statistical significance in rolling window correlations, and applications to ecological time series. Ecological Informatics 60, 101379. <URL: doi: 10.1016/j.ecoinf.2021.101379>.

Polanco-Martínez, J. M. (2020). NonParRolCor: an R package for estimating rolling window multiple correlation in ecological time series. Ecological Informatics, 60, 101163. <URL: doi: 10.1016/j.ecoinf.2020.101163>.

Polanco-Martínez, J. M. (2019). Dynamic relationship analysis between NAFTA stock markets using nonlinear, nonparametric, non-stationary methods. Nonlinear Dynamics, 97(1), 369-389. <URL: doi: 10.1007/s11071-019-04974-y>.

Telford, R.: Running correlations – running into problems. (2013). <URL:
https://quantpalaeo.wordpress.com/2013/01/04/>.

Examples

 # Code to test the function "rolcor_estim_1win"
 # Defining the 'NonParRolCor' parameters 
 # Number of Monte-Carlo simulations (MCSim), please use at least 1000. 
 # WARNING: MCSim=2, it's just to test this example! 
 MCSim <- 2 
 Np    <- 2  # Number of cores 
 X_Y <- rolcor_estim_1win(syntheticdata[1:350,], CorMethod="pearson", 
                   widthwin=3, Align="center", rmltrd=TRUE, Scale=TRUE, 
                   MCSim=MCSim, Np=Np, prob=0.95)