MHVG2MTS: Multilayer Horizontal Visibility Graphs to Multivariate Time Series Analysis

This is the original implementation corresponding to the DGP analysis presented in the paper "MHVG2MTS: Multilayer Horizontal Visibility Graphs to Multivariate Time Series Analysis".

Requirements

• R (>= 3.5.0)

Configurations

Data

• All data sets can be found in folder data/.
  • Data Generating Processes (DGP), the set of 6 linear and nonlinear bivariate time series models
    • bwn_models: White Noise models, the independent and correlated bivariate time series processes
    • var_models: VAR(1) models, the weak and strong bivariate autoregressive processes
    • garch_models: GARCH(1,1) models, the weak and strong generalized autoregressive conditionally heteroscedastic processes
    • the GDP are stored in .RData files and are in the following format:
      • list of matrix of ts objects, ie. mts, for each DGP
    • the csv files contains an instance example of each respective bivariate time series models
• All the measures computed from MHVG's of the DGP's can be found in frolder results/.
  • DegreeSeq, the intra/inter/all-layer degree sequences and the ratio degree sequences
  • DegreeDistribution, the intra/inter/all-layer degree distributions and respective the mean's and sd's degree distributions
  • GlobalFeatures, the intra/inter/all-layer global feature vectors (average degree, average path length, number of communities and modularity) and the relational feature vectors (average ratio degree and intra/inter/all-layer Jensen–Shannon divergense)
• All the empirical results also can be found in frolder results/.
  • PCA_results, the intra/inter/all-layer global features and the relationa features PCA analysis
  • Clustering_results, the clustering analysis the global and relations feature vectors

Source Files

• libraries : contains all required packages
• aux_code/ : contains all auxiliary functions
• info_data : contains some auxiliary data informations
• main_DGP : contains data parameters and runs the procedures to simulate Data Generation Processes
• main_bts : contains procedures to analyse ACF's and CCF's of the DGP
• main_local_features : runs the main procedures for the empirical evaluation of degree distributions of DGP
• main_global_features : runs the main procedures for the empirical evaluation of global and relations features of DGP
• main_clustering : runs the main procedures for the experimental evaluation of DGP clustering based of MNet features