Detecting direct causality in multivariate time series: A comparative study

The concept of Granger causality is increasingly being applied for the characterization directional interactions in different applications. A multivariate framework estimating essential order to account all available information from time series. However, inclusion non-informative or non-significant variables creates estimation problems related ‘curse dimensionality’. To deal with this issue, direct measures using variable selection and dimension reduction techniques have been introduced. In comparative work, performance an ensemble bivariate domain assessed, focusing on measures. particular, types high-dimensional coupled discrete systems are used (involving up 100 variables) robustness series length noise examined. results simulation study highlight superiority measures, especially systems.