GWO-KGC: A Novel Time Series Causal Analysis Scheme

Existing methods cannot depict the intrinsic correlations of time series accurately. Aiming at the above problem, this paper proposes a novel time series causal analysis scheme called GWO-KGC, which incorporating grey wolf optimization algorithm into kernel-ganger causal analysis. Specifically, the model removes redundant and unrelated variables, and also recognizes the complex nonlinear relationships between variables, thus providing an optimal subset of input features. Meanwhile, in order to verify the correctness of the causal analysis results, the GWO is introduced to the echo state network, to validate the influencing factors. The simulation results show that GWO-KGC model can accurately identify the main influencing factors of the selected variables, which guarantees enhanced prediction accuracy.