Figure 8
From: Time series causal relationships discovery through feature importance and ensemble models
Pearson correlation at different lags for a few of the time series given by Eq. (2). There is no strong correlation (close to 1) of either \(\textbf{X}^{1}\) or \(\textbf{X}^{7}\) with some of their drivers (rows), even though we know by definition in Eq. (2) there is an influence from the drivers. This shows that, even with a low correlation between drivers and the target, this does not mean the absence of causal relationships.
