Closed oleg-kachan closed 2 years ago
Hi @oleg-kachan, IDTxl offers basic estimators for mutual information from two (sets of) variables (have a look at the demo or the Wiki page for details). However, we don't implement estimators for the total correlation.
I would close this issue since it is more about usage and not reporting a bug. If you have further questions or would like to discuss this question in depth, consider our Google Groups mailing list. If you have trouble with the code, please reopen or start a new issue.
Thanks, Patricia
Can I compute total correlation, i.e.
TC(X_1, X_2, X_3) = \sum_i=1^3 H(X_i) - H(X_1, X_2, X_3)
with IDTxl? Isn't is idtxl.multivariate_mi.MultivariateMI?
If so, how can I compute just the above value, without any excessive computations such as statistical tests, p-values, different lags, etc.? Just for 3 univariate vectors/time series X_i, with no lags?
Thanks.